Why we backed Display.dev
- Jul 3
- 2 min read
Display.dev founders Ott & Carl (from left)
Work is shifting from documents to artifacts — and those artifacts will be created across many different models and tools. Here's where our thesis and this investment meet.
Most of the value AI creates today never leaves the agent. A model drafts a document, writes a spec, pulls the data — and it sits in a terminal or a temp folder, invisible to the team that actually needed it. That problem is where our thesis and this investment meet.
Work is shifting from documents to artifacts, and those artifacts will be created across many different models and tools.
The old unit of knowledge work was the document a person wrote. The new unit is the artifact an agent produces, often with a human giving direction rather than typing every word. That shift is already underway. What's missing is the infrastructure around it — a place for the output to live, be reviewed, and improve.
How early we are
The numbers show how early we are. The distance between people chatting with AI and agents doing real work is going to close quickly. As it does, companies will be buried in agent-generated output with nowhere to put it and no way to give feedback on it.
~5M active users on OpenAI's Codex
~1B users on ChatGPT
No single lab wins forever
There's a second force we keep underwriting: no single lab wins forever. The best model changes month to month, and open source keeps closing the gap. Any serious bet on this layer has to be model-agnostic, or it ages out the moment the leaderboard moves. Display.dev is built that way from the start — a document created in Cursor can be edited in Codex and refined in Claude Code.
What the product does
What the product does is simple to describe. Every agent gets to publish a document with its own URL. Colleagues open it and comment, the agent reads those comments, and it publishes a new version in response. Google Drive, but for agents. Pricing is per company rather than per seat, which is the right shape for output that scales with compute, not headcount.
The team
Then there's the team. Ott and Carl spent years at Pipedrive learning how to build software teams actually adopt. And they had paying customers on day one — at pre-seed, that's the strongest signal I know that a need is real rather than theoretical.
We think the home for agent output will be one of the foundational layers of the next decade of software. Display.dev is building it. That's why we're in.
Outlast Fund I AIF KS is supported by ALTUM and co-financed by the European Union through the European Regional Development Fund (ERDF) as part of Latvia's National Development Plan 2021–2027.
Comments