Joe just started in billing/ — his work touches yours.
He's writing a Stripe webhook handler that consumes JWTs. You're mid-refactor on the JWT library. Without context, his code will break tomorrow.
For engineering teams shipping with AI
Your team's reasoning lives in AI sessions now. Lore captures it automatically.
Every decision, dead end, and tradeoff your team makes in Claude Code, Cursor, or Codex — captured locally, sourced to the exact session moment, shareable in one click.
Local CLI · processes on your machine · 90 seconds · free for individuals
The problemGitHub shows what shipped. Linear shows what was planned.
GitHub shows what shipped. Linear shows what was planned.
Neither shows what your team was actually thinking.
The reasoning behind every shipped line of code now happens inside an AI session. It used to evaporate. With Lore, it doesn't.
19%
of engineer time spent searching for info or asking colleagues
Stack Overflow / McKinsey~9 mo
for new senior engineers to reach full productivity
Industry standard onboarding ramp70 / 17
of AI users say tools helped them personally — only 17% say it helped the team
2025 AI productivity surveysYour engineers think out loud now. Lore listens.
How it works
No new process. No engineer time. The data is already there.
Your team keeps coding in Claude Code, Cursor, or Codex. Lore does the rest, locally, on each engineer's machine.
01
Ingest
A local CLI reads sessions from Claude Code, Cursor, and Codex on each engineer's machine. Nothing leaves until they say so.
02
Structure
Lore extracts decisions, dead ends, requirements, and open questions as typed objects. Each one stays sourced to the exact session moment it came from.
03
Surface
Your team gets a live workspace: who's working on what, every decision sourced, every thread's reasoning history, and related work surfaced in context.
What you get
Stop re-deriving what your team already figured out.
Five surfaces, one source of truth: your team's actual reasoning, structured automatically and never authored by hand.
01
Decision Log
Every architectural call your team has made — auto-extracted, sourced to the session moment, citable in PRs.
When a teammate asks 'why did we pick Postgres over Dynamo?', the answer is the exact session message where the decision happened. Not a summary. Not a doc someone remembered to write. The real reasoning, alternatives and all.
02
Live Collaboration
See your team's thinking in real time — and join in while it's hot. On reasoning, not just documents.
- Avoid. A banner fires before you write code that breaks a teammate's in-flight refactor.
- Initiate. See who's deep in the problem you have opinions on. Send a 10-min huddle invite with context already attached.
- Pull. Say “I'm stuck on Stripe webhook idempotency.” Lore routes you to the teammate who solved it last week.
- Drop in. A senior notices a junior making a risky call in the live feed and hops in unprompted.
03
Cross-engineer handoff
When the work passes between engineers, the context passes with it.
Sarah hits EOD mid-debug. Tom picks it up tomorrow. He doesn't restart the conversation — he inherits Sarah's reasoning, the things she ruled out, and the open thread she left behind. No re-explaining. No lost context.
04
Cross-repository memory
Solving the same problem twice is a tax. Lore charges you once.
Lore notices when work in one repository structurally resembles work done elsewhere. 'Sarah solved a near-identical webhook retry problem in billing last week — view her approach?' Surfaces before you start, not after you finish.
05
Team search
Ask 'why do we handle Stripe webhooks this way?' Get the why — distilled, sourced, and summarized, with the session moment one click away.
The how is in the code. The why — the reasoning, the tradeoffs, the alternatives ruled out — used to evaporate. Lore makes it searchable in plain English, sourced to the session message it came from.
Auto-organized
Your work, connected. Without you connecting it.
Solo or team, no tagging required. Lore connects related sessions and decisions automatically, then surfaces the overlap when it matters.
01
Related work, inferred.
Lore uses the cwd, files touched, repositories, and decisions to find related sessions. No tagging. No setup.
02
Overlap, learned.
Lore notices when reasoning in one session matches another — shared code, shared decisions, shared people. Across your own work or across your team's.
03
Surfaced just in time.
Not a graph you have to maintain. Lore drops a "related decisions" link into the session where it matters, then gets out of the way.
Example
You start a webhook handler for Twilio status callbacks. Lore surfaces: "3 decisions in your Stripe webhook handler are structurally similar — idempotency check, retry-on-429, dead-letter routing. Skip the signature-verification one; Twilio uses a different scheme. View the relevant 2?"
Learned, scoped, and surfaced in context — instead of authored, static, and decayed. Lore connects related reasoning by reading what you did.
Live Collaboration · Avoid
The collision nobody catches today.
Sarah is refactoring JWT in auth-service. Joe starts writing a JWT consumer in billing-service. Neither knows. Lore does.
Without Lore
10:47am
Sarah opens a Claude Code session in auth-service. She decides to swap jsonwebtoken for jose. Nothing leaves her machine yet — no commit, no PR, no Slack ping.
11:03am
Joe opens Claude Code in billing-service. He needs a Stripe webhook handler that verifies JWTs. His agent reads the current auth code — written against jsonwebtoken — and confidently builds on it.
11:30am
Sarah pushes her PR. Joe is still writing against the old library. Tomorrow's merge conflict is locked in. Half a day of rework — nobody knows yet.
With Lore
10:47am
Lore silently structures Sarah's reasoning as she works.
11:03am
Joe's session boots. ⚠ A banner: Sarah is in JWT right now. The code you're about to read is changing. One click: load her decision into your context. Sarah receives the same notification — Joe was alerted about her work.
11:30am
Joe's webhook uses the new library from the start — no conflict, no rework. Lore captures his decision and links it to Sarah's. Two related calls on the Decision Log, citable in tomorrow's PR review.
What everyone else misses
Other tools see the output. Lore sees the thinking.
ToolWhat they seeWhat they miss
GitHubThe 200-line diff that shippedThe 3 approaches that didn't
Linear / JiraThe ticket was “Done”The reason the simple fix took 4 days
Notion / ConfluenceThe doc someone remembered to writeEverything no one remembered to write
Hyper / GleanSynthesized summary of org chatterThe exact session message where the decision was made
Anthropic native memoryOne engineer's accumulated contextEveryone else's
LoreThe reasoning that produced the workNothing. The data is already there.
Trust mechanic
You decide what your team sees.
Lore runs locally on your machine. You see your own digest before anyone else. Edit, redact, or hold any session back. The manager view is a byproduct of work engineers control — not a surveillance layer.
- 01Local-first ingestion. Sessions are processed on the engineer's machine. Nothing reaches the team until they approve it.
- 02Edit before share. Every digest is editable. Redact specific lines, hold a session back, or share only the decisions.
- 03Secrets auto-scrubbed. API keys, env vars, and obvious secrets are removed before anything is shared.
- 04Engineer-controlled visibility. New threads follow your account setting: workspace visibility for workspace members by default, and private otherwise. You can change any thread to private, workspace, or public.
Your team's best thinking is already happening. Start capturing it today.
Free forever for individuals. Free for teams under five. Install in 90 seconds.