Sqwish Bets on Customer Data Customization as Decision Model Use Surges
- Published
- Source
- TechCrunch
More than one in four requests to Vercel’s AI model router went to decision models in October, according to the article. These models use an LLM’s general intelligence to choose among preset options without generating output tokens, and can report confidence in their choices. Sqwish, a U.K.-based startup developing the kettei fat decision model, announced a $3 million pre-seed round. Co-founders Ushnish Sengupta and Federica Freddi said they had adapted architectures originally developed for small LLMs after TypeSafe’s Jev models emerged. Microsoft also announced that Microsoft-Decision-1.1 was rolling out for wide use on Azure immediately.
The article identifies data calibration and performance in real operating conditions as central challenges for decision models. TypeSafe CEO Diogo Almeida said statistically adjusted synthetic data was the hardest part of building Jev. Sqwish also uses synthetic data, but differentiates its offering with post-training tools that let customers use their own production data to improve model judgment on specific tasks. The approach positions customization as a potential way for a smaller developer to compete with hyperscalers and frontier-model providers.
Sqwish’s stated test is whether kettei fat performs in customers’ specific contexts and improves through continued tuning; the article says it has performed well on some benchmarks but provides no scores. Microsoft-Decision-1.1 and Jev were described by Microsoft engineering VP Achint Srivastava as performing strongly across accuracy, reliability, speed, and cost efficiency, including on held-out benchmarks. No further timeline was disclosed at the time of publication.
Original report
As decision model use skyrockets, is customization the next frontier?
TechCrunch · Tim FernholzThis briefing is an independently written summary based on publicly available reporting and is provided for industry information and news discovery. The original report and source publication are credited and linked where applicable. RobotToday does not claim ownership of third-party source material.
Rights concerns? Contact [email protected] with the relevant URL and details. We will review the matter and take appropriate action where warranted.
More related news
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
-
11
RobotToday Weekly November 9–13, 2026 RobotToday · 3 days ago
- 12
Related Suppliers
Microsoft Corporation (Robotics)
Microsoft provides robotics developers ROS-on-Windows support, the AirSim open-source simulation platform, and Azure cloud services for robot fleets and AI.
ZhiCheng AI (Hangzhou) Technology Co., Ltd. 智澄英达(杭州)科技有限公司
Hangzhou humanoid robot maker of the TR-series (TR4/TR5/TR6) and open-source Chengling world model.
Wayve Technologies Ltd.
London autonomous-driving company founded 2017 by Amar Shah and Alex Kendall; develops end-to-end embodied-AI driving software, valued at $8.6 billion.
AZURE Photonics USA Inc.
Manufacturer of precision optical components and machine vision lenses for industrial automation, robotics, and inspection applications.
Azure Printed Homes
California maker of robotically 3D-printed modular studios, ADUs and homes from recycled plastic composite, with 100+ structures delivered.
Woven by Toyota
Tokyo-based, Toyota-owned developer of automated driving software and the Arene vehicle platform, with about 2,500 staff and Woven City test site.
Wuxi Longsheng Technology Co., Ltd. 隆盛科技
Wuxi auto-parts maker (EGR systems, motor cores) building a robot business via subsidiary Weihan's harmonic reducers and joint actuators.
FMC³ Robotics
Ingolstadt, Germany embodied-AI firm whose RoboBrain software and DATOS haptic-glove kit run on partner robots; MoU with Vector and PEM Motion in 2026.
Synthium
Synthium is a Singapore-based, Antler-backed startup that crowdsources gamified simulations to generate training data for humanoid robot AI models.
Antioch
New York simulation software firm founded 2025; its cloud platform calibrates digital twins of robots and runs thousands of parallel tests, $40.5M raised.
Cambrian Inc.
AI 3D-vision system that gives robot arms 6D pose estimation and path planning for bin picking, assembly and cable handling.
Mecka
US-based Mecka sells first-person human activity data and tools for robot learning, plus robot deployment support; USD 60M Series B led by Sequoia in Oct 2026.
Robotics needs a service framework.
RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.
Join the RobotToday community on LinkedIn
Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.