Root out false information before agents build on it
Wrong information spreads through a company like a poisonous seed. Knowledge Health finds it across your docs and code before your team and your agents build on it.
Notes from Falconer
Essays, announcements, and research from the team behind Falconer.
Documentation has a new purpose
Agents now handle the step-by-step tasks docs used to explain, so documentation has to serve people and agents at once. Public Docs, in beta now, publishes customer-facing documentation from the same knowledge base your team already uses internally.
Rolling out the new Falconer knowledge platform
Falconer gives you and your customers complete, specialized intelligence: an internal knowledge base and external documentation in one beautiful package.
Falconer is launching a new usage-based pricing plan
One shared credit balance for AI features, Knowledge Health, and paid public-docs add-ons gives teams more flexibility and a clearer view of spending.
Knowledge as code
Only one breakout AI use case is making a significant impact on knowledge work: coding, because code is versioned, tested, and machine readable. Treating company knowledge the same way, centrally stored, versioned, and observable, is how the rest of the organization gets that same leverage.
Spaces: focused work, connected knowledge
One knowledge tree gets noisy as a company grows, but separate workspaces create silos. Spaces give each team a focused view while preserving the context, permissions, and decisions that connect everything, with no migration required.
Why everything breaks at 150 people
Cerebras's post on building their internal knowledge base hit 2.5M views. From someone who led knowledge management at Stripe and Uber: why knowledge bases break around 150 people, what a great one looks like, and how to roll one out.
Docs-as-code as simple as Google Docs
Customer guides, onboarding flows, API references: they live in the codebase, where the people who write them can't easily edit. So they keep a working copy somewhere else and maintain two versions of the same thing, forever.
Search, rebuilt: how Falconer finds the right answer
Company search is hard: the corpus is too large to browse and too small to learn ranking from. Here's how Falconer's search page and answer engine find the right answer.
Agent loops are fine, retrieval is what you're paying for
Inefficient retrieval is burning your token budget. When the first search comes back incomplete, context grows and the bill follows.
Skill documents: shared team instructions for every AI tool
Falconer Skill documents let teams create, share, manage, and reuse AI instructions across Falconer and MCP-connected tools like Claude, Cursor, and Codex.
Stop freeloading off open source
Most teams quietly work around the rough edges of the open source libraries they depend on. We decided to fix them at the source instead, contributing patches upstream to Tiptap, the framework behind Falconer's editor.
How to generate a changelog in 10 seconds
Changelogs used to take hours each week to assemble across GitHub, Linear, Slack, and meeting notes. Falconer turns the same week of context into a changelog in seconds — already cited, audience-aware, and ready to send.
Falconer agent now speaks git
The Falconer agent can now answer questions about your code's history: who wrote what, when, and why. It calls git directly against your connected repos to do code archaeology, release diffs, regression hunts, and ownership lookups, all in a chat.
Ditching the agent sandbox for AWS S3 Files
How we gave the Falconer agent git access (log, blame, diff between arbitrary refs) by mounting a shared NFS filesystem backed by S3 Files across our ingest and UI services. A walkthrough of the storage choice, the Pulumi quirks, and how we made the repo sync robust and reliable.
How others build agent memory, and what I took from each
ChatGPT, Claude Code, and Letta have each built production memory systems for AI agents. Looking at the differences shaped how I built agent signals for Falconer.
Knowledge Health: observability for your knowledge base
Every company runs on written knowledge, and almost none of them know how healthy theirs is. Knowledge Health puts a single, live score on it and shows you exactly what's dragging it down: contradictions, stale docs, coverage gaps, and redundancies.
Agent Personalization: an agent that knows you
Falconer now tailors every response to who you are. Agent Personalization builds a lightweight, transparent profile of your role, team, and preferences — fully editable, with every attribute traced back to its source.
Your context is more than training data
Everyone has access to the same frontier models. Your competitive advantage is institutional context — the decisions, tradeoffs, and battle scars inside your four walls. Here's why curating that context is now existential.
Falconer Update: Full self-driving docs
Falconer Update keeps your documentation in sync with your codebase automatically. Toggle it on for any document and choose Review mode for human-in-the-loop edits, or Full Self-Driving mode to let Falconer handle it entirely.
Falconer Generate: from repo to doc set in minutes
Falconer Generate turns a connected GitHub repo into a structured documentation set, helping teams get documentation started faster.
The source of truth for high-speed teams
Our mission is to capture all of your important context, keep it up to date, and make it easy for you to deploy it wherever you want: your teammates, your customers, your coding agents.
Rethinking data ingestion as a DAG
How we reduced data ingestion time from hours to minutes by reimagining our pipeline as a directed acyclic graph. This post covers the architectural shift from async workflows to job queues, the migration strategy we used to preserve behavior, and the observability patterns that helped us identify and isolate bottlenecks at scale.
LLM-as-a-Courtroom
How we built a multi-agent courtroom simulation to decide when code changes require documentation updates—and why the legal system is humanity's best framework for binary decisions under uncertainty.