Meet Polaris.
It's like Gong for Claude Code sessions.

Bring your local Claude Code sessions straight into a collaborative Slack channel. Polaris works in the background to automatically capture and document the AI's entire execution path including all prompts, responses, and tool calls in real time. Teammates can watch the live log stream, intervene with inline commands via Slack, or audit the complete thought process later.

# webapp
Polaris
Claude Code icon Alice working on auth
Martha writing docs
Claude Code icon Bob building payments

One channel, every session

# webapp
A
Aliceauth10:31

implement auth middleware with RS256

agent.alice10:31

Creating src/middleware/auth.ts...

B
Bobpayments10:32

add Stripe webhook handler

agent.bob10:32

Creating src/api/webhooks/stripe.ts...

B
Bob→ auth10:33

@agent.alice don't forget rate limiting on that endpoint

A
Alice→ payments10:33

@agent.bob use the shared secret from vault, not env vars

agent.alice10:33

Good call. Adding rate limiter...

alice — claude code
> /polaris join #webapp
Connected as alice/auth
> implement auth middleware with RS256
Creating src/middleware/auth.ts with RS256 JWT verification...
bob via slack — don't forget rate limiting
Good call. Adding rate limiter...
bob — claude code
> /polaris join #webapp
Connected as bob/payments
> add Stripe webhook handler
Creating src/api/webhooks/stripe.ts...
alice via slack — use the shared secret from vault
Pulling webhook secret from vault...

What you can do with Polaris

Start streaming your session

> /polaris join #webapp

Every prompt, response, and tool call streams to your team's Slack channel in real time.

Pull a teammate into your session

> /polaris tag @bob I need your input on this auth approach

Bob gets notified in Slack with full context. His reply appears inline in your terminal.

Catch up on a teammate's session

@polaris summarize alice last 2h

Get a summary of what happened — what was built, what decisions were made, what's still in progress.

Redirect an agent from Slack

@agent.alice use RS256, not HS256 — we need asymmetric keys

Your message lands directly in Alice's coding session. The agent picks it up and adjusts.

Attach session context to a PR

> /polaris attach-pr #482

Adds a session transcript — prompts, decisions, and reasoning — to the pull request. Reviewers see the "why," not just the "what."

Search past sessions

@polaris search "webhook secret rotation"

Find when and why a decision was made, across all sessions.

Why Polaris

Your prompts are the most valuable artifact in your engineering workflow.

Every prompt carries intent, context, and decision-making that no commit message can reconstruct. Today, all of it vanishes when the session ends.

AI sessions are invisible

When someone's coding with an AI agent, the rest of the team has no idea what's happening. It's pair programming behind a locked door.

Knowledge evaporates

The full reasoning chain — why the agent chose one approach over another, what it tried and rejected — disappears with the session. Only the final code survives.

Feedback comes too late

Teammates see the pull request, not the process. By then the agent has already committed to a path that one Slack message could have redirected.

Multiple agents, zero shared context

Teams running concurrent AI sessions have no coordination layer. Alice's agent doesn't know what Bob's agent is building, leading to conflicts and duplicated work.

Polaris fixes this. Every prompt, every response, every tool call — captured, streamed to your team in real time, and stored as permanent memory. Teammates can intervene mid-session. Agents across sessions share context through the hub. Nothing is lost.

The vision

Tobi Lutke, CEO of Shopify, recently described a future where every AI interaction in an organization flows through a shared, observable stream — what he calls "the shop floor." The vision he articulates is strikingly close to what Polaris already does.

Article

Learning on the Shop Floor

by Tobi Lutke · May 9, 2026

Read on X →

The idea behind Polaris was born independently, but the convergence isn't a coincidence. As AI agents become central to how software gets built, the need for this layer — variously called a context graph, a memory layer, institutional memory, or decision traces — becomes inevitable. A persistent, searchable record of every prompt, every decision, every session. Polaris is building that layer.

Pricing

Start free. Upgrade when you're ready.

Unlimited users on every plan. No per-seat pricing.

Free

$0 forever

Try Polaris with your team. No credit card required.

Get started
  • Unlimited users
  • 1,000 prompts / month
  • 5 GB data captured
  • 7 days session history
  • Community support

Pro

$29 /month

For power users who rely on Polaris daily.

Get started
  • Unlimited users
  • 5,000 prompts / month
  • 25 GB data captured
  • 30 days session history
  • Community support

Team

$49 /month

For teams that need capacity, history, and support.

Get started
  • Unlimited users
  • 10,000 prompts / month
  • 50 GB data captured
  • 90 days session history
  • Priority support

Enterprise

Custom limits, SSO, audit logs, compliance, and dedicated support.

Contact us
All plans include unlimited users and real-time Slack streaming. No credit card required for Free.

How it works

1

Connect

Sign up and connect your team's Slack workspace. Polaris uses Slack as the collaboration layer — no new apps to learn.

2

Install

npx @lightupai/polaris on your machine sets up hooks, MCP server, and logs you in.

3

Collaborate

/polaris join #your-channel links your Claude Code session to Slack. Every prompt and response streams live — teammates reply there and their messages appear inline.

Ready to try it?

Set up takes less than two minutes.

Sign up with Google Join the community