AI agent instruments promise to siphon out a few of the drudgery from each day workflows, however most organizations are hesitant to undertake them but, harboring a urgent concern: knowledge safety. Massive enterprises with commerce secrets and techniques, corporations in extremely regulated industries, and authorities businesses have thought greater than twice about bringing in AI instruments out of concern that their — or worse, their prospects’ — knowledge might inadvertently be compromised, or used to coach basis fashions.
Canadian AI agency Cohere is taking purpose at assuaging these issues with its new AI agent platform dubbed North, which guarantees to allow non-public deployment in order that enterprises and governments can preserve their and prospects’ knowledge protected behind their very own firewalls.
“LLMs are only as good as the data they have access to,” Nick Frosst, co-founder and CEO of Cohere, mentioned throughout a demo of North. “If we want LLMs to be as useful as possible, they have to access that useful data, and that means they need to be deployed in [the customer’s] environment.”
As a substitute of utilizing enterprise cloud platforms like Azure or AWS, Cohere says it may well set up North on a company’s non-public infrastructure in order that it by no means sees or interacts with a buyer’s knowledge. North can run on a company’s on-premise infrastructure, hybrid clouds, VPCs, or air-gapped environments, Frosst mentioned.
“We can deploy literally on a GPU in a closet that they might have somewhere,” he defined, including that North was designed to run on as few as two GPUs.
Cohere claims North additionally contains safety protocols like granular entry management, agent autonomy insurance policies, steady red-teaming, and third-party safety exams. And, it meets worldwide compliance requirements like GDPR, SOC-2, and ISO 27001.
Greater than non-public deployments
Cohere, which has up to now raised $970 million, most not too long ago at a $5.5 billion valuation, mentioned it has already piloted North with some prospects comparable to RBC, Dell, LG, Ensemble Health Companions, and, as TechCrunch reported final yr, Palantir.
Techcrunch occasion
San Francisco
|
October 27-29, 2025
North mirrors many AI agent platforms proper out of the field. Its chief options are chat and search, which let customers get solutions to buyer assist inquiries; summarize assembly transcripts, write advertising copy, and entry data from each inside sources and the net. Frosst added that every one responses embrace citations and “reasoning” chains of thought so staff can audit and confirm the output.
The chat and search features are powered by present Cohere expertise, like Command (its household of generative AI fashions), and Compass (its multimodal search tech stack). Frosst mentioned North is powered by a variant of its Command mannequin that’s educated for enterprise reasoning.
“It goes beyond just Q&A and gets into doing work for you. So, [North] has a bunch of asset creation. It can make tables, it can make documents, it can make slideshows. It can do a bunch of market research,” Frosst mentioned.
It’s value noting that in Could, Cohere acquired Ottogrid, a Vancouver-based platform that develops enterprise instruments for automating high-level market analysis.
Like different AI agent platforms, North can hook up with present office instruments like Gmail, Slack, Salesforce, Outlook and Linear, and combine with any Mannequin Context Protocol (MCP) servers to entry industry-specific or in-house purposes.
“As you build confidence by chatting to the model, there’s like a smooth transition that happens between using this as an augmentation to using it as an automation,” Frosst mentioned.
