boost.ai Presis moves agent routing and safety decisions to a self-hosted model
boost.ai has announced Presis, a model hosted on its own infrastructure for routing, scope, safety and post-conversation review classifications in conversational agents.
AT A GLANCE
Key points
- Presis classifies routing, scope, safety and conversation-review decisions.
- The model is hosted on infrastructure controlled by boost.ai.
- A phased platform-wide rollout is planned for the coming weeks.
What is changing?
boost.ai introduced Presis on November 10, a fine-tuned, self-hosted classification model for conversational agents in regulated industries. Presis handles the classification steps that determine which part of an agent should receive a request, whether a question is in scope, whether an answer is safe to show to an end user, and how completed conversations should be reviewed.
According to the company’s internal tests, the model is at least 2.5 times faster than the models it replaces, with accuracy gains of 0.5 to 7.9 percentage points depending on the classification feature and an average 33% reduction in token costs. These are provider-reported measurements. Presis is scheduled to roll out across the boost.ai platform over the coming weeks.
Explore the original sources
Source date: 10 November 2026.
Published by AI Radar: 10 Nov 2026, 15:06 (Istanbul, UTC+3).
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.
Loading comments…