1. Market size: How many teams actually need self-hosted vs. managed? (Feels like the TAM is smaller than serverless alternatives) 2. Positioning: Are you targeting "we can't afford Kafka" or "we can't trust cloud"? Different buyer psychology. 3. Next moves: Thinking about protocol support beyond Kafka? Or doubling down on that niche? Ask because I'm running validation with 4 frontier models (GPT-5, Sonnet, Opus, Gemini Pro) on infrastructure plays—helps founders pressure-test positioning before they build. If you want a second opinion from actual ML systems instead of just your co-founder's bias, happy to help. No strings attached.
TypeStream looks solid. Good luck with the launch.
Managed is an opportunity for sure, but it's complex for a boostrapped solo dev like myself. Supporting Helm charts and reference docker-compose files would be a start I think.
For the Kafka deployment, it a lot of places where I worked, to get Kafka streams, it was never just Kafka. It's say MSK, Schema Registry, tying those together, config-as-code for the topics, and setting this up in the local dev environment. That was a major pain to advocate for the first time this the benefits of streaming would be obvious.
3. For Typestream, Kafka is abstracted away pretty much entirely so I'm actually not positioning this towards people who already have Kafka deployed heavily. It's for those who do dual writes in their code, writing cronjobs for syncing, and have other secondary datastores.
I'm very much testing out the positioning by talking to whoever I can! What are you thinking?