boost.ai Launches Presis, A Self-Hosted AI Model Built for Speed & Accuracy in Regulated Industries

The fine-tuned classification model now powers agent routing, scope, and safeguard decisions across the boost.ai platform, holding steady even during peak-demand hours.

SANDNES, Norway, Nov. 10, 2026 — boost.ai, a global leader in conversational AI for regulated industries, today announced the launch of Presis, a fine-tuned, self-hosted language model that handles the classification decisions in and around every AI agent conversation, including which part of the AI agent should respond, whether a question is in scope, whether an answer is safe to show to an end-user, and how conversations are reviewed once they've finished.

Based on internal testing, Presis is at least 2.5 times faster than the models it replaces, with the difference most noticeable during peak-demand hours, when third-party capacity is most likely to slow down, and in voice AI deployments, where routing and safeguard checks have to complete before a response can even begin.

"For regulated organizations, every automated conversation matters – speed and reliability are core to how they build trust with their customers," said Jerry Haywood, CEO of boost.ai. "Presis reflects what we've learned from a decade of working inside those industries, and it gives us one more way to earn the trust they place in us every day."

Boost.ai has spent a decade deploying conversational AI inside banks, insurers, and other highly-regulated organizations, environments where an incorrect or delayed answer can result in a compliance problem, not just an inconvenience. That experience shaped the decision to give customers a dedicated model for the decisions inside every conversation, rather than relying on general-purpose, third-party models designed to do everything for everyone.

"A model tuned specifically for classification can do things that a general-purpose model can't, including staying fast under peak load and improving steadily because we control every part of the stack it runs on," said Rasmus Hauch, CTO of boost.ai. "With Presis, we can now offer faster, more consistent answers where it counts, while building the reliability that customers expect from boost.ai even deeper into the platform."

Because Presis runs on infrastructure boost.ai controls end-to-end, routing, scope, and safeguard decisions are no longer exposed to the outages, timeouts, and capacity swings that can affect third-party providers. Classification costs are lower too: on average, internal testing sees a 33 percent reduction in token costs, and Presis is more accurate than the models it replaces on all five classification features it powers, by 0.5 to 7.9 percentage points depending on the feature.

Presis is rolling out across the boost.ai platform in the coming weeks. The company is also evaluating a similar approach for generative features in the future.

SOURCE boost.ai