News

Startups

Artificial Intelligence

Backboard.io Secures Pre-Seed Funding to Solve Artificial Intelligence Memory Gap

Rob Imbeault’s new venture, Backboard.io, has raised pre-seed funding to build a foundational memory layer that eliminates the stateless "amnesia" common in current AI systems.

Rob Imbeault’s new venture, Backboard.io, has raised pre-seed funding to build a foundational memory layer that eliminates the stateless "amnesia" common in current AI systems.

Rob Imbeault’s new venture, Backboard.io, has raised pre-seed funding to build a foundational memory layer that eliminates the stateless "amnesia" common in current AI systems.

NewDecoded

Published Jan 8, 2026

Jan 8, 2026

4 min read

Image by backboard.io


Ottawa-based Backboard.io has closed a selective pre-seed funding round led by Mistral Venture Partners to address the critical amnesia problem in artificial intelligence. The investment round includes participation from N49P, Garage Capital, and Developer Capital, bringing together partners with deep experience in AI infrastructure. Founded by Rob Imbeault, who previously helped scale Assent to a $1.4 billion valuation, the company aims to transform AI from stateless demonstrations into persistent software. Modern large language models are fundamentally stateless, meaning they lose context and forget user interactions once a session ends. This "LLM amnesia" creates a significant hurdle for developers who must currently rely on fragile workarounds to simulate continuity. Backboard.io is building a memory-native infrastructure layer that allows AI to retain state, reason over long timelines, and improve through continued interaction.

The startup’s technology has already demonstrated a significant technical lead, ranking first on the LoCoMo benchmark with a 90.1 percent accuracy score. This performance indicates that AI agents using Backboard can reliably recall specific details and user preferences without the hallucinations typical of current systems. The platform already supports orchestration across more than 2,000 different models, providing a unified memory solution for the entire ecosystem.

Imbeault’s return to the startup world is driven by the belief that real value is created in foundational infrastructure that works quietly and consistently. He draws on his experience building enterprise-grade systems at Assent to focus on reliability and scalability from day one. The goal is to move beyond the AI wrapper trend and create a standard architectural requirement for the professional AI stack.

The funding round was intentionally kept small to focus on partners who understand long-term infrastructure bets. Mistral Venture Partners, which has backed companies like Cohere and Klipfolio, emphasized that Backboard represents a foundational layer rather than a simple application. Other investors like Garage Capital bring experience from backing industry leaders like Groq and Substack.

With the initial funding secured, the Backboard team is now focused on hardening the platform and earning developer trust through execution. The roadmap involves shipping updates and listening to early users to ensure the memory layer becomes an invisible but essential utility. Imbeault notes in the official announcement that if done correctly, Backboard will eventually fade into the background while simply making AI work better.


Decoded Take

Decoded Take

Decoded Take

The Evolution of the AI Infrastructure Stack

The emergence of Backboard.io signals a maturation of the AI industry from the model-centric phase to the infrastructure-centric phase. While the last two years focused on increasing the raw intelligence of models, the bottleneck has shifted to how that intelligence is applied in persistent enterprise workflows. By solving the amnesia problem at the architectural level, Backboard.io is positioning memory as a distinct utility layer, much like the database was for the early web. This shift indicates that the winners of the next AI wave will be those who can provide the glue that makes AI reliable enough for real-world labor.

Share this article

Related Articles

Related Articles

Related Articles