We use cookies to understand how you use our site and improve your experience. Cookie policy
Last updated:
The Palette context layer connects to your tools, interprets what's happening, and maintains a living map of your organization that both people and AI can read. Here's the full flow.
The context layer connects to the tools your team already uses: Slack, Jira, Linear, GitHub, Notion, Google Calendar. You choose exactly which channels, projects, and repos Palette can see. Nothing is accessed unless you explicitly add it.
Connecting tools takes a few minutes. No data migration, no new workflows to learn.
Raw events from your tools get ingested, scored, and interpreted. This isn't a data dump or an activity feed. The context layer judges what matters and translates noise into meaning.
Example: a PR merges, a Linear project moves to "In Progress", and a cross-team sync gets scheduled. Instead of three separate notifications, the context layer produces one signal: "Product launch prep is ramping up, new onboarding flow shipped and GTM sync scheduled for Thursday."
The difference matters. Events tell you what happened. Signals tell you what it means.
All those interpreted signals feed into your context, a living map with three layers:
Everything is plain English. Every claim has a confidence score so you know what's solid and what needs confirmation. Your context updates continuously as new signals come in.
Palette Desktop has the context layer built in. Agents working in your shared folders read it automatically.
Other AI tools (Claude, ChatGPT, Cursor, Conductor, others) read it via the Palette MCP server. Install it once and your AI starts every session already knowing your org context. No more pasting background into every prompt.
Your team reads it too. A few minutes of plain English instead of an hour of Slack archaeology. New hires get context in days, not months.
Three sources feed the system:
| Source | What it provides | Example |
|---|---|---|
| Connected tools | What happened | PR merged, issue moved, meeting scheduled |
| Human input | What it means | "Sprint was smooth" or "We're blocked on legal" |
| Public data | External context | Competitor funding, market shifts |
The Signal Engine interprets all of it, scoring relevance and extracting meaning.
Your context stores it all, structured, confidence-scored, always updating.
From there, it's consumed two ways: AI tools read it via MCP to start with full context. People read it as plain English to stay aligned without meetings.
The result: your team's AI tools are context-aware, your people are aligned, and nobody had to sit through a status update to get there.
When you add Slack or Linear to your own Claude, your AI only knows what you have access to. You've given it your silo with better search. When every team connects their tools to the context layer, everyone's AI gets context from across the org, tools and channels they're not in, because those teams opted in. That's not just more data. It's a network effect: every team that connects makes every other team's AI smarter. See the MCP page for the full picture.