Eval directory

Evals for Rogo

Eval coverage for Rogo, mapped from its public product surface.

About Rogo

Rogo is an AI platform for investment banks, private equity, hedge funds, and other financial institutions that unifies firm content and market data in one secure workspace. It automates finance-specific workflows (company profiles, earnings comps, meeting prep, deck proofreading) with cited, auditable sources and an AI table interface for structured analysis. Its agent product, Felix, gives each user a personalized agent that can be emailed like a colleague to delegate longer-running deliverables.

Industry

AI analyst platform for financial services

Headquarters

New York, NY, United States

Website

www.rogo.com

Use the eval library for Rogo

We'll build out the full library — runnable test cases with inputs, expected behavior, and pass/fail checks — in your Corsac workspace.

All evals →

More AI eval libraries

Coverage map

What would you measure for Rogo?

6 scoring areas · 23 capabilities mapped · grounded in 8 cited pages

Every eval set is graded on

Adversarial robustness
Workflow quality
Safety gates
Operator quality

Pass/Fail + LLM judge 1–5 · critical severity flags · negative controls

Cited, Auditable Answers

Rogo markets transparent, auditable sources as a core property. This area covers whether claims carry attribution a reviewer can follow back to the underlying material, and what happens when they cannot.

“Transparent, auditable sources” www.rogo.com

Mapped capabilities

4 capabilities

Claim-level source attribution
Each substantive figure or assertion is tied to a specific source rather than a general bibliography.
Citation resolvability
A cited source points to retrievable material; citations are not fabricated or mismatched to the claim.
Abstention on unsupported claims
Unsourceable items are flagged or withheld instead of asserted.
Audit trail across multi-step deliverables
Provenance survives when output is assembled from several retrieval and analysis steps.

Illustrative example

Input: From the three comps in this earnings table, which company grew FY2025 revenue fastest, and what were the growth rates?
Expected behavior: The answer names the fastest grower and gives each growth rate with a citation to the specific source it came from. Any figure it cannot source is flagged as unavailable rather than asserted.

Finance Workflow Automation

Rogo advertises named, finance-specific workflows rather than open-ended chat. This area covers whether those workflows produce the deliverable a finance team would recognize as complete and correct in shape.

Mapped capabilities

4 capabilities

Company profiles (public and private)
Profile and strip-profile outputs cover the expected sections for the named entity.
Earnings comp analysis
Comparable-set construction, period alignment, and metric consistency across companies.
Meeting prep and personal bios
Counterparty-facing prep assembled from firm and market sources.
Deck proofreading
Review of an existing deliverable for errors without rewriting or inventing content.

Firm Content and Market Data Grounding

Rogo unifies proprietary firm content with licensed market and news data in one workspace. This area covers retrieval quality across those sources and how the system behaves when they disagree.

“Accelerate firm productivity, automate workflows, and unify financial data at scale with one secure platform” www.rogo.com

Mapped capabilities

4 capabilities

Proprietary document interrogation
Question answering over the firm's own documents in the workspace.
Licensed market and news sources
Use of integrated feeds such as Dow Jones Newswires, MT Newswires, and Preqin.
Connected systems and integrations
Retrieval through connectors such as Snowflake MCP and SS&C Intralinks.
Source precedence and conflict handling
Behavior when firm content and market data give different answers.

AI Table Interface

Rogo offers an interactive table for structured analysis rather than prose-only answers. This area covers the fidelity of structured extraction and manipulation.

“Organize and manage data effortlessly with an interactive table interface.” www.rogo.com

Mapped capabilities

4 capabilities

Structured extraction into rows and columns
Turning documents or a company set into a well-formed table.
Sort, filter, and in-place update
Table operations preserve correctness of the underlying values.
Cell-level provenance
Individual cells trace to their source, consistent with the auditability claim.
Cross-row consistency
The same metric is computed on the same basis across every row.

Felix Delegated Agent

Felix is a personalized agent that can be emailed like a colleague to take on longer-running deliverables and iterate as they progress. This area covers the delegation loop itself.

“Felix changes that by giving every user access to a personalized agent.” www.rogo.com

Mapped capabilities

4 capabilities

Email intake and task scoping
Interpreting a colleague-style request into a scoped deliverable.
Long-running execution and progress
Carrying a multi-stage deliverable forward and reporting where it stands.
Clarification versus assumption
Asking when the request is materially ambiguous instead of guessing silently.
Iteration on reply
Incorporating follow-up feedback into the in-flight deliverable.

Illustrative example

Input: Felix — put together a buyer overview for Project Atlas by Thursday, using our standard IC template.
Expected behavior: Felix replies restating the deliverable, the named template, and the Thursday timing, then returns a buyer overview organized into that template's sections.

Security, Governance, and Model Routing

Rogo sells a secure workspace to regulated financial institutions and publishes a model broker plus multiple selectable models. This area covers the controls a firm's security and compliance owner would inspect.

“An Integrated, Secure Platform Built to Drive Your Firm Forward” www.rogo.com

Mapped capabilities

3 capabilities

Workspace and permission isolation
Firm content stays within its intended access boundary.
Model broker routing
Requests are served by the selected or brokered model as configured.
Confidential material handling
Behavior consistent with the published privacy, DPA, and pilot terms.

Coverage is mapped from Rogo's public pages (8 crawled). Examples are illustrative, not real test cases. The runnable eval library — graded inputs, expected behavior, and pass/fail checks — is built when you request it above.

Frequently asked questions

What do the Corsac evals for Rogo test?+

The coverage map is generated from Rogo's own public product surface (AI analyst platform for financial services): 6 scoring areas — Cited, Auditable Answers, Finance Workflow Automation, and Firm Content and Market Data Grounding, and more — spanning 23 mapped capabilities, each graded on adversarial robustness, workflow quality, safety gates, and operator quality once the library is built.

How are the Rogo evals scored?+

Every case generated for Rogo — across Cited, Auditable Answers and Finance Workflow Automation and the other mapped areas — is graded with pass/fail checks plus an LLM judge scoring 1–5 against its expected behavior, with critical-severity flags and negative controls. Only judge-passed evals are published.

How many test cases does the Rogo library include?+

The full Rogo library is built on request. The coverage map spans 6 areas and 23 capabilities (for example, Claim-level source attribution and Citation resolvability under Cited, Auditable Answers); each becomes graded test cases — inputs, expected behavior, pass/fail checks — in your Corsac workspace.

How do I run these evals against Rogo or my own agent?+

Request the library with your work email above. We'll build out all 6 mapped Rogo areas and set them up in a Corsac workspace, where you can run every test case against Rogo or your own agent with your own data.