Nastia maintains an internal platform layer that powers the consumer product. We are working towards opening parts of it to a small number of research collaborators and commercial partners. There is no general-availability public API at this time.
What we may offer
The platform exposes primitives that have proven hard to build well from scratch:
- Long-form memory
Persistent, retrievable, structured memory across sessions, with policies for what to retain and what to forget. - Real-time multimodal generation
Streaming text, voice, and image synthesis with sub-second time-to-first-token latency. - Routing layer
Per-turn selection from a long tail of specialised conversational models. - Safety filters
Domain-specific safety layers tuned for social AI workloads.
Who we work with
We partner with research labs, universities, and a small number of product teams whose work overlaps with ours. We are most interested in collaborations that benefit users of the consumer product, advance the state of the art in companionship AI, or stress-test our safety and alignment work in new domains.
Apply for access
If you have a use case for the platform, write to [email protected] with a description of what you would build, the data you would touch, and the expected volume. Include a link to your team or your prior work. We answer when there is something to say.
Roadmap
A public, self-serve API is on our roadmap. We do not yet have a date. We prefer to release fewer things, well, than to ship a public API before the platform is ready for one.