Last updated: 2026-05-21
Cohesion is a YC Spring 2026 AI agent system for hedge fund analysts. It tracks earnings, podcasts, and X posts autonomously. Live with $10B+ AUM. Free trial.
Cohesion is a YC Spring 2026 AI agent platform for public equity investors, built by ex-hedge fund analysts. It autonomously tracks earnings reports, company podcasts, and X posts to surface investment signals for 10+ hedge funds with $10B+ AUM. Pricing is not public; access starts with a free 20-ticker trial at cohesionplatform.com.
Cohesion is an AI agent platform built for public equity investors and hedge fund analysts, founded in San Francisco as part of Y Combinator Spring 2026. The company was started by Devon Krapcho, a former hedge fund analyst at Long Path Partners with five-plus years covering software companies, and Matthew McBrien, who previously built AI tools for investors at T. Rowe Price and worked in security at AWS. The platform is designed to replace the manual, time-consuming process of monitoring dozens of companies across fragmented data sources. The core product is a set of AI agents that continuously track public companies by monitoring earnings reports, news, podcasts, and social media posts on X (formerly Twitter). Analysts historically spent hours each week scanning these sources by hand and still missed material signals. Cohesion agents run this process autonomously and surface differentiated insights directly to the analyst, cutting time-to-signal from hours to minutes. Cohesion is purpose-built for public equity investing workflows at long/short and long-only fundamental equity funds. The platform is already live with more than 10 hedge funds collectively managing over $10 billion in assets under management, making it a credible choice for small and mid-sized funds that want analyst-grade coverage without scaling headcount. It is not designed for retail investors or quant funds. Pricing is not publicly listed. Access begins with a free trial: submit a list of 20 tickers to get started at cohesionplatform.com. The platform runs entirely on web with no desktop or mobile apps. Given the early-stage nature of the company and its YC backing, pricing is negotiated directly with the team.
No public pricing. Free trial: submit 20 tickers to start at cohesionplatform.com. Custom pricing negotiated directly with the team.
Cohesion is a YC Spring 2026 AI agent platform built for public equity investors at hedge funds, available at cohesionplatform.com. Founded by Devon Krapcho (ex-Long Path Partners) and Matthew McBrien (ex-T. Rowe Price), it deploys AI agents that autonomously monitor companies across earnings reports, podcasts, and X posts. The platform is live with more than 10 hedge funds managing over $10 billion in combined AUM. It replaces the hours analysts spend manually tracking coverage lists across fragmented, decade-old platforms.
Cohesion does not publish public pricing as of May 2026. Access starts with a free trial where you submit up to 20 stock tickers to get started at cohesionplatform.com. Paid plans are custom-priced and require direct contact with the founding team. No standard monthly or annual pricing tiers have been disclosed, which is typical for early-stage YC companies selling to institutional fund clients who expect negotiated terms.
Cohesion core features include autonomous tracking of earnings reports and SEC filings, monitoring of company mentions in podcasts, and real-time tracking of X (Twitter) posts mentioning covered companies. The platform generates differentiated investment insights without requiring manual queries or setup from analysts. Users submit a ticker list and the AI agents begin monitoring immediately, surfacing signals that would otherwise require hours of manual research each week.
Cohesion is not free as a permanent plan. It offers a free trial where users submit up to 20 stock tickers to experience the platform before committing to a paid arrangement. After the trial, pricing is negotiated directly with the Cohesion team. There is no self-serve free plan with ongoing access, and the platform is positioned as an institutional research tool rather than a consumer or retail investing product.
The closest alternatives in 2026 are AlphaSense, which offers a broader market intelligence library including broker research, expert call transcripts, and SEC filings with 15 years of institutional history; Hebbia, which focuses on document analysis and end-to-end workflow automation for financial research; and Seeking Alpha, which provides crowdsourced analyst research and earnings event coverage. ChatGPT handles ad-hoc queries but does not provide autonomous always-on company monitoring.
Cohesion is best for fundamental equity analysts and portfolio managers at long/short and long-only hedge funds managing between $100 million and $5 billion in AUM. It is especially valuable for funds with analysts covering 20 or more companies simultaneously, where manual monitoring creates a significant weekly time burden. It is not a good fit for quantitative funds that need structured data feeds or APIs, or for retail investors without institutionalized coverage processes.
No public API documentation has been published by Cohesion as of May 2026. The platform is web-only with no publicly documented programmatic access or data export endpoints. No MCP (Model Context Protocol) support has been announced. Given the company early stage (team of 3, founded 2026), API and integration capabilities are likely limited; contact the Cohesion team directly for enterprise integration inquiries.
Cohesion and AlphaSense serve different parts of the equity research workflow. AlphaSense has 15 years of history with a large library of broker research, expert call transcripts, and SEC filings, making it the established choice for comprehensive research coverage at large institutions. Cohesion (YC 2026) focuses on autonomous monitoring of non-traditional datasets like podcasts and X posts, which AlphaSense does not index natively. Choose AlphaSense for breadth and enterprise compliance; choose Cohesion for always-on monitoring of alternative signals without manual querying.