@mercury builds the best banking products for entrepreneurs. They also celebrate the entrepreneurs building those companies.
They interviewed @capitaltruist, about the vision and the messy challenges behind building @falconer_ai.
Full article below 👇
I co-founded and then left a 7-figure consultancy because it felt like painting the Golden Gate Bridge over and over.
When a unicorn client wanted to pay us a hefty sum to clean up their Notion workspace, I said no. "I can do it, but it'll quickly turn to shit again."
Their response: "I don't care. It's a trash fire."
This was the catalyst for starting Falconer. I talked to about 100 engineering managers, tech leads, and founders about their internal docs. They were customers of Notion, Glean, and Confluence. The top complaints were related to document lifecycle management. "What's the source of truth? How does a document get updated? How can I find it quickly?"
The vision for Falconer needed to be fundamentally different, not just a beautiful Confluence. It all came down to accuracy and curation.
Notion is the everything app. Glean helps you search everything. Confluence is...well I'm not sure what Confluence is for.
Falconer is the accuracy app, obsessed with knowledge quality and curation. The app that lets you find the right thing, not just return everything.
Falconer is:
1. A knowledge graph built for accuracy
2. An AI-native library that curates your knowledge for you
3. A context monitoring and observability tool
None of our tools, UX, or aspirations mattered until we showed results.
Customer love has shown up in our Slack Connect channels and internal usage dashboards, but we needed public numbers to prove it. It's validating to share the winning results of our unique approach and architecture. And there's still so much more to come.
the team can say RTFM because Falcon's already captured the answer. the manual stays alive whether anyone remembers to update it or not.
@falconer_ai 😎
this is so well phrased. how you build human capital at your enterprise is what will set you apart in the age of AI. context is key when everything else is so volatile.
incredibly proud of what we’re building at @falconer_ai helping companies multiply their human capital and capture context
Everything you'd want in a markdown editor and more.
Built to scale. AI-native. Context aware.
- Multiplayer comments: synced and resolvable
- Author states: edit, suggestion + view modes
- Edit history: diff and roll back
- Multi-edits
@payabli has built some of the best product docs in the world.
Their customers see it every day. Internally, they wanted to build the best writing culture, too. Knowledge was spread across Confluence, Google Drive, and Slack. Finding an answer often meant pulling an engineer out of their work.
Their move from Confluence to Falconer took just two days.
The impact showed up in new behavior. Teams that never had time for documentation — HR, finance, marketing — started writing more things down. And engineers now rely on Falconer instead of getting trapped in endless Slack threads.
writing a self-review usually eats a whole weekend. this time mine took ~15 minutes.
and it came out more thorough than anything i'd have written from memory.
@falconer_ai pulled from six months of changelogs, PRs, and incident reports, so nothing got left out. i was confident every contribution made it in, even the wins i'd forgotten i shipped. all of it cited.
Onboarding a new engineer can take weeks of explaining how the codebase works. And when a senior engineer leaves, that context walks out the door.
Falconer Generate fixes the handoff, and it doesn't take a documentation sprint.
Point Falconer at any repo and get an initial doc set back in minutes: project overview, get started, architecture overview, contributing guide, and glossary.
Falconer also answers questions about the codebase directly, so new hires ramp from the docs instead of your staff engineer's calendar.
i found a substack called sentimental being (@yearlyblues is it you???) and it’s gently been undoing something in me. only 8 posts but i keep coming back to them like a favourite blanket. it’s the kind of writing that wraps around you and makes you feel less alone in your own head. whoever you are behind these words, you’ve reached me, truly. thank you.
By now we've realized that Plan Mode gives us more reliable results with Claude and other coding agents.
But iteration on raw .md files is hard when you don't have an editor.
Being able to collaborate with agents and humans in the same space is the real unlock!!!
@falconer_ai i let a falconer skill prep my standup updates and now i show up sounding so on top of it that even i'm impressed. do i remember doing the work? no. does Falconer? yes.