Startup Sqwish raises $3 million in pre-seed funding for decision model kettei fat
British startup Sqwish has raised a $3 million pre-seed investment to develop the decision model kettei fat, which offers companies tools for fine-tuning on their own production data.
British startup Sqwish has announced a pre-seed investment round of $3 million to develop a decision model called kettei fat. The company was founded by Ushnish Sengupta and Federica Freddi, who met as postgraduate students at Cambridge. According to the source, they were developing small language models even before the Jev model from TypeSafe appeared, and later reworked their architecture into a decision model.
According to the description in the source, decision models are language models that do not generate output text, but are fine-tuned for fast and cheap decision-making among predefined options while also stating a confidence level for their choice. They are used in production environments where the speed of a decision matters more than the detailed reasoning provided by slower and more expensive frontier models.
According to founder Sengupta, Sqwish differentiates itself by offering built-in post-training tools that allow customers to improve the model's judgment using their own production data, similarly to how the company trains its own model. Sengupta stated that the company views benchmarks more as "an invitation to try the model" than as proof of quality, since real verification only comes with deployment in a specific context.
Why it matters
For companies considering automation of fast decisions in production systems, a new vendor is emerging that offers tools for fine-tuning the model on their own data, which could reduce costs compared to deploying more expensive frontier models for simple tasks of choosing among predefined options.
Relevant practical impact
What this means
For a business
The emergence of a new category of "decision models" offering continuous fine-tuning on companies' own production data expands the range of tools available for cheap and fast automated decision-making, without having to rely on more expensive frontier models.
DevelopmentCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.