Documentation

Getting started with Cognify

Cognify is the layer between your AI conversations and your organization's knowledge. This guide takes you from zero to your first structured, cited knowledge base in about five minutes.

Quick start

Create your account

Sign up with email and password. An organization and a default workspace are created for you. Multi-factor authentication (TOTP) can be enabled from Settings.

Import a conversation

Go to Import and paste any AI conversation — Cognify auto-detects the provider and message boundaries. Or upload a file:

conversations.json from a ChatGPT data export
Claude conversation exports
Markdown or plain-text transcripts

Duplicate imports are detected by content hash — you'll be linked to the existing conversation instead of creating a copy.

Run analyses & parsing

Open the conversation and pick from 20+ analyses (summaries, decisions, risks, requirements, PRD, meeting minutes…) in the AI Processing tab, or run the non-interpretive Structural Parsing engine. Every extracted item is cited back to the exact source message — click a citation chip to jump to it.

Promote, search, export

Promote extraction results into first-class knowledge items, explore them in the knowledge graph, find anything with hybrid search (⌘K), and export conversations with their analyses to markdown, JSON, HTML or CSV.

REST API

Everything in the UI is available over a versioned REST API at /v1. Authenticate with a bearer token or a scoped API key (create one in Settings → API keys). Org-scoped routes take an X-Org-Id header. The full OpenAPI spec is served at GET /openapi.json.

Endpoint	Description
POST /v1/auth/register	Create an account + organization; returns access & refresh tokens.
POST /v1/conversations/import/paste	Paste any AI conversation; provider & format auto-detected.
POST /v1/conversations/:id/analyze	Queue AI analyses (executive summary, decisions, risks, PRD…).
POST /v1/conversations/:id/parse	Queue structural parsing (topics, Q&A, entities…) with citations.
GET /v1/search?q=	Hybrid lexical + semantic search across conversations & knowledge.
GET /v1/knowledge/graph	Typed knowledge graph: nodes and edges across all conversations.
POST /v1/conversations/:id/export	Export markdown, JSON, HTML or CSV — inline content response.

Example: import via API

curl -X POST "$API_URL/v1/conversations/import/paste" \
  -H "Authorization: Bearer $ACCESS_TOKEN" \
  -H "X-Org-Id: $ORG_ID" \
  -H "Content-Type: application/json" \
  -d '{
    "workspaceId": "<workspace-id>",
    "title": "GTM strategy session",
    "raw": "User: How should we price the Team tier?\nAssistant: ..."
  }'

API keys are scoped and hashed at rest; the full key is shown exactly once at creation. Rotate keys from Settings → API keys.

Ready to build your memory layer?

Free tier — 10 conversations a month.

Get started