Rather than joining the global push to build more AI hardware, Backboard.io is betting on doing more with what already exists while ensuring data remains within customer environments. The company introduced four new developments designed to tackle rising AI costs, data exposure to external providers, and the growing use of unauthorized tools. Built in Nepean within Prime Minister Mark Carney’s riding, the platform was created by a team entirely composed of graduates from Canadian post-secondary institutions.
BackboardQuant: more AI per GPU
BackboardQuant (BBQ) compresses AI models by up to 70% with functionally no loss in quality. In testing, compressed models retained functionally full-precision performance while running up to 2.7x faster. The result is one GPU doing the work of two or three, easing the data-centre crunch and lowering cost to serve. Developed in Backboard’s research lab, it ships built into the company’s enterprise deployments.
Backboard Studio: frontier coding for up to 90% less
Backboard Studio matches or beats the leading coding tools from the largest AI labs at up to 90% lower cost. On Terminal-Bench 2.1, the independent standard for agentic coding, Backboard Studio scores 79.8% running Claude Opus 4.8, ahead of the frontier labs’ own results on the neutral public harness (GPT-5.5 at 78.2%, Opus 4.8 at 74.6%). Running the open-source GLM 5.1, it clears 70%, frontier-class performance with no proprietary model required. A built-in token optimizer cuts frontier model usage by up to 30% on like-to-like comparisons. It runs on cloud or self-hosted, so proprietary code never leaves the building.Â
Nash: thousands of AI models in one app
Nash, gives users access to thousands of AI models across text and images in a single chat app, while keeping their memory out of the models’ hands. For enterprises, it is the direct answer to shadow AI: one sanctioned app, better than what employees would find on their own. Available now for consumer and enterprise use.
World #1 in AI memory
Backboard ranks number one on LoCoMo and LongMemEval, the two leading independent tests of AI memory. The results are public and reproducible: LoCoMo, LongMemEval. With Backboard, that memory stays inside the customer’s environment, under their control.
The mandate is there, from both the public and the private sector. They want a Canadian option they can trust. We are here, and you do not need to wait. For too long the story was that Canadians don’t buy from Canadians. We don’t believe that is true. People simply hadn’t heard about us. So here we are. Major enterprises around the world are already working with us. There are no more excuses. Work with us.
Rob Imbeault, Co-founder CEO of Backboard.io, formerly co-founder of Canadian SaaS Unicorn, Assent.
Sovereign by design
Backboard’s full stack, the API, the application layer, and the models, can run inside a customer’s own cloud. Data never leaves and the system runs under the customer’s governance. For governments, hospitals, banks, and critical infrastructure, sovereign AI becomes the default, not a compromise.The message is clear. Canada does not need to wait for sovereign Al. It is already here, and it is winning.
