faq · 33 answers
Frequently asked questions.
The questions teams ask before adopting Pulse, answered straight. If something here isn't covered or contradicts what you find elsewhere on the site, tell us on /contact and we'll fix it.
About Pulse
What is Pulse?
Pulse is a company brain. It indexes the work that actually happens across 25 connectors, Slack, GitHub, GitLab, Notion, Linear, Jira, Asana, Salesforce, Stripe, Microsoft 365, Drive, meeting transcripts, and more, then turns it into a permission-aware graph of decisions, commitments, owners, and open questions that a team can ask, navigate, and act on.Who is Pulse for?
Teams of 5 to 500 people. The sweet spot is any team that has outgrown one Slack workspace, accumulated three or four different sources of truth, and wants a memory layer that respects existing access controls rather than a chat box that ignores them.What problem does Pulse solve?
Three problems. Decisions disappear into Slack scrollback the day they are made. Context lives in tools that don't talk to each other, so onboarding a new person takes weeks of asking around. Existing AI tools either ignore who is allowed to see what, or wrap a chat box around a doc index and call it knowledge. Pulse fixes all three: it keeps decisions as structured records, is ACL-mirrored so it only shows people what they're allowed to see, and asks for approval before it acts.Is Pulse a chatbot?
No. Pulse can answer questions, but the value is the decision history and the approval-gated actions that hang off it. A chat box is the easy part of building this category. The hard parts are turning work into structured records, mirroring ACLs from every source tool, and getting the confidence score honest enough to trust.
How it works
How does Pulse build the knowledge graph?
Connectors pull in source data on a schedule, and instantly through webhooks where a tool supports them. Pulse reads each item and pulls out the decisions, commitments, action items, PR reviews, and meeting outcomes, along with their owners, deadlines, and how they connect. It is stored in Postgres with pgvector for semantic retrieval. Every item keeps a link back to its source and the access control list of the document it came from.How does Ask work end to end?
Five steps, and you can see all of them on every answer. First Pulse gathers the sources you're allowed to see. Then it ranks them by relevance and recency. Then it attaches a citation to each sentence. Then it writes the answer with Anthropic Claude, OpenAI, or Google Gemini on an endpoint that keeps nothing. Finally it scores its own confidence from how well it has done on similar questions in your workspace, and holds back rather than guess when the evidence is thin.How does Pulse decide what to surface in a briefing?
Each briefing is built from weighted sections (overnight decisions, commitments coming due, stuck PRs, customer signal, and answers that have started slipping). You can turn each section up or down, pin things to always include, and mute things you never want to see. It respects your delivery channel and quiet hours, so briefings land when you're ready for them, not at 3am.What is the decision graph?
Every decision Pulse finds becomes its own record, with the rationale, the owners, the sources behind it, and the things that depend on it. You go straight to a decision instead of re-Googling it. It's editable, so when Pulse gets a connection wrong or merges two things that should be separate, your fix teaches it to do better.Where does Pulse host data?
Pulse runs on Vercel and Supabase Postgres, served from a single primary region today. EU data residency is planned for Enterprise, not yet available. Your content and everything Pulse builds from it live in that Postgres database, with row-level rules that keep each company's data separate, so one company's records can never reach another's. No data leaves the primary region except for exports you trigger yourself.
Integrations
Which tools does Pulse integrate with today?
Twenty-five connectors live today across code (GitHub, GitLab), tasks (Linear, Jira, Asana, Monday, Trello), docs and files (Notion, Confluence, Drive, Dropbox, Airtable), CRM and revenue (Salesforce, Zoho, Stripe, Apollo), meetings (Zoom, Fireflies, Granola), enterprise (Microsoft 365, Gmail), plus Slack, Calendar, Miro, Canva, and Clay. A Push API lets you send custom sources in from any API key with write scope.Are connectors read-only?
Connectors are read-only by default. Writes happen through a separate agent-actions surface that requires explicit approval per action, allowlist enforcement on the recipient or target, and per-policy rate limits. Read scopes and write scopes are requested separately so a workspace can adopt Pulse for retrieval first and enable agent actions later.What if a tool we use isn't on the list?
Three options. (1) The Push API lets you send any data in over HTTPS, authenticated with a bearer API key that has write scope. (2) A self-hosted bridge to run a connector inside your own VPC, connecting back to Pulse over an encrypted authenticated link, is on the roadmap, not yet shipped. (3) For high-priority asks the Pulse team prioritizes new connectors based on customer demand. Tell us what you need on /contact.Can Pulse use my company's existing search index?
Pulse does not depend on an upstream search index. It builds its own structured picture from your source content. That said, Pulse's MCP server can be called by any AI tool you already use, so your existing assistants can query Pulse without ripping out anything you already have.
Security and trust
Can my coworker see things I shouldn't see?
No. Slack channel privacy, Confluence page restrictions, GitHub repo permissions, and Drive file ACLs replicate into Pulse during connector sync. Notion is the one connector that can't be mirrored per page (its API exposes no per-page permissions), so Notion content is scoped to the installing admin or the whole workspace, chosen at connect time. Every answer, search, briefing, and agent draft is filtered through these permissions before you see it. When Pulse spots context you don't have access to, it tells you and offers a one-click request-access flow that goes back to the source tool. Access requests, agent reads, and API calls are written to the audit log, with who, what, the outcome, and when.Will Pulse train AI models on our data?
No. Pulse calls Anthropic Claude, OpenAI, and Google Gemini under API terms that bar training on the traffic we send; the providers may retain inputs briefly (up to 30 days) for abuse monitoring. Your content stays in your own space, every query scoped to your workspace's tenant id. Nothing is ever pooled across companies to train a model. The optional benchmarks feature shares only blurred percentile bands, and only once enough companies have opted in.What compliance posture does Pulse hold?
SOC 2 Type II is in audit prep, targeting 2027 H1. GDPR-aligned data handling, a DPA available on /dpa, and the sub-processor list on /legal/subprocessors. ACL mirroring, audit logging, BYOK for Anthropic, and per-company key isolation are enforced in code, not just promised in policy.What happens when an agent action goes wrong?
Five minute undo on most external writes: Pulse calls the source tool's own delete to retract the message, comment, or page. A few create-only writes have no clean delete in the upstream API (a new Linear issue, for instance) and are flagged non-revertible up front rather than promising an undo that can't happen. Admins can also restrict each action kind to an allowlist (the recipient's email domain for Slack DMs, the channel for channel messages, the repo for GitHub, the team for Linear); when one is set, every write is checked against it before it executes. Rate limits cap how much can happen at once. After 5 minutes the log entry is permanent.Can a custom agent read data it shouldn't?
No. A custom agent reads only what its creator can see, set by an explicit grant: a team, a Slack channel, a source type, or a single document. It's fail-closed, so an agent with no grant reads nothing, and a grant can never widen access past what the person who set it can see. Reads are checked against each document's own permissions at the moment the agent reads, and a sub-agent inherits the intersection of every agent in the chain. Scope filters narrow which topics an agent searches, but the grant is the permission boundary.Can we bring our own model key?
BYOK Anthropic ships today on the Enterprise plan. Paste a key from /app/admin/byok and every Claude call (answering, decision extraction, the devil's advocate, drafting) runs through it. Keys are encrypted at rest with AES-256-GCM, decrypted only when a request is made, and only the last four characters are kept as a hint. Google Gemini, AWS Bedrock, and Azure OpenAI BYOK are self-serve too on Enterprise: add your credentials at /app/admin/byok, test the connection, and matching calls route to your own key or cloud.
Skills and AI features
What is a Skill?
A Skill is a scoped, reviewable procedure that does one job. Drafting a renewal note. Preparing a one-on-one. Writing a launch summary. Each Skill is a SKILL.md with explicit triggers, promoted through a DRAFT, SHADOW, ACTIVE lifecycle: in shadow mode Pulse compares the Skill's output against what your team actually did, and only a Skill whose match rate clears the bar goes live. Skills are versioned and exportable.Where do Skills come from?
Two paths. (1) Auto-extracted: Pulse observes a recurring pattern in the workspace and compiles the steps into a draft SKILL.md. A human reviews and promotes it. (2) Hand-authored: a teammate writes the Skill from scratch on the Skills surface. Both paths produce the same artifact and go through the same shadow-mode match-rate gate before activating.How accurate is Pulse's AI?
Its confidence is tuned per workspace and per topic. When Pulse says 87 percent confident, it's right about 87 percent of the time on that kind of question in your workspace. It re-tunes every week from your thumbs up and thumbs down. And it marks every passage as either backed by a source or inferred, so you can always tell the difference.Does Pulse work with Claude Desktop and Cursor?
Yes. Pulse follows the Anthropic Agent Skills standard. SKILL.md tarballs work in Claude Desktop and Cursor today. Pulse also runs a native MCP server with fifteen read tools (search_company, get_decision, find_expert, get_pulse, the on-demand workspace snapshot, plus the filesystem-style pulse_ls / pulse_tree / pulse_grep / pulse_glob, and more), seven opt-in capture tools, three personal-memory tools (pulse_remember / pulse_recall / pulse_forget, private to you), and the propose tools pulse_propose_action + pulse_propose_config, which draft actions and config changes into the human approval inbox without executing anything, all callable from any MCP-compatible client. Auto-capture is available via a Claude Code SessionEnd hook or a background daemon, and auto-captured sessions stay private to you.Can a team write their own custom agents?
Yes, and not just admins. Any workspace member can build and manage their own agents at /app/agents: a trigger, a prompt, the action it proposes, and an explicit grant for what it's allowed to read. A member's agent is bounded to that member's own access. Agents draft into the approval inbox; auto-execute stays owner-only. Admins still manage workspace-wide agents and policies under /app/admin/agents and /app/admin/action-policies.Does Pulse track per-person productivity?
It never computes a productivity score or a ranking for a person, watches typing speed, or reads anyone's screen. What it does show is work context: a person's page carries what they own, what they've touched across tools, and an activity heatmap, and that exact view is visible to every workspace member equally, including the person it describes. Transparency, not a manager-only lens. Sentiment is team-level only, suppressed below five senders; no individual's sentiment is shown to anyone. Recognition (karma) has no rankings. Managers do get a work-state briefing of their direct reports (what each owns, owes, is blocked on, and is deciding, plus the manager's own private 1:1 notes), scoped to what the manager can already see and carrying no scores or trends: a memory aid for 1:1s and unblocking, not a measurement tool. Stewardship, the opt-in feature that keeps facts correct on an owner's watch, shows workspace totals only, and a test fails the build if a per-person breakdown is ever added.
How Pulse compares
How is Pulse different from Glean?
Glean is enterprise search, optimized for an organization of thousands with a search-bar shaped problem. Pulse is a process graph for teams of 5 to 500, built for remembering decisions, answering with honest confidence, and acting only with approval. If a team has tens of thousands of documents and wants a search box, Glean is a strong choice. If a team has decisions to remember and work to act on, Pulse is the right shape. Full comparison at /vs/glean.How is Pulse different from Notion AI?
Notion AI is great inside Notion, and on plans with connectors its search reaches tools like Slack, Drive, Jira, and GitHub. What comes back is search results; it does not keep decisions or commitments as structured records with rationale, owners, and citations. Pulse pulls from all the sources a team actually uses and connects them into a process graph. Full comparison at /vs/notion-ai.How is Pulse different from Coworker AI?
Coworker AI is an enterprise agent platform with its own organizational memory and autonomous agents. Pulse is a memory and action layer where every external write waits in a human approval inbox, and any MCP client (including a chat companion) can call it. The honest difference is the execution model: draft-then-approve versus agents that run on their own. Full comparison at /vs/coworker.Why not just build this on top of ChatGPT or Claude?
The AI model is just one of five steps. The hard work is in the other four: retrieval that respects ACLs at the document level, ranking it by relevance and recency, putting a citation on every sentence, and scoring confidence in a way that's specific to your workspace and topic. A bare AI call gives a confident-sounding answer with no audit trail and no way to say it isn't sure.
Getting started
How long does setup take?
Three phases. About 30 minutes of conversational interview with a founder (moving into the product as guided onboarding), so it isn't empty on day one. About 10 minutes of OAuth handshakes for the connectors that match the team's stack. 2 to 6 hours of backfill in the background before answers are dense. Useful answers land within minutes for recent activity; the long tail of historical decisions takes the rest of the first day.What does a Pulse rollout look like?
Three pilot tracks are available. Knowledge first (read-only, briefings + Ask, no actions) ships in week one. Action layer (drafting Slack DMs, Linear tickets) enables in week two with explicit allowlists. Custom Skills (codify a team's recurring rituals) ship over the following two weeks. The whole rollout fits inside a one-month pilot with a clear success metric agreed up front.Do we need to migrate any data?
No migration. Pulse reads from existing tools via OAuth. The source of truth stays in Slack, GitHub, Notion, and the rest. If a team disconnects Pulse, the source tools are untouched and Pulse's mirrored copy is deleted on a 30-day soft-delete grace (or immediately on request).How does support work?
Weekly office hours and email at support@pulsehq.tech, staffed by humans, not a model. First-response times depend on your plan: Pro is 8 hours in business hours, Enterprise is 2 hours, with a 1-hour priority track on Enterprise that triages by severity (P1 within 1 hour, P2 within 4, P3 within one business day) and 24/7 phone. Full channel details at /help.
Still have a question?
Office hours every Tuesday at 17:00 UTC and every Thursday at 09:00 PT. Or grab a slot from /contact and we'll answer in writing within one business day.