Unicorn Founder Rob Imbeault Returns to Build Backboard.io; The Memory Layer for AI

Secures Strategic Backing from Mistral Venture Partners, n49p, Garage Capital and DevCap.

TORONTO, ON  — Backboard.io, an AI infrastructure company building a memory-native layer for Large Language Models (LLMs), today announced a substantial pre-seed funding round co-led by Mistral Venture Partners (Cohere, Klipfolio) and N49P (Spellbook, EvenUP). The round includes key participation from Garage Capital (Groq, Substack) and Dev Cap (Unified, Glowtify).

Founded by Rob Imbeault, co-founder of a SaaS unicorn Assent, Backboard.io marks his return to the operator’s seat to address one of the most pressing bottlenecks in generative AI: the lack of stateful, persistent, and portable memory.

Imbeault previously helped scale Assent from a basement startup into a global supply chain compliance platform valued at more than $1.4 billion. He is now applying the same enterprise-grade rigor around scalability, security, and data reliability to the fragmented AI infrastructure stack.

Backing Rob was an easy decision. Backboard isn’t another AI wrapper. It is foundational infrastructure. Rob has the vision and operational discipline to make Backboard the standard for how AI systems remember and retrieve information.Raif Barbaros, Partner at Mistral Venture Partners

The Mission: Memory-Native AI Infrastructure

Despite rapid advances in model performance, today’s AI systems suffer from a fundamental limitation: amnesia. Once a session ends, context is lost, forcing developers to rely on complex and brittle workarounds to preserve state across interactions.

Backboard.io addresses this challenge by providing a unified, portable memory layer for AI, enabling persistent context and stateful orchestration across more than 2,000 models. Backboard’s memory system has already demonstrated industry-leading performance, achieving 90.1% accuracy on the LoCoMo (Long Conversational Memory) benchmark.

Building Assent taught me that unsexy infrastructure is where real enterprise value is created. AI today is powerful, but fundamentally stateless. We’re building permanent memory infrastructure that allows AI systems to learn and evolve with users over time. To do that, we partnered with investors who understand deep technology and long-term bets.Rob Imbeault, Founder and CEO of Backboard.io

About Backboard.io

Backboard.io is a developer-first AI infrastructure platform designed to serve as the persistent state layer for Large Language Models (LLMs). Currently ranked #1 on the LoCoMo benchmark, the platform enables true memory-native AI systems, moving beyond simple conversational history to retain high-fidelity, long-term context required for complex SaaS tools and autonomous agents. Backboard.io supports multi-model orchestration across more than 2,000 LLMs, providing a portable memory layer that works across providers rather than locking developers into a single vendor. By combining persistent AI memory, context management, and stateful orchestration, Backboard allows teams to deploy a secure, production-ready AI stack in minutes. The company is focused on solving the industry’s “stateless” problem: giving AI the ability to remember, retrieve, and build on past interactions to support sophisticated, multi-step workflows.

About Backboard.io

THE WORLD’S SMARTEST AI MEMORY

Backboard.io is a developer-first AI infrastructure platform designed to bring stateful, persistent memory to Large Language Models (LLMs). The platform enables memory-native AI systems, allowing applications to retain long-term context across sessions, users, and models. Backboard.io supports multi-model orchestration across more than 2,000 LLMs, providing a portable memory layer that works across providers rather than locking developers into a single vendor. The platform combines persistent AI memory, context management, and stateful orchestration to help teams build more reliable, scalable, and production-ready AI applications. Built for enterprise and developer use cases, Backboard.io allows teams to deploy a secure AI stack with persistent memory in minutes using a single configuration. The company is focused on solving one of the core challenges in generative AI infrastructure: enabling AI systems to remember, retrieve, and build on past interactions over time.