Boosted.ai, a specialist in agentic AI for investment management, has announced Alfa Prime, a multi-model AI investment committee that combines large-scale signal monitoring, purpose-built research agents, and structured debate between independent AI models. The system is designed to help investors identify, investigate, and challenge investment ideas within a framework defined by the investment team itself.

"The AI committee debates, challenges, and recommends. People direct and decide."

Drawing on Boosted.ai's more than eight years of quantitative machine learning experience, Alfa Prime can monitor millions of market, fundamental, and research signals, surface the ones that warrant attention, and deploy specialized AI agents to investigate them before a thesis is tested across multiple model families. The Toronto-based company previously built Boosted Insights, a machine learning platform that lets portfolio managers apply their own data and variables to quantitative models, and Alfa Prime extends that institutional lineage into agentic, multi-model research.

Alfa Prime reflects a broader shift taking shape across investment management, as increasingly capable AI models move from answering research questions to participating more directly in the research process itself. That shift is already visible across the industry, with banks, asset managers, and data vendors rolling out generative AI tools for tasks such as summarizing filings, earnings transcripts, and research notes. Boosted.ai believes the shift will change how investment ideas are discovered, tested, challenged, and ultimately evaluated. As an initial application of Alfa Prime, the company is opening the system to a limited number of institutional partners, including funds and asset managers that want the approach configured around their own mandate, criteria, and house view.

"We didn't build Alfa Prime because the market needed another AI tool," said Joshua Pantony, Co-Founder and CEO of Boosted.ai. "The jump in what these models can do over the last year is real. In our own testing, Alfa Prime was able to identify higher-quality investment opportunities at roughly twice the rate of our baseline analyst workflow.¹ But a performance benchmark is not an investment process. What it showed us is that the process itself needs to evolve. One model giving you an answer is not enough. The future is an investor setting the objective, multiple models building the bull and bear cases and challenging each other, and an AI swarm working together to optimize the outcome."

A committee that debates, challenges, and revisits the thesis

Most AI tools are designed to answer a question once. Alfa Prime, by contrast, is built to pressure-test an investment idea from multiple angles.

The system assigns independent AI models to research and argue competing bull, bear, and moderating views, drawing on underlying financial data, filings, transcripts, research, and other relevant evidence. Models challenge competing assumptions, identify gaps in the evidence, and refine their conclusions across multiple rounds. Different model families participate so that the analysis is not dependent on the reasoning patterns or blind spots of any one model — an approach that carries over a practice long standard in quantitative investing, where independent models are ensembled to reduce reliance on any single model's errors.

A moderator then evaluates where the models agree, where they disagree, and why. The resulting analysis is distilled into a citable investment memo that surfaces supporting evidence, unresolved questions, key risks, and the assumptions most likely to change the conclusion. For institutional investors, documentation of this kind carries practical weight: investment decisions and their supporting rationale are subject to internal review and, in many jurisdictions, external regulatory scrutiny.

Alfa Prime produces bull, base, and bear cases along with an indication of conviction based on how the debate evolves. Rapid convergence can signal a more strongly supported thesis, while sustained disagreement can highlight uncertainty or areas requiring deeper investigation.

The investor remains involved at both ends of the process. People define the question and the investment framework at the outset, and they inspect, challenge, and ultimately decide what to do with the analysis. The AI committee debates, challenges, and recommends. People direct and decide.

Opening Alfa Prime to select institutional partners

Boosted.ai will initially work with a small cohort of institutional partners to configure Alfa Prime around each firm's investment methodology. Rather than producing a generic market view, Alfa Prime is shaped around a firm's mandate, research framework, proprietary data, portfolio context, and house view. The launch lands in a field where established data and analytics providers and a wave of startups are all offering AI research tools to institutional investors.

The company is deliberately limiting the number of relationships it pursues, including bespoke implementations and, where appropriate, more exclusive arrangements around specific strategies, markets, or applications.

"Every firm we talk to can see where these models are headed," Pantony added. "We'd rather work deeply with a small number of serious partners than sell a widget to everyone. Investment firms have spent decades competing to put the smartest people in the room. We're entering a world where the smartest thing in the room may not be a person. That changes what an investment edge looks like."

Pantony has published an essay outlining why Boosted.ai believes recent advances in AI require a new approach to fundamental research, and what that shift could mean for investment decision-making.