Kerva is making product creation autonomous. All of our time goes into building real products and the agentic loops behind them: finding needs, shaping solutions, designing and shipping software, testing it in the world, and turning repeated work into workflows agents can run.
This page is where we share some of that work.
Each article is generated from topics discussed in our internal meetings. The goal is not to publish transcripts. Meetings are messy, detailed, and full of context that only matters inside the team. Instead, AI helps sift through the discussion, find the useful threads, and turn them into short notes that other builders may find helpful.
That feels like the right format for us. We are builders, and all of our energy goes into the work itself. But if we only build privately, we lose the chance to share what we are learning along the way. AI gives us a way to keep the overhead low: the meetings already happen, the raw material already exists, and the writing can be drafted without pulling the team away from the work.
The filter is simple. A topic is worth sharing if it contains a useful tradeoff, a mistake, a surprising result, a product decision, or a technical detail that another builder could apply. Some weeks that might be about reducing the cost of an AI workflow. Some weeks it might be about why an agent failed at testing. Some weeks it might be about killing an idea.
This is building in the open, but in a practical way. Not everything belongs outside the company, and not every update is interesting. The useful material is usually buried inside the work itself.
AI helps us dig it out.
Author: Kerva
