Filot on the AI Panel at Unbundled & Uncovered New York

Filot Co-Founder and CEO Bjorn Birkelund took the stage at Unbundled & Uncovered New York 2026, hosted by Substantive Research at the Metropolitan Club. The conference is the industry's leading forum on the transformation of investment research, and the panel brought together founders and executives from across the research technology landscape to confront a pointed question: why is so much AI investment in research failing to pay off?

The Panel

Bjorn shared the stage with David Nable of BlueMatrix, Thomas Li of Daloopa, Rahul Ravikrishnan of Eidosmedia, Rahul Rekhi of Rogo, and Gavin Skinner of Aiera, in a discussion moderated by Brijesh Malkan of Singletrack.

The conversation opened with a hard truth: most AI tools are not actually delivering ROI in investment research. Generic AI offers plenty of customization on the surface, but matching the nuanced sophistication of how a specific analyst actually works, and doing it efficiently, takes a far more tailored system. In this world, "nearly right" isn't useful.

The real unlock comes when AI molds to each analyst's unique way of working rather than forcing the analyst to adapt to the tool, and does it hands-off enough to scale across an entire firm. In modeling, that means an AI that can be tuned in minutes to follow each analyst's precise methodology, accelerating model builds and updates across the entire coverage universe, down to formula linking, table structure, sheet layout, and projection approach.

Determinism in a Non-Deterministic World

The most important thread of the discussion centered on how AI can be used deterministically in highly regulated, high-stakes domains like investment research, when LLMs are non-deterministic by nature.

Filot's answer is a multi-layer auditing approach, something most AI products leave as an afterthought: deterministic validation tests combined with granular auditability down to each individual data point. Trusting the output of an AI matters, but being able to efficiently audit the results matters just as much, because that is what allows firms to trust outputs at scale.

"Nearly right isn't useful in this industry. The firms adopting AI successfully are the ones that can verify every number, not just admire the output. That is why we built auditing into Filot as a first-class layer rather than an afterthought, and it is a big part of why highly regulated enterprises across the buy and sell side have adopted the platform." — Bjorn Birkelund, Co-Founder and CEO of Filot

Why This Conference Matters

Unbundled & Uncovered sits at the intersection of research economics and research technology. The audience is not evaluating AI as a concept; directors of research and heads of desk are deciding which platforms clear the bar for institutional deployment. Thanks to Mike Carrodus and the Substantive Research team for another stellar conference. We look forward to the next one in London.