Intelligence
Relationship Intelligence Relationship Engagement Index Superbrain AI Archetype Bayesian Deal Probability AI Moat Score GTM Command Centre Revenue Miner Say/Do Ratio
Visualise
Relationship Groups Relationship Planner Opportunity Orbit Client Cockpit Bucket Game Perfect Timing
Act
Personalised Communication Effortless Action Automations Network Scanner Deal Rooms Angles Tab
Methodology & AI
AI Coaching Sales Methodologies Sales Canvases Fear Matrix Meeting Intelligence Manifesto Generator AI Partner Assist
More
Product Tour Integrations
Solo Consultants Fractional Executives Boutique Consultancies Agencies
Services Blog Changelog About Sign in Talk To The Founder
Home / Wiki / AI & Automation

How Superbrain learns from feedback

Updated May 2026

You'll learn how to: Tell the difference between Nynch memory, saved skills, draft feedback, and model training.

Time: 2 minutes.

Prereqs: None.

The short version

Nynch does not retrain the underlying AI model from your feedback.

Superbrain learns by saving useful context inside Nynch, then using that context in future answers, drafts, and recommendations. That context can include memories, rules, saved skills, content drafts, recommendation feedback, and outcome patterns.

What happens when you give feedback

  1. Give feedback in plain English.

    Example: "For future AI compliance posts, make them more founder-led and less legal-sounding."

  2. If you want the preference to stick, say so clearly.

    Use phrases like:

    • "Remember this for future posts."
    • "Add this to my LinkedIn content style."
    • "Always avoid that tone."
    • "Update my skill with this."

  3. Assist decides the right place to save it.

    • Memory or rule: Best for broad personal preferences and working style.
    • Skill: Best for repeatable drafting instructions, such as LinkedIn posts, comments, emails, proposals, or meeting prep.
    • Draft revision: Best for improving only the current post or current piece of content.
    • Superbrain learning: Best for tracking whether suggestions were accepted, ignored, completed, or successful.

  4. Skill edits require confirmation.

    Assist can propose a skill update, but it should not silently rewrite a saved skill. You will be asked before the skill changes.

What this means for content feedback

If you refine one LinkedIn post and say "make this shorter", Nynch will use that instruction for the current draft.

If you say "remember this for future LinkedIn posts", Nynch can treat it as a durable style preference.

If you say "update my LinkedIn content skill so future posts are shorter", Assist should propose the skill change and ask you to confirm.

What Superbrain learning means

Superbrain tracks what Nynch recommended, what you did with it, and what happened next. That helps future recommendations become better timed, more useful, and less noisy.

That is different from model training. The AI provider is not being retrained on your private feedback. Nynch is using your saved context to guide future work.

If something goes wrong

  • Symptom: "I corrected a post but future drafts did not change." -> Fix: Say "remember this for future posts" or ask Assist to update the relevant skill.
  • Symptom: "Assist says it updated a skill without asking." -> Fix: Check Settings > Skills. Skill changes should require confirmation.
  • Symptom: "I only wanted to fix this one draft." -> Fix: Use the content composer feedback normally, but do not say "remember this" or "always".
  • Symptom: "I want to see or remove stored memory." -> Fix: Ask Assist what it remembers about the topic, or review the available memory and skills settings for your workspace.

Related: Create a new AI skill | Use a saved AI skill | Configure Growth Assistant agent autopilot.