Take It With You: What You Get If You Leave Nynch | Nynch Skip to content
Use cases Product Pricing Strategy Day Resources Sign in Talk to the founder
Take it with you

If you leave Nynch, everything you put in leaves with you.

Ask any vendor holding your company's context what you would get to take with you if you ever leave. This page is our answer. It is generated from the same list the export runs on, so it cannot promise more than the export does.

Definitions version 2026-10-20. Export format 1.0.

What you get.

One folder you can open without Nynch. Spreadsheets for the records, plain Markdown for the memory Nynch keeps about each relationship and account, and the definitions that say what every stage and status means.

Folder layout

  • README.md
  • manifest.json
  • data/<dataset>.csv
  • data/<dataset>.jsonl
  • memory/relationships/<name>-<id>.md
  • memory/accounts/<name>-<id>.md
  • memory/learnings/*.md
  • transcripts/<meeting-id>.md
  • proposals/<id>.json
  • proposals/<id>.md
  • semantics/nynch-bd.ossie.yaml

Formats

  • CSV (UTF-8, header row)
  • JSON Lines
  • Markdown with YAML front matter
  • Apache Ossie YAML

Who can start it

A workspace owner or admin, after two-step verification. Every export is written to the audit log.

How long it takes

  • Under two minutes for a few hundred people.
  • Ten to fifteen minutes for about 20,000 people and 60,000 conversations, across two or three background runs. A run that stops resumes from where it stopped.
  • Ready to download for seven days, then deleted.

See it before you need it. The sample export is a real export of a made-up workspace: twelve people, four companies, their deals, memory files, a meeting transcript and a proposal. Every name is generated and every address ends in example.test.

Every dataset, 74 of them.

Each dataset is a CSV file and a JSON Lines file under data/. 63 of them can be loaded straight back into a Nynch workspace. Where Nynch writes a Markdown memory file from a dataset, the table says so.

Dataset What it holds Files
companies Every company (account) with its firmographics, tier and enrichment. CSV, JSONL , plus Account memory
people Every person in the workspace with their profile, status, owner and relationship scores. CSV, JSONL , plus Relationship memory
opportunities Deals: the live pipeline rail. Status is derived from outcome (null = open). Opportunities: the retiring rail, kept read-only. Same thing as a deal. The rail column tells them apart. CSV, JSONL , plus Relationship memory and Account memory
pipelines Pipelines the deals move through. CSV, JSONL
pipeline_stages Stages per pipeline. stage_type (active, won, lost) is the canonical stage mapping. CSV, JSONL
deal_people Which people are attached to which deals. CSV, JSONL
deal_stakeholders Stakeholder maps per deal: role, influence, support and concerns. CSV, JSONL
buying_committee Buying committee members per deal. CSV, JSONL
deal_activities Activities logged against deals. CSV, JSONL
deal_stage_history Every stage move with how long the deal sat in the previous stage. CSV, JSONL
deal_comments Comments on deals. CSV, JSONL
deal_lines Service lines priced on each deal. CSV, JSONL
interactions Every recorded touch with a person: emails, meetings, calls, messages and manual notes of contact. CSV, JSONL , plus Relationship memory
activities The older activity log per person. CSV, JSONL
email_messages Imported email messages linked to a person (text bodies, no HTML). Messages not linked to a person are not reachable by workspace and are left out. CSV, JSONL , plus Relationship memory
email_threads Thread-level summaries, commitments and signals. CSV, JSONL
sent_emails Emails sent from Nynch (text bodies). CSV, JSONL , plus Relationship memory
linkedin_messages LinkedIn messages imported from the extension. CSV, JSONL , plus Relationship memory
chat_messages WhatsApp, Slack and other chat messages synced to a person. CSV, JSONL , plus Relationship memory
meetings Meeting records with summaries, commitments and full transcripts (also written to transcripts/). CSV, JSONL , plus Relationship memory
calendar_events Imported calendar events and their attendees. CSV, JSONL
meeting_attendees Who attended which meeting. CSV, JSONL
notes Notes written on people. CSV, JSONL , plus Relationship memory
comments Teammate comments on people, companies and deals. CSV, JSONL
tasks Team tasks. CSV, JSONL
action_items Personal action items tied to people and deals. CSV, JSONL
commitments Promises made and received. CSV, JSONL , plus Relationship memory
genuine_commitments Commitments confirmed as genuine from conversations. CSV, JSONL , plus Relationship memory
company_people Which people belong to which companies, including secondary associations. CSV, JSONL , plus Account memory
employment_history Past and present roles per person. CSV, JSONL
contact_emails Every email address per person with consent and deliverability state. CSV, JSONL
contextual_roles What a person is to the firm on an account, agreement, project or deal. CSV, JSONL
company_signals Company signals as the workspace sees them (funding, hiring, leadership, news). CSV, JSONL , plus Account memory
contact_signals Detected changes about a person (job moves, title changes). CSV, JSONL , plus Relationship memory
social_signals Social activity signals per person. CSV, JSONL , plus Relationship memory
opportunity_signals Buying signals spotted in conversations. CSV, JSONL , plus Relationship memory
signals Revenue signals (expansion, upsell, referral angles). These are signals, not deals. CSV, JSONL
intelligence_items Intelligence items surfaced about people and companies. CSV, JSONL
engagement_scores Two-way engagement scores per person. The inputs to the warmth tier. CSV, JSONL
ideal_client_profiles Ideal client profiles. CSV, JSONL
services The services the firm sells. CSV, JSONL
proof_points Proof points and evidence the firm uses. CSV, JSONL , plus Account memory
gtm_problems The customer problems the firm solves. CSV, JSONL , plus Account memory
competitors Competitors and how the firm wins against them. CSV, JSONL
manifestos Positioning manifestos. CSV, JSONL
custom_fields Custom field definitions. CSV, JSONL
custom_field_values Custom field values per record. CSV, JSONL
lead_source_options The workspace lead-source list. CSV, JSONL
relationship_memory The synthesised relationship memory per person: history, facts, what not to assume. CSV, JSONL , plus Relationship memory
account_memory The synthesised account memory per company. CSV, JSONL , plus Account memory
agent_memories Durable memories Nynch agents kept about people, workflows and preferences. CSV, JSONL
knowledge_entities Knowledge-graph entities for people, companies and deals. CSV, JSONL
known_facts Known facts: relationships between people, companies and deals with their evidence. CSV, JSONL , plus Relationship memory
beliefs Buyer beliefs the firm has catalogued. CSV, JSONL
belief_observations Each time a belief was observed in a conversation. CSV, JSONL , plus learning
deal_learnings Patterns learned from won and lost deals. CSV, JSONL , plus learning
proposal_learnings Patterns learned from proposals. CSV, JSONL , plus learning
decisions Decisions recorded with their reasoning. CSV, JSONL
introduction_requests Warm introduction requests and their outcomes. CSV, JSONL
team_intro_requests Introduction asks between teammates. CSV, JSONL
relationship_paths Warm paths to target people through connectors. CSV, JSONL
circles Circles: the people around one target, each asked to help open a conversation. CSV, JSONL
circle_members Members of each circle. CSV, JSONL
proposals Proposals with their full content (also written to proposals/). CSV, JSONL
proposal_versions Every proposal version as a frozen document. CSV, JSONL
proposal_options Priced options inside proposals. CSV, JSONL
proposal_artifacts Where each proposal was exported to (links to the generated documents). CSV, JSONL
agreements Commercial agreements per account. CSV, JSONL
agreement_versions Accepted versions of each agreement with frozen terms. CSV, JSONL
agreement_lines Service lines inside accepted agreement versions. CSV, JSONL
projects Delivery projects per account. CSV, JSONL
campaigns Outreach campaigns and their steps. CSV, JSONL
campaign_enrollments Who was enrolled in which campaign and what happened. CSV, JSONL
company_briefs Company briefs written for prospects. CSV, JSONL

What stays behind, and why.

662 internal tables are left out. Most are working copies, caches or scores Nynch can rebuild from the records you do get. Each one is listed below with its reason, so you can check the list rather than take our word for it.

Operational state (128 tables)

Private reviewed repair audit and authorization state. Never generic relationship context or retrieval evidence.

deal_activity_repair_receipts, historical_pipeline_repair_items, historical_pipeline_repair_plans, legacy_commercial_repair_receipts

Retired Academy community forum. Product surface removed. Rows retained. Not relationship context.

academy_forum_categories, academy_forum_comments, academy_forum_posts, academy_forum_votes

Add/Skip decisions feeding the exec monitor trust model. Feature-internal learning state.

exec_suggestion_decisions

Ambiguous historical Deal outcome repair queue. It is review workflow state, not customer-authored relationship context.

deal_outcome_repair_queue

Append-only Daily Relationship Queue outcome and recipe-learning ledger. Never generic AI context or RAG material.

daily_relationship_assignment_events

Append-only Deal outcome landmarks for leakage-safe model evaluation. The canonical Deal already supplies customer-facing outcome context.

deal_outcome_events

Backend-only merge values for campaign emails from website sign-ups (the calculator). The same result is written as a contact note, which is the copy Assist reads.

contact_signup_results

Bulk processing job and item state. User-facing records live in their destination sources.

bulk_job*

Calculated product reliability scoreboard snapshots for super-admin owners. They contain operational evidence only, never relationship context.

feature_reality_status_snapshots

Campaign jobs-page watch state. User-visible journey state remains canonical on campaign_enrollments, pending_campaign_actions, and the drafted email.

campaign_job_watches

Candidates held back pending a past-roles retry. Detected people surface as new_exec_suggestions once resolved.

exec_monitor_held_candidates

Champion job-change watch state: user champion marks and last-checked times. The findings live on contact_job_changes and the alerts on ai_alerts.

champion_watch

Coalescing flags and Gmail catch-up cursors for push-triggered syncs.

sync_push_triggers

Contact merge audit trail. The surviving contact is canonical.

contact_merges

Content-workspace run health and counters. Source ideas and drafts are indexed separately.

content_analysis_status

Customer-owned Resend provider event receipts and webhook ingestion ledger. Campaign delivery outcomes are canonical on campaign_email_delivery_claims.

resend_campaign_event_receipts

Customer-visible website-scan progress receipt. It records process state only. Safe website signals remain on the company and public cache contracts.

company_website_scan_receipts

Daily Relationship Queue generation state. The user-facing people and relationship evidence remain on canonical Contact and Company records.

daily_relationship_days

Database query performance telemetry.

query_performance_log

Delivery receipts for the 30-minute meeting brief (claimed, delivered, skipped). The brief content lives on meeting_prep_docs and notifications.

meeting_brief_deliveries

Derived contact meeting-count repair cursors are not relationship evidence.

contact_meeting_rollup_checkpoints

Derived daily ranking and workflow state. Canonical Contact, Company, signal, and action records remain the relationship truth.

daily_relationship_assignments

Derived prospecting signal response weights. Canonical outreach and reply evidence remains on queue and interaction records.

prospecting_signal_response_weights

Email enrichment execution log. Accepted data lives on contacts.

email_enrichment_log

Email generation execution log. Sent/draft content has canonical sources.

email_generation_log

Email send execution state. Messages and sent emails are canonical.

email_sends

End-of-day synthesis execution logs.

eod_synthesis_logs

Enrichment work queue and retry state.

enrichment_jobs

Firmographic queue control, job receipts, and health ledgers. Accepted customer-facing fields and provenance are stored on companies.

company_enrichment_*

Frozen ranking snapshots derived from canonical company, signal, relationship, and activity records. Avoid duplicating that evidence in AI context.

focus_company_members

Granular campaign delivery and tracking telemetry. Campaign outcomes are canonical.

campaign_email_*

Grok Bot campaign assignment queue. User-visible journey state remains canonical on campaign_enrollments, pending_campaign_actions, and contact notes.

campaign_agent_jobs

Historical employer-link remediation queue and audit status. Verified relationship truth remains on contacts and companies.

contact_company_review_queue

How each target-market tier was built (filters, estimated size, confidence). The tiered companies themselves are canonical company records.

market_definitions

Immutable engagement correction audit. Read only through bounded parent-authorised lifecycle history.

client_engagement_events

Import progress state. Safe connection status is exposed through the connected-integration context wrapper.

integration_import_progress

Internal identity-resolution conflict evidence. Canonical user context remains on contacts.

contact_identity_conflicts

Internal progress ledger for the bounded company identity shadow scan. Never relationship context.

company_identity_observation_scan_state

Internal reversible identity-link ledger. Canonical user context remains on contacts.

contact_identity_links

Internal shadow identity evidence and reviewer labels. Canonical user context remains on companies.

company_identity_observations

Last contact-duplicate scan receipt per workspace for honest empty states. Accepted people remain on contacts.

workspace_dedupe_scan_state

Learned skip patterns a user undid for the exec monitor. Feature-internal learning state.

exec_skip_pattern_overrides

Lease and retry state for governed derived-signal calculation. It contains process health, not customer relationship knowledge.

derived_signal_recalc_queue

Merge audit events. The surviving contact is canonical.

contact_merge_events

Merge audit trail. The surviving company is canonical.

company_merges

Multi-key identity graph (LinkedIn vanity/opaque/conversation, email, phone, name) for match/orphan attach. Canonical person context remains on contacts.

contact_identity_keys

Offline release-gate receipts for Today readiness. Operational validation metadata.

daily_relationship_readiness_validation_receipts

One-off website pre-qualification runs over market candidates. Not monitoring. The resulting companies are canonical records.

market_web_check_runs

Operational no-resend delivery and history-reconciliation ledger. Canonical sent content is stored in sent_emails.

outbound_email_send_attempts

Pending new-executive suggestions for the exec monitor. Surfaced through its own alert card, not Assist, until the feature is proven.

new_exec_suggestions

Per-candidate website check outcome and a short quote, kept for the add step. Canonical company records carry the durable context.

market_web_check_results

Per-contact relationship-memory refresh queue and rewrite timestamps. The memory itself lives on contact_ai_profiles.

relationship_memory_refresh_state

Per-contact scan receipts proving the fan-out ran. Detected signals live in contact_signals and ai_alerts.

signal_scan_receipts

Per-mailbox receive bookkeeping (workspace, last sync, last error) for secondary mailboxes.

email_account_sync_state

Per-mailbox sync lease so push, drain and hourly poll never overlap.

mail_sync_leases

Per-run progress and spend tracking for the weekly new-exec detector. Operational, never relationship context.

exec_monitor_progress

Per-user approval rate for the exec monitor trust model. Feature-internal state.

exec_monitor_trust

Per-user generation status and credit accounting. Materialised queue records hold the user-facing result.

daily_relationship_generator_runs

Per-workspace watch allowance settings (per-person cap, private-watch toggle). Configuration, not relationship context.

workspace_watch_settings

Point-in-time Focus market readiness diagnostics. Canonical companies, signals, and ICP/service records remain the source of market truth.

focus_market_snapshots

Principal-independent private Storage deletion receipts for removed Audit introduction videos. Cleanup state must never become generic AI context.

value_audit_video_cleanup_tombstones

Privacy-safe intent, provider-attempt, quality, and cost receipts for email resolution. Accepted addresses live on Contacts. These ledgers are operational evidence, not AI context.

email_resolution_*

Private actor-owned email and meeting attribution receipts. Human attachments are not generic Assist or retrieval evidence. Source access remains scoped to the attaching person and workspace.

deal_communication_links

Private actor-scoped relationship snapshots. Read only through the permissioned history RPC

contact_member_histories

Private audience snapshots and resumable import progress. Canonical journeys are campaign enrollments.

campaign_audience_imports

Private durable Value Audit queue, lease, retry, and quality receipts.

value_audit_jobs

Private engagement idempotency payload and exact replay receipts. Never generic AI context.

client_engagement_receipts

Private historical market reset recovery snapshots. Never relationship context.

focus_reset_20261009

Private history discovery cursors. Operational progress, not generic CRM evidence.

relationship_memory_discovery

Private per-alert selection and planned-action state. The customer-facing company alert carries the evidence and roster. This table must not become generic AI context.

company_alert_contact_actions

Private publication and draft delivery receipts. Canonical activity is exposed by Today.

launch_kit_asset_receipts

Private source-backed continuity excerpts. Read only through the actor-authorised evidence RPC.

relationship_memory_evidence

Private target-market before-images and compare-and-set state. Never relationship context.

target_market_reset_rows

Private target-market confirmation and recovery receipts. Never relationship context.

target_market_reset_operations

Private transaction-bound recovery context. Never relationship context.

target_market_restore_context

Private unfinished composer state. The sent message becomes canonical in sent_emails. Drafts must not influence AI context or search.

email_drafts

Private upload and lifecycle ledger for Audit introduction videos. The approved Audit snapshot carries the public-facing media reference. Upload state must never become generic AI context.

value_audit_video_assets

Private Value Audit batch receipts, dispositions, and counts.

value_audit_batches

Private visitor no-resend ledger. Never relationship context.

visitor_email_delivery_intents

Provider push subscription ids, expiries and per-subscription client secrets for mail and calendar sync. Never AI context.

sync_push_subscriptions

Provider sync cursors and progress only. Campaign engagement is read through the bounded Focus Companies workflow, not generic Assist context.

zenabm_engagement_syncs

Provider, cost, and stage receipts for Daily Relationship Queue generation. Operational evidence, not relationship context.

daily_relationship_receipts

Public authority-scan rate limiting.

authority_scan_rate_limits

Public endpoint rate limiting.

public_rate_limits

Redacted contact-to-company reliability ledger. It records outcome and refresh state only. Never relationship context.

company_link_live_proofs

Redacted operator-attended Chrome proof ledger. It records only outcome, extension version, source commit, and time. Never relationship context.

extension_supervised_proofs

Repair audit trail. Canonical company/contact records hold the corrected result.

company_mismatch_audit

Retired Academy events calendar. Product surface removed. Rows retained. Not relationship context.

academy_events

Retry queue for deleting unreferenced storage objects after failed uploads. Canonical customer records and their live storage references remain the source of truth.

storage_orphan_cleanup_queue

Retryable per-source generation work and provider receipts. Materialised queue records hold the user-facing result.

daily_relationship_generator_jobs

Reviewed recovery audit and idempotency state is not relationship context.

deal_customer_recovery_receipts

Rotation state for which watched company was last checked. Operational scheduling state, not relationship context.

exec_monitor_company_checks

Scheduled-job locks and processor state.

cron_*

Sealed Today readiness evaluation evidence and verdicts. Operational proof, not generic AI context.

daily_relationship_readiness_evaluations

Send ledger for inbound audit emails: recipient, attempts, and failure reason. The audit itself is canonical on company_briefs and the person who asked is canonical on the captured lead. This row only records whether it reached them.

inbound_audit_deliveries

Server-owned Fathom delivery and polling checkpoints are operational state. Meeting content remains on meeting_transcripts.

fathom_webhook_state

Service-only archive of duplicate email rows removed by the one-row-per-message cleanup.

email_message_dedupe_archive

Service-only campaign transfer lineage. User-visible journey state remains canonical on campaign_enrollments.

campaign_enrollment_transfers

Service-only cursor per activity source for the team re-match backfill: last id, counts and pass timestamps. Holds no customer content. The re-attached activity lives on its own sources.

team_activity_rematch_progress

Service-only ledger for handing the records of a departed teammate to their successor: planned and moved counts per record kind, the resume step and the ids of rows a guard refused. Operational handover state, not relationship context.

member_offboarding_handover*

Service-only ledger for the Monday team summary: stored facts, the single AI narrative and a resume lease. Derived from deals, contacts and intro requests, which stay canonical. Never relationship context.

team_weekly_summary_runs

Service-only per-recipient delivery state for the Monday team summary (notification id, email intent). Delivery bookkeeping, not relationship context.

team_weekly_summary_deliveries

Service-only private calendar delivery identity and recovery receipts are not relationship context.

private_calendar_prep_deliveries

Service-only record of a disconnected mailbox (owner, provider account, address) so its stored mail can be re-attached when the same user reconnects it.

email_account_tombstones

Service-only resume cursors for batch scans (kg-sync, ai-data-hygiene). Operational state, never relationship context.

system_scan_cursors

Service-only run ledger for a supervised workspace tenancy move: table checkpoints, batch counts and the parked identity keys of records being merged. Operational migration state, not relationship context, and the parked columns briefly hold an email or LinkedIn URL that must never become generic AI context.

workspace_tenancy_move_*

Service-run queue that removes the mail and calendar data synced from a disconnected mailbox or calendar.

mailbox_data_purges

Short-lived enrichment debit reservation and settlement state. Accepted profile data lives on contacts and durable balance movements live in credit_transactions.

contact_enrichment_credit_reservations

Short-lived working list for a winnable-market refresh spanning several calls. Removed when the refresh finishes. Company records carry the tiers.

market_som_refresh_progress

Singleton cap/cursor control for the nightly signal fan-out. Operational, never relationship context.

signal_scan_config

Snapshot of statuses changed by the one-time 2026-10-14 meeting insight sweep, kept so the sweep can be reversed. Internal migration state, never relationship context.

meeting_insight_sweep_20261014

Telegram send/delivery log. Conversation history is canonical.

telegram_message_log

Temporary dedupe candidates. Accepted results live on contacts.

contact_dedupe_candidates

Temporary enrichment staging. Accepted contact data is canonical.

pending_contact_enrichments

Throttle state for refreshing derived prospecting signal response weights. Canonical outreach and reply evidence remains on queue and interaction records.

prospecting_signal_response_refresh_state

TypeSafe Jev shadow-mode agreement ledger: compares Jev decisions with LLM decisions in background, never affects user-facing outputs. Operational diagnostic evidence only, not relationship context or AI input.

jev_shadow_comparisons

UI dismissal state, not relationship knowledge.

flow_dismissals

User-visible monthly focus-list workflow state. Canonical company, signal, deal, task, and campaign records already provide the underlying context.

focus_company_cohorts

Webhook delivery and dedupe log.

webhook_processing_log

Which pre-written webinar posts an owner removed. The posts themselves remain canonical on content_drafts.

campaign_post_dismissals

Which teammate owns each exec in the New in Role campaign. Ownership state. The person themselves lives in contacts.

exec_campaign_enrollments

Who was @mentioned in a record comment. Notification routing state, not relationship context.

record_mentions

System internals (82 tables)

RAG internal storage

rag_document_chunks, rag_documents

Admin-defined per-step conversion targets for the growth/case-study funnels. Configuration data, not relationship context.

growth_funnel_goals

Anonymous per-step page views for the public funnels. Analytics only, not logged-in relationship context for Superbrain indexing.

growth_funnel_page_views

Assist source-access registry, not customer data.

superbrain_source_registry

Automation failure/debug log. Operational, not relationship context.

automation_failures

Background LinkedIn-search workqueue drained by the Chrome extension. Operational state, not relationship data.

linkedin_search_jobs

Chrome extension acceptance-check observability. User-facing relationship outcomes are stored on contacts and drip_connect_queue.

linkedin_connection_check_runs

Chrome extension parse telemetry. Operational observability.

extension_parse_telemetry

Config: sources that stay searchable but are never embedded (mirrors contract sources without the rag surface).

rag_indexing_exclusions

Connected-app execution audit logs

mcp_outbound_call_logs

Connected-app pending action state. Operational.

mcp_pending_actions

Connected-app tool execution logs. Operational observability.

mcp_execution_logs

Connector sync run logs. Operational observability, not relationship data.

integration_sync_logs

Daily ad spend/clicks per traffic source for the funnel-economics (EPC/ROAS/CAC) dashboard. Marketing analytics, not logged-in relationship context for Superbrain indexing.

funnel_ad_daily

Daily Google Analytics 4 metrics (sessions, source/medium, conversions) synced by sync-analytics-daily. Company-level analytics, not per-user relationship data.

ga4_daily

Daily Mixpanel product-analytics metrics (event counts, unique users) synced by sync-analytics-daily. Company-level analytics, not per-user relationship data.

mixpanel_daily

Database migration metadata

schema_migrations

Demo walkthrough infrastructure, product demos, anonymous view/step analytics. Not relationship data.

demo_*

Derived routing table holding one public web address per Company. Assist already reads Audit content through company_briefs. The slug adds no relationship context and would only invite fabricated links.

value_audit_company_addresses

Durable LinkedIn post signal-extraction queue. User-facing signals are stored in intelligence_items and raw evidence in linkedin_posts.

linkedin_post_signal_jobs

Email import analysis/classification retry ledger and dead-letter visibility. User-facing intelligence is stored on emails and email_threads, not indexed from this operational failure log.

email_analysis_failures

Exit-overlay email opt-in log. The captured email becomes a contact via find_or_create_contact (which Superbrain already indexes). The log itself is operational.

video_optins

External provider id to Deal mapping that makes integration upserts idempotent. The Deal carries the customer-facing data.

deal_provider_identities

First 30 Days course instrumentation: dedupe ledger for stall nudges (one row per user, gate, step). Delivery bookkeeping, not relationship context.

course_nudges

First 30 Days course instrumentation: per-week weekly-review completion rows used to count consecutive reviews for the operator_day30 graduation gate. Analytics only, not relationship context.

course_weekly_reviews

High-volume granular playback telemetry (play/pause/seek/progress) for drop-off curves. Not relationship context.

video_watch_events

High-volume internal audit trail

entity_audit_log

Idempotent Actions evidence ledger. User-facing targets and scored results are exposed through weekly_action_plans and canonical Say/Do summaries.

action_activity_events

Immutable frozen Say/Do score snapshots derived from weekly_action_plans and the Actions evidence ledger. A cache that stops a finalized weekly score from drifting, not a source of relationship context.

action_say_do_snapshots

Immutable published calculator versions addressable by opaque link token. Public snapshots carry no answer results. They are definition+metadata snapshots only.

client_tool_public_snapshots

Internal activation-funnel instrumentation (time-to-value milestones for the admin Activation dashboard). Analytics only, not logged-in relationship context for Superbrain indexing.

activation_milestones

Internal Assist pilot cohort and outcome ledger. It is operational measurement, never relationship context or AI source material.

assist_pilot_*

Internal automation queue

automation_event_queue

Internal background rescore requests raised by ICP changes and repair backfills. Operational queue state, not relationship knowledge.

icp_rescore_requests

Internal browser identity attribution. Not AI context.

video_engagement_bindings

Internal calculator draft state: generation claims, definitions and status. User-facing calculated results are never persisted. This table is plumbing for the draft-ready transition.

client_tool_drafts

Internal dedup/audit log for hot-chat alerts (notify-hot-chat). Not relationship data.

chat_alert_log

Internal Jungler pilot dedupe and provider audit ledger. User-facing signals are stored in social_signals, linkedin_contact_engagements, and linkedin_social_leads.

jungler_pilot_events

Internal kill-switch and cap configuration for the guarded enrichment cron. Operational safety config, not relationship knowledge.

enrichment_cron_control

Internal latency telemetry for the post-meeting orchestrator

post_meeting_processing_metrics

Internal pilot playback telemetry and session capability hashes. CRM reads use scoped summary RPCs, never raw AI retrieval.

video_engagement_sessions

Internal pilot publishing configuration, not relationship context.

video_engagement_placements

Internal prospecting cache for recovered direct employers from recruiter job posts. User-facing evidence is stored on contacts and queue source metadata.

prospecting_recruiter_recoveries

Intro batch job state. The intros themselves live on contact activity.

intro_request_batches

LinkedIn drip work queue. Relationship outcomes live on contacts and linkedin_messages.

drip_connect_queue

Meeting-prep work queue. The user-facing artifact is meeting_prep_docs.

meeting_prep_queue

Notification dedup keys. Pure infrastructure.

telegram_notification_dedup

One-off audit snapshot from the 2026-05-19 LinkedIn message dedup cleanup migration. Backoffice recovery trail, not relationship data.

_linkedin_dedup_audit_*

One-off audit snapshot from the 2026-05-25 LinkedIn message misattribution cleanup migration. Backoffice recovery trail, not relationship data.

_linkedin_misattribution_audit_*

One-off work list for re-sorting mis-labelled email (2026-10). Holds only email ids. The result is stored on email_messages.

email_resort_queue

Operational completion ledger for the imported-email analysis retry worker. Outcomes belong in email messages, threads, and the failure ledger.

email_analysis_retry_runs

Operational Firecrawl monitor state. User-visible website changes are stored in company_signals and intelligence_items.

company_website_watches

Operational queue/status for background Rhythm-brief research prep. The prepared research is stored on contacts.signaliz_data, not here.

rhythm_brief_prep

Operational run history for website monitoring. User-visible signals are stored in company_monitoring_snapshots and intelligence_items.

company_monitoring_runs

Per-user on/off and ordering for the sources feeding the daily Today list. UI configuration, not relationship context.

daily_source_preferences

PostgreSQL system tables

pg_*

Pre-signup Network Revenue Audit lead capture (anonymous scan results + assembled deliverable) for the free audit lead magnet. Hydrated into onboarding at signup, not logged-in relationship context for Superbrain indexing.

connector_sessions

Precomputed Growth Agent opening turns, keyed by lead magnet and greeting branch. A latency cache over text the assistant wrote itself. Holds no visitor and no relationship context.

growth_agent_greeting_cache

Private invitation retry deduplication keys. User-facing activity lives in action_activity_events and sprint totals.

connection_sprint_action_receipts

Private observation-to-invitation retry mappings. Not user-facing activity.

manual_invitation_observation_receipts

Public /growth founding-member funnel leads and per-step analytics. Not logged-in relationship context for Superbrain indexing.

growth_funnel_leads

Public authority lead-capture records, not logged-in relationship context.

authority_score_leads

Public Commercial Spine lead magnet submissions and generated reports. Not logged-in relationship context for Superbrain indexing.

commercial_spine_leads

RAG embedding job queue

rag_embedding_jobs

RAG indexing queue

rag_index_queue

RAG storage, queues, caches, feedback, and ranking internals.

rag_*

Raw embeddings

content_embeddings

Request rate-limit log with non-UUID IDs

request_rate_log

Scan session state rows. Operational.

scan_sessions

Scoped resume capability hashes, not relationship context.

video_engagement_links

Scoring work queue state. Operational, not relationship data.

contact_scoring_queue

Scrape run log. Detected signals live in social_signals/linkedin_* sources.

social_scrape_log

Short-lived job bookkeeping for Strategy generations. The finished work lives in its own table. This row is plumbing, not relationship context.

strategy_generation_jobs

Skill version snapshots. The active skill content is indexed via company_skills/user skills sources.

skill_versions

Supabase internal tables

supabase_*

Superbrain configuration and status metadata

superbrain_*

UI flow session telemetry. Product analytics, not relationship data.

flow_sessions

Unmatched participant triage queue. Resolved participants become contacts.

unmatched_meeting_participants

Voice-of-market marketing pipeline. Backoffice, service-role only, not relationship data.

vom_*

Webhook debug logs

webhook_logs

Withdrawn LinkedIn invitations and their re-invite block, used by the connection queue. The user-facing record is the contact timeline entry written with each withdrawal.

linkedin_invitation_withdrawals

Security-sensitive (60 tables)

Private repeatable growth state is available only through owner-scoped or explicitly consented readers.

growth_cadence_rules, growth_contributions, growth_event_changes, growth_event_variants, growth_events, growth_evidence_links, growth_invitation_evidence, growth_invitations, growth_materialization, growth_occurrences, growth_plan_versions, growth_plans

Confidential paid-diagnostic engagement state is read only through its own contracts, never as generic AI context.

client_diagnostics

Confidential paid-diagnostic sources, evidence, findings and history are read only through their own contracts, never as generic AI context.

client_diagnostic_*

Connected mailbox account and token metadata.

email_accounts

Connected-app keys, grants, subscriptions, logs, and OAuth state.

mcp_*

Dependent rows captured with a Company delete so Undo can restore them. Never AI context.

company_delete_contract_secondary_items

Ephemeral private transaction authorization for canonical Account creator rehoming. It is a mutation fence, never customer knowledge or AI context.

account_creator_rehome_authorizations

Ephemeral transaction authority for advancing the campaign sender rotation counter during the trusted pool update. It is a security fence, never customer knowledge or AI context.

campaign_sender_counter_authorizations

Ephemeral transaction authority for assigning immutable campaign sender snapshots during the trusted pool update. It is a security fence, never customer knowledge or AI context.

campaign_sender_assignment_authorizations

Ephemeral transaction authority for closing current Account owners during archive or merge. It is a security fence, never customer knowledge or AI context.

account_lifecycle_owner_close_authorizations

HMAC-only erased-identifier tombstones. Never user or AI context.

core_erased_*

Immutable outcome correction snapshots are audit evidence and must not enter generic Assist retrieval.

zenabm_outcome_link_reviews

Mailbox identities and provider configuration.

email_identities

Non-PII privacy-erasure proof. It is an operator compliance record, never relationship evidence or AI context.

personal_account_erasure_tombstones

OAuth tokens and pending authorization state.

oauth_*

Observed outcome links are scoped to explicit same-company deals and remain outside generic Assist retrieval until a dedicated reader is validated.

zenabm_outcome_links

Per-action execution receipts proving an automation step ran exactly once. The resulting record change is the customer-facing truth.

automation_action_receipts

Pre-delete Company snapshot retained only to power Undo. Never AI context.

company_delete_contract_snapshots

Privacy-bounded mutation audit ledger. Exposed only through the policy-scoped receipt contract, never as AI context.

core_mutation_receipt*

Private blocked-validation replay details for Deal close and correction. They must never become generic AI context.

deal_close_failure_replay_payloads

Private campaign mailbox routing configuration is not generic relationship context.

campaign_sender_pool

Private human coaching and delivery state must never become AI knowledge or public content.

human_coaching_*

Private idempotency and replay ledger for Account owner mutations. Never relationship evidence or AI context.

account_owner_assignment_receipts

Private immutable Deal close and correction replay payloads. Canonical commercial truth remains on the Deal and its outcome event.

deal_outcome_mutation_replay_payloads

Private owner feedback. No generic AI indexing.

growth_review_decisions

Private reviewed drafts, task and send receipts, and usage events are not relationship evidence.

meeting_follow_through*

Private tenant-bound terminal RPC responses for exact retry replay. Product state remains on the Account or Deal. This ledger is never AI context.

client_contract_command_results

Private transaction-local authorization fence for the exact proposal delivery receipt transition. Never customer knowledge or AI context.

proposal_delivery_mutation_context

Private transaction-local scope for exact campaign privacy erasure. Never customer knowledge or AI context.

campaign_privacy_erasure_scopes

Private visitor opt-out capability state. Never relationship context.

visitor_email_suppressions

Provider-sourced account metrics are exposed only through the authenticated Focus Companies workflow until a workspace-scoped Assist reader is validated.

zenabm_account_engagement

Raw ABM review rows include unfiltered provider payloads. Assist may read only the tenant-scoped view that returns user-reviewed rows and omits provider evidence.

zenabm_review_actions

Restricted keyed erasure registry material. Never user or AI context.

core_erasure_*

Row-level payloads of a pre-delete Company snapshot, retained only to power Undo.

company_delete_contract_snapshot_items

Service-only workshop applicant emails, payloads, token hashes and referral receipts are outside CRM tenant scope. Never generic AI context.

workshop_applications

Service-role-only allowlist for the invite-only OAuth signup gate. It is an access-control fence holding third-party emails, never customer knowledge or AI context.

approved_signup_emails

Single-row shared secret proving a request came from our own pg_cron jobs. Owner/service-role only. Never customer knowledge or AI context.

internal_cron_auth

Stored terminal responses for replayed Contact contract calls. The saved Contact is the customer-facing truth.

contact_contract_replay_payloads

Stored terminal responses for replayed Deal delete/restore calls. The Deal row is the customer-facing truth.

deal_lifecycle_replay_payloads

Stored terminal responses for replayed Deal stage moves. The Deal stage is the customer-facing truth.

deal_stage_move_replay_payloads

Stored terminal responses for replayed Opportunity delete/restore calls. The Opportunity row is the customer-facing truth.

opportunity_lifecycle_replay_payloads

Telegram account-linking credentials.

telegram_link_*

Telegram connection credentials and status.

telegram_connections

Telegram session state.

telegram_sessions

Transaction-local erasure authority rows. Access control state, not relationship context.

growth_erasure_scopes

Unmatched and conflicting provider identities require explicit human review and must not enter generic Assist retrieval.

zenabm_identity_reviews

User AI provider configuration and encrypted credentials.

user_ai_*

Webhook endpoints and secrets.

user_webhooks

Zapier authorization and connection credentials.

zapier_*

Regenerable AI output or derived score. The facts it was built from are exported. (54 tables)

aha_moments, ai_alerts, ai_playbook_suggestions, ai_playbooks, ai_recommendations_log, authority_reports, bucket_analytics, calendar_event_insights, campaign_learnings, campaign_step_outcomes, checklist_ai_suggestions, company_data_suggestions, company_features, competitive_mentions, competitive_positions, connection_recommendation_feedback, connection_recommendation_weight_adjustments, connection_recommendations, contact_data_suggestions, contact_features, contact_icp_matches, contact_icp_score_history, contact_intelligence_cards, content_recommendations, contract_alerts, deal_ai_suggestions, deal_coaching_actions, deal_intelligence, deal_velocity_recommendations, duplicate_suggestions, email_learnings, email_outcomes, fear_busters, feed_items, fox_assessments, icp_briefs, intelligence_signals, leaderboard_metrics, meeting_email_drafts, meeting_prep_docs, network_pulses, nudges, pipeline_stage_analytics, prospect_predictions, prospect_scores, sentiment_analysis_log, service_improvements, service_knowledge_log, social_topics, strategy_asset_scores, team_streaks, trend_scans, weekly_digests, whale_dossiers

AI internals (52 tables)

Agent capability registry. System configuration, not user data.

agent_registry

Agent handoff queue state. Operational.

agent_handoffs

Agent runtime, queue, checkpoint, schedule, and execution state.

agent_*

Agent scratch context store. Durable knowledge lives in agent_memories.

agent_context_store

AI feedback telemetry

ai_recommendation_tracking

AI response cache

ai_cache

AI runtime, cache, evaluation, budget, and provider internals.

ai_*

AI-drafted account version of a pitch deck awaiting human approval, same class as value_audit_outreach_drafts. Letting the model read its own unapproved deck copy back would turn generated claims about a buyer into recorded fact. What the account actually said stays canonical in meeting_transcripts, deal_stakeholders and the Deal research dossier.

deal_pitch_renders

AI-drafted outreach awaiting human approval, same class as company_briefs. Unapproved drafts must not feed AI context or search, or the model would read its own unsent output back as fact. The underlying relationship facts stay canonical in contacts, companies, and company_signals, and a sent message becomes canonical in sent_emails.

value_audit_outreach_drafts

AI-generated commercial offer hypotheses awaiting human judgement. Blocked until output quality is proven. An offer the user promotes becomes canonical in services.

offer_opportunities

Append-only Today ranking exposure and outcome events for model evaluation. The Today queue records remain the customer-facing state.

today_rank_events

Bucket recommendation queue state. Accepted outcomes live on contact_buckets.

bucket_recommendations

Deal-intelligence calibration and model internals.

dii_*

Deal-intelligence resonance work queue. Operational.

dii_resonance_queue

Dedupe similarity scores, not user-authored relationship context.

contact_similarities

Derived per-offer contact suggestions. The canonical relationship record is the contact itself. These rankings are regenerable AI output.

offer_shortlist_members

Derived stage-baseline priors and counts. They are calculation internals, not customer-authored Deal context. Customer-facing reads use the governed projection.

deal_outlook_priors

DII stage medians. Model internals.

dii_user_stage_medians

DII statistical priors. Model internals, not user content.

dii_user_priors

Durable Account Research retry/resume state. User-facing reports are stored in ai_cache and surfaced through the reasoning-agent, not indexed as relationship context.

account_research_jobs

Durable analyze-intelligence work queue. User-facing intelligence is stored in intelligence_items and opportunity_signals, not indexed from this operational queue.

intelligence_analysis_queue

Durable retry state for automatic closed-Deal ICP learning. The resulting evidence lives in icp_learning_events.

icp_closed_deal_learning_queue

Evaluation data

ai_golden_sets

Evaluation tables

ai_evaluation_*

Generated candidate hypotheses awaiting queue admission. Blocking them prevents unreviewed output from reinforcing later AI context.

daily_relationship_candidates

Heartbeat calibration telemetry.

heartbeat_effectiveness

Heartbeat proactive-scan run history. Outputs surface as ai_alerts/action_items.

proactive_scan_runs

Internal ICP refinement work and decision history. The current ideal_client_profiles row is the canonical Assist context, so legacy Golden suggestions and pending AI output must not feed generic intelligence.

icp_learning_events

Internal ICP suggestion-application history. Assist reads the current ideal_client_profiles row and canonical closed Deals instead of recycling historical or legacy Golden refinements as intelligence.

icp_refinement_history

Legacy Golden Standard pattern retained temporarily for production rollout compatibility. Automatic ICP intelligence now comes from canonical closed Deals.

user_golden_icp

Legacy probability history produced by the retired visible scorer. Its event model is not trusted and it cannot be used as generic AI evidence.

deal_probability_history

Metrics buffer

ai_metrics_buffer

Model prediction telemetry used for calibration. Not user-searchable content.

prediction_log

Multi-agent coordination events. Operational observability.

agent_coordination_log

One Offer Creator generation run: input snapshot, status, and AI cost metadata. Operational run history. The resulting offers live in offer_opportunities.

offer_ranking_runs

Per-call AI cost ledger (our cost, billing-internal)

ai_call_costs

Per-call data-provider cost ledger

data_provider_calls

Playbook scan run snapshots. Suggestions surface via ai_playbook_suggestions.

playbook_scan_snapshots

Probability calibration ledger containing shadow candidate observations and resolved outcomes. It remains evaluation-only until a model promotion contract explicitly approves a user-visible projection.

deal_prediction_outcomes

Prompt templates

ai_prompts

Prospect planning chat and an unexecuted research plan. It holds what the user asked for and what the assistant suggested, not verified facts about any person or company, so reading it back would present planning guesses as relationship context.

prospect_conversations

Provider enrichment cache. Accepted fields live on contacts and companies.

enrichment_cache

Quality metrics

ai_quality_metrics

Signal identity-verification audit ledger. Customer reads use current_company_signals.

company_signal_trust_assessments

Standard Derived Signal model output and diagnostics. Only explicit approved projections may become product context. Raw envelopes never become generic AI evidence.

derived_signal_snapshots

Statistical model-evaluation reports and shadow operational health. These internals cannot become customer relationship evidence or generic AI context.

derived_signal_validation_runs

Temporary service-suggestion cache. Accepted services live in service_bundles, services, and service_tiers.

service_suggestions_cache

Theme extraction cache keyed by question hash. Derived cache, not source content.

theme_extraction_cache

Usage tracking

ai_usage

User evaluation of a derived interpretation. It must not be recycled as source evidence or generic AI context.

derived_signal_feedback

Vendor unit prices

data_provider_unit_costs

Workspace usage-credit ledger, prices and background job state

usage_credit_*

Product configuration, personal preferences or app housekeeping, not business memory. Workspace name and plan are in manifest.json. (45 tables)

automation_rule_executions, automation_rules, automation_templates, buckets, calendar_display_preferences, contact_buckets, contact_field_layout, contacts_blacklist, content_triggers, crm_views, daily_pulse_sessions, deal_monitor_config, decision_templates, discovery_schedules, meeting_prep_settings, mentions, methodology_stages, nudge_dismissals, nudge_preferences, pipeline_automation_rules, pipeline_stage_requirements, profiles, revenue_goals, rhythm_sort_sessions, signal_definitions, signal_preferences, signal_types, stage_checklist_items, style_profiles, sync_preferences, team_activities, team_relationship_moments, time_blocking_preferences, tracking_workspaces, user_email_styles, user_integrations, user_intelligence_settings, user_methodologies, user_rules, warming_cadence_progress, warming_cadences, weekly_action_plans, weekly_time_blocks, worksheets, workspaces

Tracking, telemetry and working copies of the retiring opportunities rail. The records they describe are exported. (44 tables)

campaign_email_tracking, company_monitoring_snapshots, company_signal_feedback, contact_bigmarker_events, contact_education, contact_identity, contact_job_changes, contact_reminders, contact_scoreapp_results, contact_sources, deal_checklist_responses, deal_close_date_history, deal_room_item_events, deal_room_view_notifications, deal_room_views, deal_stage_responses, documents, email_attachments, email_open_tracking, external_events, inferred_email_connections, link_clicks, linkedin_contact_engagements, linkedin_posts, linkedin_social_leads, newsletter_stats_snapshots, newsletter_subscribers, opportunity_calls, opportunity_documents, opportunity_stage_history, opportunity_stakeholders, political_map_snapshots, signal_attributions, social_comment_drafts, social_engagement_queue, social_engagements, telegram_conversation_history, tracked_companies, tracking_high_value_email_alerts, tracking_links, user_documents, video_views, videos, visitor_events

Configuration (43 tables)

API keys and OAuth tokens

integration_config

Chrome extension feature configuration.

extension_config

Client-authored marketing framework slides. Content the client writes for their own audience, not relationship context.

framework_wheels

Coach-curated skill templates copied into company_skills at signup. Service-role only, not relationship data.

partner_skill_templates

Coach/affiliate partner records (service-role only) used by stripe-webhook to seed content into attributed workspaces. Not relationship data.

partners

Company-level GTM value drivers (pain relievers and gain creators). Workspace strategy configuration, not relationship contact data.

gtm_value_drivers

Connected-app OAuth client grants. Security configuration.

mcp_oauth_client_grants

Connected-app OAuth metadata and encrypted tokens

mcp_user_connections

Core record source registry used for database validation and UI labels. Product metadata, not relationship context.

record_source_types

Custom-domain setup state for the public AI assistant. Not relationship data.

workspace_assistant_domains

Derived signal-weight configuration. Canonical feedback and deal outcomes remain the auditable learning evidence.

focus_signal_learning_weights

Derived soft-ranking weights for Daily Relationship Queue strategies. Canonical assignment events hold the auditable first-party outcomes.

daily_relationship_generator_learning

Global workshop admission, closing-time and open-state configuration is outside CRM tenant scope. Never relationship context.

workshop_campaign

Historical private presentation and webinar content. Not relationship context.

pov_asset_variant_revisions

Historical private profile facts and sales scripts. Not relationship context.

launch_kit_sales_revisions

Internal feature flags

feature_flags

Internal onboarding state

onboarding_progress

Internal transactional email templates.

system_email_templates

Junction table linking services to their selected value drivers. Derived from services and gtm_value_drivers config. Not relationship data.

service_value_driver_links

Per-user list of people hidden from the post-commenter panel. A display preference, not relationship context.

post_commenter_exclusions

Per-workspace branding + alert thresholds + lead owner routing for the public AI assistant.

workspace_assistant_configs

Per-workspace slide styling (colours, font, logo). Configuration, not relationship context.

framework_brand_kits

Private newsletter and video drafts with source research. Not relationship context.

launch_kit_longform

Private presentation and webinar drafts and generation context. Not relationship context.

pov_asset_variants

Private profile facts and sales scripts. Not relationship context.

launch_kit_sales_assets

Private publishing destination configuration.

launch_kit_publishing_settings

Private strategy writing, generation state and revision history. Not relationship context.

launch_kit_sprints

Private UI layout configuration.

command_centre_layout

Service-only limits and scheduling controls for Daily Relationship Queue generation. Configuration, not relationship context.

daily_relationship_generator_config

Service-only reversible briefing preference snapshots. Backup configuration, not relationship context.

briefing_test_preference_snapshots

Service-only temporary briefing test controls. Operational configuration, not relationship context.

briefing_test_controls

Short-lived OAuth verifier state

mcp_oauth_state

System rule catalog. Installed and customized rules live in user_rules.

rule_templates

System skill catalog. Installed skills live in user_skills and company_skills.

skill_library

System skill templates. Installed skills live in user_skills and company_skills.

skill_templates

Temporary workspace capability switch for the pre-launch Value Audit pilot. It grants access only and contains no relationship context.

value_audit_preview_workspaces

The consultant's own published writing. Read through the dedicated voice-context loader for drafting prompts, never generic AI context.

workspace_content_items

The workspace's own core problem statement. Read through the Strategy loader for generation prompts. The same statement stays on the profile for Assist.

workspace_strategy_profiles

User-installed Preferences and Playbooks use dedicated scope-aware loaders and explicit consultation. Never expose them through generic context, search, or RAG.

user_skills

Workspace Preferences and Playbooks use dedicated scope-aware loaders and explicit consultation. Never expose them through generic context, search, or RAG.

company_skills

Workspace typical-vs-opposite strategy drafts. Adopted rows reach Assist only through the dedicated GTM context reader, never generic table dump or RAG.

gtm_market_contrasts

Workspace-owned Value Audit generation and rendering configuration. It contains instructions rather than customer or relationship facts and must not enter assistant context.

value_audit_domain_packs

Your Content source registry (blog, Substack, own LinkedIn). Configuration, not relationship context.

workspace_content_sources

Assist and agent conversations and run state. The memories they produced are exported as agent_memories. (21 tables)

account_agent_assignments, agent_activity_log, agent_project_outputs, agent_projects, agent_schedules, agent_settings, agent_tasks, agent_workflow_runs, assist_conversations, assist_projects, assist_reviewed_zenabm_actions, assist_workpieces, assistant_actions, chat_messages, coaching_commitments, coaching_sessions, conversations, deal_review_commitments, deal_review_sessions, notification_events, notifications

Prospecting, content-studio and Academy working tables. Anything that became a person, company or deal is exported there. (20 tables)

academy_courses, academy_lessons, academy_progress, benchmark_snapshots, business_knowledge, connection_sprints, connector_scores, content_brief_hooks, content_drafts, content_outlines, content_publish_jobs, content_themes, content_video_drafts, pending_campaign_actions, prospect_deep_signals, prospect_outreach, prospects, sprint_opportunities, sprint_runs, testimonial_requests

Reusable templates and packaging that only make sense inside the product. The services, proposals and campaigns built from them are exported. (20 tables)

bundle_tier_services, bundle_tiers, campaign_templates, email_templates, email_templates_ai, gtm_customer_profiles, gtm_problem_links, lead_magnets, linkedin_conversation_labels, linkedin_labels, linkedin_lead_sources, linkedin_snippets, linkedin_snoozes, linkedin_split_inboxes, proposal_templates, service_bundles, service_templates, service_tiers, user_deal_room_blocks, user_deal_room_templates

Sign-in and authorisation state (19 tables)

Invitation delivery receipts and retry state. Recipient addresses and process health, not relationship context.

workspace_invitation_delivery_attempts

Invitation tokens

workspace_invitations

Permission mappings

workspace_members

Private paid activation and failure receipts are operational evidence, not relationship context.

paid_purchase_activation_journeys

Private Team application decisions are authorization state, not relationship context.

team_checkout_approvals

Records which actor may change a custom field definition. An authority mapping, not relationship context.

custom_field_definition_delegations

Removed SSO providers and their verified domains. Gates converting members to password accounts

workspace_sso_removed_connections

Role assignments

workspace_user_roles

Role definitions

user_roles

Security audit logs

workspace_access_log

Service-managed commercial capability assignments, not relationship context.

workspace_capability_grants

Service-only checkout intents contain billing identity and payment session identifiers.

stripe_checkout_attempts

Service-only immutable purchase-to-account receipts are authorization state, not AI context.

paid_subscription_owners

Service-only paid-checkout login delivery receipts can temporarily contain a bearer link and must never become AI context.

paid_checkout_login_email_receipts

Service-only Team-upgrade cancellation receipts contain billing identifiers and must never become AI context.

team_upgrade_proration_receipts

Service-only workspace billing provenance is not relationship context.

workspace_subscription_claim_receipts

Single-use paid-checkout login handoff receipts contain authentication state and must never become AI context.

paid_checkout_handoffs

SSO domain verification tokens

workspace_sso_domains

SSO identity provider configuration

workspace_sso_connections

Buyer-facing rooms and partner-network records shared with other accounts. The proposal and deal behind them are exported. (14 tables)

deal_room_items, deal_room_members, deal_room_sections, deal_rooms, esignature_documents, network_partner_invitations, network_partners, partner_exchange_deal_link_events, partner_exchange_deal_links, partner_exchanges, partner_profiles, proposal_edit_feedback, proposal_option_services, proposal_refinement_messages

Billing (13 tables)

Billing and Stripe plan configuration.

subscription_plans

Billing details

subscription_addons

Billing usage meters

usage_meters

Credit balance and billing state.

user_credit_wallet

Credit ledger movements. Billing data, not relationship context.

credit_transactions

Current Value Audit allowance reservation and consumption state.

audit_usage_operations

Immutable Value Audit allowance event ledger.

audit_usage_ledger

Internal credit pricing configuration.

credit_cost_config

Offer Creator trial/paid entitlement and quota counters. Billing state, not relationship context.

offer_browser_entitlements

Stripe customer, subscription, and charge data.

stripe_*

Usage metering events.

usage_events

User billing

user_addons

Workspace usage metering.

workspace_usage

Client-foundation ledgers (billing periods, movements, handover and close events). The agreements, versions, lines and projects they roll up are exported. (13 tables)

client_close_events, client_deal_services, client_handover_run_events, client_handover_run_steps, client_handover_runs, client_project_health_checkpoints, client_project_lifecycle_events, commercial_agreement_lineages, commercial_agreement_rebase_requirements, commercial_billing_schedule_periods, commercial_external_finance_mappings, commercial_line_movements, commercial_project_allocations

Nynch commercial state (9 tables)

Append-only commercial classification and correction history. Reporting may reconstruct it explicitly, but generic AI context must use the canonical Deal motion rather than historical authority records.

deal_commercial_motion_history

Effective-dated Account owner history. Generic AI context must use the canonical current Account projection, not reconstruct authority from historical rows.

account_owner_assignments

Private approved-proof selection for Audit Playbooks.

audit_playbook_version_proofs

Private Audit Playbook lifecycle configuration.

audit_playbooks

Private consultant method and outreach configuration.

audit_playbook_versions

Private immutable evidence provenance for customer-facing Value Audits.

value_audit_evidence_packs

Private source excerpts, URLs, confidence, freshness, and visibility for Value Audits.

value_audit_evidence_items

Private standing account hypotheses (pains, competitors, buying roles) drafted from audit research. Direction for generation, never assistant context: an unverified hypothesis read back as fact is exactly the fabrication class the registry exists to block.

value_audit_account_plans

Private standing sector research reused across Value Audits. Market items enter audits only through the evidence pipeline, never assistant context.

value_audit_sector_briefs

Nynch administration (8 tables)

Administrative tables

admin_*

Company revenue ledger, every Nynch Stripe payment classified by SaaS/services split. Super-admin only. Company finance data, not user relationship context.

revenue_events

Customer support issue workflow. Admin-facing operational data, not relationship context.

bug_reports

Daily company-wide KPI snapshot (Nynch Top 20). Internal founder/admin business metrics, not user relationship context.

company_kpi_snapshots

Internal database-audit findings, not customer relationship context.

audit_results

Internal product roadmap, not customer relationship data.

roadmap_items

Private screenshots for customer support issue workflow. Admin-facing operational evidence, not relationship context.

bug_report_attachments

Support content and conversations use the dedicated support/wiki path.

support_*

Derived from proposals, which is exported. (5 tables)

proposal_delivery_attempts, proposal_delivery_receipts, proposal_export_reconciliation_events, proposal_generation_operations, proposal_generation_plan_usage_attempts

Firm memory that carries only a user id, not a workspace id. Exporting it by membership would put the rows a teammate holds in another workspace into this one, so it waits until the table carries workspace_id. (5 tables)

knowledge_base_items, methodologies, outcome_patterns, testimonials, user_case_studies

Import working tables (4 tables)

Import job tracking

csv_import_jobs

Temporary bucket-game/import staging. Accepted contacts are canonical.

pending_contacts

Temporary dedupe candidates

csv_import_match_candidates

Temporary import data

csv_import_rows

Derived from deals, which is exported. (1 table)

client_engagement_states

Derived from proposal_learnings, which is exported. (1 table)

proposal_learning_plan_usage_attempts

Raw signal evidence. The verified view is exported as company_signals from current_company_signals. (1 table)

company_signals

Columns never exported.

Inside the tables you do get, these columns are always left out.

embedding (unless switched on) Vector embeddings. They are rebuilt from your text by any AI tool, so they stay out unless you switch them on.
*_token Access tokens for connected accounts such as your inbox and calendar.
oauth_data Sign-in data for connected accounts.

Other people's words.

Transcripts, emails, chat and LinkedIn messages contain the words and personal data of people outside the firm. The firm owns these records and is their data controller.

Treat the export the way you treat your inbox. Keep it somewhere secure and handle it under your own data protection policy.

Open definitions.

Every export includes semantics/nynch-bd.ossie.yaml. It says, in Apache Ossie format, what each lifecycle stage, warmth tier and opportunity status means and how it is worked out from the data, so another tool can read your export the way Nynch does.

The definitions are licensed under the Apache License 2.0. The file ships inside every export, including the sample above.

Talk to the founder

30 minutes. On your real data. Not generic slides.