Rogo

Language model reasoning over more than 50 million financial documents is the product, and the Felix agent carries deal and investment work end to end. Take the models away and the customer is left with data subscriptions it already had.

Outbound outreach to prospective buyers in a live deal is among the agent's described duties, along with personalized teasers and deal status updates committed into the firm's tools. Contacting counterparties carries confidentiality and market conduct consequences, yet no approval gate before outreach or sign off on a generated memo is described.

Auditable Excel models and a publicly described dataset integration approach give a reviewer structure to work with. Evaluation results, benchmark performance and a validation summary are unpublished, as is support for a customer's own review of the system.

Executives at Truist Securities, Nomura and Baird are quoted, and usage runs to more than 150,000 queries a day from over 50,000 bankers and investors. Testimonials and figures never meet: the quotes carry no number, and the ten hours a week saved, from OpenAI's customer story, is an aggregate.

A categorical commitment stands on the security page: private customer data never trains or updates the models. Each customer's data sits in a siloed environment under zero trust access controls with end to end encryption.

Deal content is the sensitive payload: data rooms, cap tables and buyer lists, material non public information whose leak moves markets. The privacy policy effective June 30, 2026 covers the website only and leaves enterprise data to written agreements that are not published, and retention for deal content is unstated.

SOC 2 and ISO 27001 are listed with regular third party audits and vulnerability tests, and a security portal at trust.rogo.ai is linked. Neither the SOC 2 type nor the audit period is on the page, and whether the portal releases reports without an agreement is not shown.

Users are broker dealers and advisers bound by recordkeeping, supervision and information barrier rules, and nothing published shows how agent outreach and generated materials enter supervised records. Compliance standards list the EU AI Act without a risk class or role, and separation between institutions on opposite sides of one transaction is not evidenced.

Companies and transactions are the subjects, so accuracy is the governance question here. No error rate for extraction or analysis is published, nor any process for catching a wrong memo once it sits in a live deal file, and the security advisory board covers security rather than models.

Numbers come from retrieved data rather than model arithmetic, and every claim cites its source, so a banker can check output before a client sees it. Rogo publishes no guarantee, error rate or remedy for a memo an agent filed into a live deal that proves wrong.

OpenAI's customer story assigns each model a job: GPT 4o for conversation and analysis, o1 mini for structuring data and o1 for evaluation and synthetic data, with GPT 5 an announced extension. Cloud host, subprocessors and the retention terms on which deal content reaches the model provider are undisclosed.

Integration runs both ways: S&P Global, Crunchbase and FactSet feed in, with SharePoint and customer relationship systems named on the homepage, and models, memos and decks are written back into the tools where deal teams work. A partner's own page confirms the data integrations.

Cloud delivery comes with no named host, region, residency option or transfer mechanism, and model calls sit in the processing path. Confidential transaction material makes that gap an early review question.

Every route ends at a demo request, with seat cost, tiers and minimums unpublished. Finance professionals placed with customer teams during onboarding add a second cost dimension, bundled or not, that nothing published explains.

Investment banks, private equity firms and investors at more than 350 institutions make up a deep vertical, with expansion announced in Europe and Asia Pacific. Nothing is published for commercial banks, insurers or lenders.