Framer AI AgentsDesign and publish professional sites with AI
Pulse
Your company's permission-aware, proactive and agentic brain
70 followers
Your company's permission-aware, proactive and agentic brain
70 followers
Most AI tools start each session blank and answer with no way to check them. Pulse is your company's brain. It connects Slack, email, docs, meetings, and 25 tools into one permission-aware memory, then answers anything with a citation on every line. Decisions get captured once and never re-explained. It finds the right expert, runs agents you approve, builds reports from real data, and works by voice. And it lives inside Claude, Cursor, and ChatGPT over MCP.
Payment Required
Launch Team / Built With
Pulse
Pulse
@apoorvjain25 Amazing launch , Checking it soon
Pulse
@priya_kushwaha1 Thank you, means a lot.
decisions that outlive who made them is the real bet. when one gets reversed, does the old call retire, or stay cited so you see the shift?
Pulse
@andrewzakonov
Love this question, because "decisions that outlive who made them" is close to word-for-word why I built this.
The old call never retires. Every decision stays in the archive permanently with its evidence trail, who made it, and links back to the original threads. When a new decision contradicts an old one, Pulse detects the contradiction (a nightly job scans for exactly this), and if you ask about the topic it shows both views, marks which one is current, and explains how it decided: recency, authority, and specificity, in that order. So you see the shift, not just the latest answer.
The reversal itself becomes useful history too. Each decision carries an outcome-check date, and when it arrives Pulse re-evaluates the call against what actually happened and surfaces "this aged badly" so the lesson lands. And when someone drafts a similar decision later, the devil's advocate feature argues against it citing your own past reversals.
Retiring old calls silently would throw away the most expensive thing a team produces: the record of what it used to believe and why it changed. Pulse keeps the whole arc.
A proactive company brain sounds useful only if it stays permission-aware when context crosses teams. The tricky case is when an answer mixes Slack, docs, and meetings: can users see why Pulse surfaced something and which source they are allowed to inspect?
Pulse
@yaroslav_stelmakh
Pulse filters before synthesis: your permission scope is rebuilt on every query, so only documents you could open yourself ever reach the model. That means in a mixed Slack-docs-meetings answer, every citation is inspectable by you, by construction, and each sentence links to its source. Two people asking the same thing get different answers scoped to what each can see, which we think is correct behaviour for a shared brain.
A citation on every line is what I'd judge this on. Company knowledge tools die the moment someone gets confidently told something wrong, and after that nobody opens them again. Permission-aware retrieval is the other half of that, and it's much harder than it sounds. Does the citation point at the original message, or at Pulse's stored copy of it?
Pulse
@rohanmanthani You're judging it on the right thing.
Honest answer: Pulse cites from its synced copy, not the live message. Anything that indexes your tools has to. The real question is whether the gap between that copy and the original is hidden or visible, and we made it visible.
Every citation shows when Pulse last verified that source, down to "under an hour ago" vs "a day ago." And one click on it re-reads the live source right then and shows you what changed since. So you're never trusting a stale copy without knowing it.
And yeah, permission-aware is the harder half. It's enforced before retrieval, not after: your access is rebuilt on every query, so anything you couldn't open yourself never makes it into the answer. Getting that wrong even once is the "nobody opens it again" problem you're describing, which is exactly why I was paranoid about it from day one.