The Linea coach

An AI tutor that remembers you.

Most AI tutors are very smart and very forgetful. Linea's coach reads your full path history every time you open a conversation. It knows what you said last week. It knows what you applied. It knows what you skipped.

Per-learner memory

Every conversation, lesson completion, stuck point, and applied insight is stored against your account. The coach reads it on every prompt.

Anchored to your path

Coach answers are grounded in the lessons you're actually in. Not generic LLM platitudes. References go to specific lesson timestamps.

Picks up the thread

The coach opens conversations with what you left unfinished. If you said you'd apply something by Friday, it asks on Saturday.

Sample conversation

You said last week you wanted to ship one eval suite to prod by Friday. Did you?

Half. Ran on a test set, didn't wire to CI yet.

Half is fine. Module 3 lesson 7 is the CI wire-up specifically — 18 minutes. Want me to queue it next?

Yes. Also, I'm worried about the eval drift problem you mentioned in module 2. Should I add monitoring before going live?

Good instinct. Before you ship, set up a weekly cron that re-runs the suite against the last week's production prompts. Module 5 lesson 4 covers exactly this pattern. I'll queue it after the CI lesson.

Got it. Last thing — how does this connect to the pricing experiment we talked about in Founders Toolkit?

It doesn't directly, but it's adjacent. Reliable evals give you the confidence to price by quality, not just by volume. We can come back to that when you finish module 5 here. Want me to flag it for the weekly check-in?

Built on Claude 3.5 Sonnet. Per-learner memory in Postgres with vector embeddings for retrieval. Every prompt benchmarked against held-out learner conversations for quality.

Try the coach in the dashboard