Diligence Dossier
An AI equity research analyst that ingests 10-Ks, earnings transcripts, news, and sell-side notes to maintain a living bull/bear thesis on a ticker over weeks, surfacing what changed and what it contradicts.
buy-side equity analysts running multi-week coverage on a stock
- Live thesis log that updates each morning with what evidence shifted overnight
- Contradiction radar that flags when a new datapoint breaks a prior assumption
- Source ledger linking every claim to the filing, call, or article it came from
- Side-by-side bull vs bear briefs generated before every earnings call
With long-context research agents now reliable enough to maintain memory across sessions, analysts who used to spend Fridays re-reading PDFs can finally outsource the synthesis to a kernel that remembers the prior week's stance.
Multiple industry reports show real buy-side pull: SimCorp's 2025 InvestOps Report finds 75% of buy-side leaders recognize AI benefits for investment analysis, and a Crisil Coalition Greenwich study shows 25% of buy-side equity traders plan to integrate AI in the next 12 months; vendors like Marvin Labs, Fiscal.ai, Quartr, and Fintool all already sell into this exact workflow.Buy Side Firms Gear Up to Integrate AI - Traders Magazine ↗Artificial Intelligence: A Growing Presence on Buy-Side Equity Trading Desks ↗
The exact 'ongoing coverage of public companies with multi-week memory' niche is crowded: Marvin Labs ($89/mo, source-linked filings + earnings calls), AlphaSense (added custom scheduled AI agents in March 2026), Quartr (live calls/transcripts), Fintool, Fiscal.ai, Stocklyst, TENK, ReadA10K, Compound, OpenHelm, aktai — AlphaSense's Tegus acquisition alone covers 250K+ expert interviews incumbents can ground on.AI Tools for Equity Research: 2026 Comparison | Marvin Labs ↗Best AI Tools for Equity Research in 2026 | Atlantis Blog ↗
Willingness to pay is real — AlphaSense is enterprise custom, Bloomberg/FactSet run $12K-30K/seat/year, Marvin Labs charges $89/mo, Quartr is mid-three-figures/seat — but pricing power for a new entrant is constrained because Fiscal.ai and Daloopa offer free tiers and most buyers already bundle AI into incumbent terminal contracts.Pricing | AlphaSense ↗AI Tools for Equity Research: 2026 Comparison | Marvin Labs ↗
Secular tailwind from agentic AI in finance is strong (CFA Institute/RPC, Greenwich, SimCorp all confirm direction), but the 'living thesis that contradicts the prior stance' feature is exactly what AlphaSense's new scheduled agents and Marvin Labs' coverage workflow target — incumbents are already shipping overlapping capability, compressing defensibility for a standalone kernel.Agentic AI for Finance: Workflows & Case Studies | RPC ↗AI Tools for Equity Research: 2026 Comparison | Marvin Labs ↗
Technically buildable today with long-context LLMs, public filings, and transcript APIs — Poirot-style open kernels and Daloopa's MCP connectors show the stack is mature — but 'living contradiction detection across weeks' requires careful memory/eval engineering and competing against vendors that already layer proprietary data (Tegus interviews, FactSet feeds) on top.AI Agents in Finance and Fintech: A Scientific Review of Agent-Based Models ↗AI Tools for Equity Research: 2026 Comparison | Marvin Labs ↗