Pulse

Product Introduction

1. Definition: Pulse is a permission-aware, AI-powered company memory and team intelligence platform. Technically, it is a graph-based knowledge management and agentic workflow system that integrates with a company's existing SaaS stack (Slack, GitHub, Notion, Linear, etc.) and AI tools (Claude, ChatGPT, Cursor) via MCP (Model Context Protocol).
2. Core Value Proposition: Pulse exists to eliminate institutional knowledge loss and operational friction by creating a single, trusted source of truth for a company's decisions, commitments, and expertise. Its primary value is transforming scattered, ephemeral team knowledge into a structured, queryable, and actionable "company brain" that accelerates onboarding, prevents rework, and surfaces critical work proactively.

Main Features

1. Permission-Aware Memory Graph: Unlike a simple document index, Pulse constructs a dynamic graph of connected entities—decisions, people, commitments, and projects—from activity across integrated tools. It replicates Access Control Lists (ACLs) from source systems (e.g., Slack channel privacy, GitHub repo permissions) to ensure users only see information they are already authorized to view. This hybrid retrieval system weighs semantic meaning, keywords, and recency.
2. Cited Answers with Honest Confidence: Every answer generated by Pulse includes inline citations showing the source (e.g., Slack thread, Notion doc, PR). It provides a confidence score (e.g., 84%) that is re-tuned weekly based on the team's own thumbs-up/down feedback, creating a per-workspace accuracy benchmark. Answers clearly distinguish between cited facts and AI inferences.
3. Proactive Surfaces & Agent Actions: Pulse continuously monitors integrated tools to surface stuck work (e.g., PRs awaiting review), stalled decisions, and upcoming commitments on a personalized Home feed. Its agent system can draft follow-up actions (Slack DMs, Linear tickets, kickoff docs) but places them in a human approval inbox, enforcing a mandatory review step with a five-minute undo window. Playbooks automate multi-step workflows.
4. Skills Compiler & Expert Finder: The platform analyzes real work activity (code commits, document edits, discussion participation) to implicitly map team member expertise. It can identify and rank the right person to answer a question or solve a problem, effectively reducing the "tribal knowledge" bottleneck and accelerating new hire ramp-up time.
5. Multi-Surface Integration: Pulse operates through multiple interfaces: a dedicated web app (app.pulsehq.tech), and deeply inside developer and AI tools via an MCP server. This allows teams to query the company brain directly from their IDE (Cursor) or preferred AI chat interface (Claude Desktop, ChatGPT), creating a seamless, context-aware workflow.

Problems Solved

1. Pain Point: Context Switching and Information Scatter. Knowledge workers lose significant productivity (cited as 9% of annual work time by HBR) toggling between an average of 10 apps daily, searching for decisions, and re-orienting after interruptions. Pulse consolidates context into a single, queryable layer.
2. Pain Point: Institutional Amnesia and Decision Debates. Critical decisions and their rationales are lost in ephemeral chat threads or outdated documents, leading to the same debates resurfacing months later. Pulse auto-captures decisions with source and rationale, creating a searchable decision library.
3. Target Audience: Engineering Managers & Tech Leads who need visibility into project bottlenecks and team expertise; Product Managers who require fast access to past rationale and feature planning context; New Hires who face a long ramp-up period (averaging 8 months); CTOs & Founders who need aggregated, real-time insights into company operations for reporting and strategy.
4. Use Cases: Accelerating New Hire Onboarding: Providing instant answers and 30/60/90 plans derived from existing company knowledge. Pre-Board Meeting Preparation: Automatically generating briefings on recent changes, stuck items, and promises due. Incident Response & Debugging: Quickly finding past failure cases, PR fixes, and internal experts for specific systems. Quarterly Business Review (QBR) Compilation: Using deep-research mode to synthesize a quarter's work, blockers, and velocity from raw data in minutes.

Unique Advantages

1. Differentiation: Versus Enterprise Search (Glean, Slack Search): Pulse is a proactive memory graph, not a passive search index. It captures "why" decisions were made and connects them to people and projects, and it drafts subsequent actions. Versus Individual AI Copilots: Pulse provides a shared, permissioned team memory that improves collective intelligence, whereas individual AI tools start each session with a blank slate and no shared context.
2. Key Innovation: The Process Graph Architecture. Instead of treating information as isolated documents, Pulse models the relationships between decisions, commitments, work items, and people. This allows navigation and retrieval based on process and causality, not just keyword matching. Combined with its strict ACL-gated hybrid retrieval and feedback-tuned confidence scoring, it creates a system of record that teams can actually trust for operational decisions.

Frequently Asked Questions (FAQ)

1. How does Pulse handle data privacy and permissions? Pulse operates on a strict principle of permission replication. It mirrors the ACLs (Access Control Lists) from connected tools like Slack, GitHub, Google Drive, and Confluence. If a user cannot access a Slack channel or a Google Doc in the original tool, they cannot see that content in Pulse. All answers and search results are filtered in real-time through this permission layer.
2. Can Pulse take automated actions without human approval? No. A core tenet of Pulse's design is "no silent auto-execution." Any external write action—sending a Slack message, creating a Linear ticket, updating a Notion page—is first drafted and placed in a user's approval inbox. A human must explicitly approve each action, and a five-minute undo window is provided after sending.
3. What AI models does Pulse use, and is my company data used for training? Pulse uses large language models (LLMs) from providers like Anthropic (Claude) and OpenAI via API. Under these API terms, the providers do not use customer data for model training. Pulse's architecture keeps customer data isolated per workspace. The company explicitly states it does not train its own models on customer data, a policy reinforced in its contract and security documentation.
4. How long does it take to set up and see value from Pulse? Setup involves three phases: a ~30-minute founder-led interview to seed context, ~10 minutes of OAuth connections to core tools, and a 2-6 hour background sync. Users can get useful answers about recent activity within minutes, while the full historical backfill and graph construction completes within the first day.
5. What is the "Pulse Score" shown on the dashboard? The Pulse Score (e.g., 82/100) is a high-level health metric for the team or projects being monitored. It is likely a composite indicator derived from factors like decision velocity, commitment completion rates, and stuck item resolution times, providing an at-a-glance view of operational flow.