Filot and the Rise of Vertical AI: Why the Future of Investment Research Is Methodology-Aware
Trading & Investing News / Ai News
For much of the past three years, artificial intelligence in finance has centered on general-purpose assistants capable of answering questions, summarizing documents, or generating text. While these tools have demonstrated impressive reasoning abilities, they have struggled with one of the most important realities of institutional investing: every investment firm has its own proprietary methodology.
The spreadsheet architecture, modeling conventions, valuation framework, source preferences, formatting standards, and analytical workflows developed over years—or decades—often represent a firm’s competitive advantage. General AI systems can accelerate individual tasks, but they typically force analysts into generic workflows that fail to preserve this institutional knowledge.
According to Filot Co-Founder and CEO Bjorn Birkelund, solving this challenge required a fundamentally different approach. Rather than replacing an investor’s process, Filot was designed to learn and replicate it.
“The methodology each investor has refined over their career—the firm’s house style—that’s where the edge comes from,” Birkelund said. “What we’ve cracked at Filot is how to take the mechanical layer—formulas, structure, tables, linking, model construction, and updates—and execute it in each investor’s exact methodology. Filot cuts modeling time by 91% on average. This has enabled analysts to either produce 10x more models in their own house methodology, or hold model count flat and reinvest the time into deeper diligence on existing positions and sharper decisions.”
That distinction represents a broader shift taking place across enterprise AI.
Shuo Chen, entrepreneur, investor, and entrepreneurship lecturer at Stanford and UC Berkeley, believes the industry is moving decisively toward specialized AI systems rather than generic assistants.
“The current shift in AI is moving away from generic assistants and instead toward vertical agents—tools designed to scale a team’s specific methodology,” Chen said. “In the investing world, a firm’s analytical process is its core intellectual property. The most successful systems won’t force a generic template on users; instead, they will replicate the unique way specific analysts and teams think and work. Filot is the leading example of this specialized approach.”
Rather than functioning as another chatbot layered on top of Excel, Filot positions itself as a financial modeling engine that operates inside institutional workflows. The platform is designed to preserve an investment team’s formulas, formatting conventions, assumptions, linking logic, validation procedures, and source preferences, allowing every model to remain consistent with the firm’s existing house style instead of introducing standardized templates.
This philosophy reflects one of the most significant trends emerging in enterprise AI. Across industries, organizations are discovering that their competitive advantage rarely comes from access to the same large language models available to everyone else. Instead, differentiation increasingly comes from encoding proprietary workflows and institutional expertise into AI systems that faithfully execute an organization’s unique processes.
For investment firms, that opportunity is particularly compelling.
Financial modeling remains one of the most labor-intensive activities in fundamental research. Analysts routinely spend hours—or days—constructing three-statement models, updating earnings, repairing broken formulas, building comparable-company analyses, and maintaining valuation frameworks. Much of this work is repetitive, highly structured, and essential, yet it contributes little directly to generating differentiated investment insights.
Filot’s objective is to automate this mechanical layer while leaving the investment judgment entirely in the hands of the analyst.
According to Birkelund, the firm’s technology allows analysts to redirect their effort away from spreadsheet mechanics and toward activities that actually generate alpha, including developing differentiated investment theses, conducting channel checks, evaluating management teams, and identifying variant perceptions before the broader market.
The platform supports a broad range of institutional workflows, including initiating coverage on new companies, updating models immediately after earnings releases, generating bespoke comparable-company analyses, repairing broken financial models, constructing private-market and credit models, and handling ad hoc modeling requests through natural language while preserving each firm’s preferred conventions. According to the company, every output is accompanied by source traceability, reconciliation checks, and validation mechanisms designed to satisfy investment committees, risk teams, and audit requirements.
Enterprise adoption appears to be accelerating. Filot reports that its technology is trusted by several Tier 1 multi-manager and multi-strategy investment firms and has expanded partnerships across institutional financial services, including a strategic relationship with TresVista. The company also states that its platform supports investment organizations collectively responsible for approximately $1 trillion in buy-side assets under management.
Perhaps the most important implication extends beyond financial modeling itself.
The first generation of enterprise AI focused primarily on increasing individual productivity. The next generation appears increasingly focused on capturing institutional knowledge—transforming years of accumulated organizational expertise into repeatable, scalable software.
For hedge funds, asset managers, and investment banks, this represents a profound opportunity. A firm’s proprietary modeling methodology, once embedded only in senior analysts and spreadsheets, can become an operational asset that scales consistently across teams without sacrificing the firm’s investment philosophy.
In that sense, Filot is representative of a broader evolution occurring throughout artificial intelligence. Rather than asking humans to adapt to AI, the most valuable enterprise systems are increasingly learning how to adapt to humans.
If that trend continues, the future of AI in financial services may not belong to the most general assistant. It may belong to the systems that most faithfully preserve—and amplify—the unique intellectual processes that investment firms have spent decades refining.
Model at the Speed of Markets – Filot
