Your week, in your hands.
Flow availability, focus fragmentation, meeting load and when you focus best, plus how your hours with AI compare to the hours without it.

Synapse joins your work signals with consented human state to show what AI is really doing to delivery and focus, without ever naming, ranking or exposing a person.

Reads work signals and AI usage from the tools your team already runs on
Engineers get private insight into their own flow and focus. Leaders get the team picture as aggregates. Nobody gets a view of anyone else.
Flow availability, focus fragmentation, meeting load and when you focus best, plus how your hours with AI compare to the hours without it.

AI adoption, delivery outcomes, AI spend and consented human-state alignment, combined into one AI Effectiveness Index that refuses to guess.

Synapse turns decisions, docs and code history into memory your agents check before they act, and holds them to the rules your team sets.
People, repositories, pull requests, issues and documents as one living graph. Restricted documents appear as locked, masked nodes.
Guard rules stop an action. Ask rules pause it for a person. Counts are per rule and never name a member.
Ask what the organization knows. The answer is the evidence, each excerpt linked to its source.
OAuth 2.1 with PKCE. Drive, OneDrive and Notion documents reach an agent only if its person can open them at the source.
Set targets on team metrics. Synapse checks hourly and posts to Slack when a line is crossed.
Every figure is drawn as exactly one of five states. Gaps are never drawn as zero, nothing is estimated, and each absence says why it’s missing and how to fix it.
Below the floor, Synapse won’t even say how many people contributed, because a count would expose who chose not to.
Connecting Google is optional and done by each member with their own account. Synapse only reads; it never creates, changes, shares or deletes anything in Google.
Synapse’s use and transfer of information received from Google APIs adheres to the Google API Services User Data Policy, including the Limited Use requirements. Google data is never sold, used for advertising or used to train generalized AI models. Read the details in our privacy policy.
Every seat comes with a monthly allowance. Overage is billed at published rates, so the bill is as readable as the dashboard.
See pricingNo. My Effectiveness and your agent decisions are visible only to you, admins included. Team pages show aggregates above the privacy floor (10 people by default, never fewer than 4) and never name, rank or drill into a person.
Activity and timing: commits, pull requests and reviews, issue titles and state changes, calendar event times and AI usage counts, plus the text of documents and channels you choose as memory. Never issue descriptions, prompts or completions. Code search reads repositories live and stores nothing.
Its figures are withheld, and the page shows the parent’s numbers with a banner saying so. A small team can still see its own view if every member opts in, and it disappears the moment anyone opts out.
Claude, Claude Code, Cursor and any client that supports remote MCP connect directly. Usage reports from Claude Code, GitHub Copilot and other AI tools come in through Connectors.
Documents from Google Drive, OneDrive and Notion reach an agent only if its signed-in person can open them at the source, checked again on every request. Slack channels and repositories your admins choose as team memory are shared across the organization.