Rogo AI Review 2026: Felix IB Research Copilot

# Rogo AI Review (2026): Felix, Pitchbooks, Models - What’s Real vs Demo Theater

You have seen the demo. A banker emails an agent; a branded deck comes back. The honest answer, as of September 2026, is that Rogo is a real product with real institutional deployments - and also a sales motion that rewards polished theater. This review separates what public primary sources support, what is vendor narrative, and what you cannot know without a paid seat and your own data. Concrete numbers get quoted; adjectives do not.

Bias note up front: Vantaige maintains a directory page at /ai-tool/rogo-ai. Treat any ranking language there as directory opinion with commercial incentive risk. This article prefers primary filings of fact (Rogo blog, Kleiner Perkins) over our own card copy. I have not sat a multi-month production seat on your firm’s MNPI stack for this piece - label every capability claim accordingly.

1. TL;DR
2. What Rogo is (and is not)
3. Felix: harness, not “a model”
4. IB workflows that are actually claimed
5. Series D and scale claims (with sources)
6. Named clients - public sources only
7. Pricing opacity
8. vs Hebbia and AlphaSense (brief)
9. Evaluation checklist
10. Demo theater vs hands-on limits
11. Watch out / when NOT
12. FAQ
13. References
14. Related from Vantaige

TL;DR {#tldr}

Rogo = vertical AI platform for high finance (IB, PE, public/private markets), outputting Excel / PowerPoint / Word in firm templates with citation/auditability claims.
Felix = Rogo’s autonomous agent layer: email-style delegation, multi-step workflows, slide annotation loops, model routing across frontier labs (Rogo’s “Model Broker” narrative).
Capital (primary): $160M Series D announced 29 Apr 2026, led by Kleiner Perkins, with Sequoia, Thrive, Khosla, J.P. Morgan Growth Equity Partners, and others named on Rogo’s post.
Named clients (public): Kleiner names Rothschild & Co, Jefferies, Lazard, Moelis, Nomura. Rogo’s own IB post names Baird, Jefferies, Lazard and cites a Baird engagement claim (95% engagement - vendor claim). Usage figures moved from Kleiner’s 35,000+ daily users to Rogo’s later 40,000+ professionals / 300+ institutions (July 2026) - treat as vendor-reported, not audited.
Pricing: not published. Any per-seat dollar figure you see online is third-party estimate or rumor until your quote lands.
When NOT: small boutiques without security/procurement capacity; teams that only need quote-level research Q&A; buyers who confuse a scripted pitchbook demo with production MNPI isolation.

What Rogo is (and is not) {#what}

Is: A finance-specific agent platform aimed at deal work - comps, models, memos, pitchbooks, screenings, diligence assist - with forward-deployed staff encoding firm workflows. Kleiner Perkins’ April 2026 note frames it as agents that “execute full workflows” and deliver in Excel, PowerPoint, and Word in the firm’s house style.

Is not:

1. A Bloomberg/FactSet replacement. Rogo’s own materials lean on connected third-party data (PitchBook, FactSet, Refinitiv and peers appear in third-party comparisons and Rogo product copy). You still buy the terminal/content licenses.
2. A self-serve $20/month sidebar. There is no public self-serve rate card.
3. Proof that junior banker headcount goes to zero. Vendor ROI stories exist; independent, standardized productivity RCTs do not (and we will not invent them).

This is a method for evaluating a vertical agent vendor under banking constraints - not a buy rating.

Felix: harness, not “a model” {#felix}

Rogo product pages and technical commentary converge on a useful framing: Felix is an agent harness (orchestration, tools, templates, evals, deployment) sitting on top of frontier models, not a single proprietary LLM you “switch to instead of GPT/Claude.”

What Rogo publicly claims Felix does:

Accept work the way a junior would - including email delegation.
Produce decks, models, memos, dashboards, sourced research from multi-step runs.
Iterate via a slides annotator: comments map to slides; edits apply in place rather than regenerating the whole deck (Rogo IB use-case post).
Attach deep-linked citations so reviewers can chase a figure to a filing line / cell / transcript quote (vendor claim - verify on your documents in pilot).
Schedule monitoring tasks by email (May 2026 product update).
Route across OpenAI / Anthropic / Google via Model Broker (Rogo positioning); Claude Opus 5 was announced as selectable inside Rogo on 24 Jul 2026.

Watch out: “model-agnostic harness” is an architecture story. It does not automatically mean your outputs are better than Claude-in-Excel or ChatGPT-for-PowerPoint on a single slide. The bet is workflow + firm memory + citations + deployment, not raw next-token quality on a homework prompt.

IB workflows that are actually claimed {#workflows}

Rogo’s July 2026 IB article is the cleanest public map. Mapped to jobs bankers recognize:

Job	What Rogo claims	What you must still own
Pitchbooks / BD decks / CIMs / valuation packs	First-draft pages in firm format; annotation-based revision; citations on figures	Narrative judgment, client politics, final QA
Models (DCF, LBO, merger, comps, operating)	Build with inputs from filings, presentations, consensus, connected feeds; cell-level source links	Assumption choice, sensitivity design, sign-off
Research / screenings	Transcript takeaways; deterministic screenings across structured sources (example filters published by Rogo)	Universe definition, edge cases, private-company gaps
Process / diligence assist	VDR sync (SS&C Intralinks called out); scheduled watches; briefing packs	Legal hold, MNPI walls, human diligence calls

Agent Library (Rogo news): shared library of purpose-built agents across IB/PE/credit/etc., plus forward-deployed custom agents. Rogo publishes large execution counts for shared and custom agents; treat volume metrics as vendor telemetry, useful as adoption signal, not as quality proof.

Series D and scale claims (with sources) {#funding}

Primary - Rogo, 29 Apr 2026: “Today we’re announcing $160M in new financing, led by Kleiner Perkins, with participation from Sequoia, Thrive Capital, Khosla Ventures, J.P. Morgan Growth Equity Partners, BoxGroup, Mantis VC, Jack Altman, Evantic and Positive Sum.” Capital use called out: deeper integrations, more forward-deployed bankers/engineers, EMEA/Asia expansion.

Primary - Kleiner Perkins, 29 Apr 2026: Confirms leading the Series D; states more than 35,000 bankers and investors use Rogo daily; names Rothschild & Co, Jefferies, Lazard, Moelis, Nomura.

Secondary (label as such): Press and analyst write-ups around the round cite ~$2B valuation and total funding above $300M, and expand client lists. If a number does not appear on rogo.com or kleinerperkins.com, do not paste it into an IC memo as fact.

Later vendor scale (Rogo, 20 Jul 2026): “More than 40,000 financial professionals at over 300 institutions.” That is a +5k users / institution count update relative to Kleiner’s April note - consistent with growth narrative, still unaudited.

Named clients - public sources only {#clients}

Allowed names for this article (appeared in primary or vendor pages we fetched):

Name	Where it appears
Rothschild & Co	Kleiner Perkins Series D note
Jefferies	Kleiner; also Rogo IB post
Lazard	Kleiner; also Rogo IB post
Moelis	Kleiner
Nomura	Kleiner
Baird	Rogo IB post (incl. “95% engagement across global IB within a few months” - vendor claim)

We are not inventing Reddit quotes, Slack leaks, or “a VP at X said.” If a bank is not on a public page, it stays off this list.

Truist Securities and other names appear in secondary roundups; verify against a primary page before you put them in a procurement brief. When in doubt, ask Rogo for a referenceable customer list under NDA.

Pricing opacity {#pricing}

Rogo does not publish a public price list. AlphaSense, Hebbia, and Rogo share this sales-led pattern in 2026 research-tool markets.

What that means operationally:

1. Budget from a scoped pilot (seats × months × integration hours × forward-deployed time), not from a blog’s guessed ARR.
2. Third-party seat estimates circulate (e.g., market-intel notes sometimes cite ~$3,300 per seat per year). Label: unverified third-party estimate, not a Rogo rate card. Real contracts swing with modules, data connectors, FDE intensity, and term.
3. Some competitor roundups allege seven-figure floors for large institutional deployments. Again: unverified unless your quote says so.
4. Rogo markets a credit center for governing spend across teams/deals/geographies (vendor feature claim). Ask to see it in the pilot with your admin role - demos often skip governance UI.

Affiliate disclosure: none. No sponsored placement. Directory page bias disclosed above.

vs Hebbia and AlphaSense (brief) {#compare}

Hold the three apart by the question each answers best:

Tool	Best question	Weak question
Rogo / Felix	“Run this IB workflow and give me a firm-templated deliverable.”	“Search 500M premium docs I don’t already license.”
Hebbia (Matrix)	“Grid-reason across my uploaded VDR / document set with cell-level citations.”	“Auto-produce a bulge-bracket pitchbook house style.”
AlphaSense	“What does the market content library say (broker, experts, filings, news)?”	“Replace my associate’s comps book build overnight.”

Many institutions will buy two of three. Treating them as mutually exclusive is how RFPs get weird. For directory cards see Vantaige’s Hebbia, AlphaSense, and Rogo pages - weigh that bias.

Horizontal Office add-ins (ChatGPT for PowerPoint/Excel, Claude for Office) compete on individual productivity at much lower friction. They do not claim Rogo’s firm-agent library + FDE packaging. Different layer of the stack.

Evaluation checklist you can paste into the Rogo (or SE) call {#checklist}

This is a method for turning a flashy deck into a measurable pilot - not a guarantee Rogo fails or passes.

Gate	Pass condition	Fail condition
Template fidelity	Your cover, fonts, footer, and exhibit styles survive without manual rebuild	Demo skin only; “we’ll customize later”
Citation chase	Random figure → source in <2 minutes for a reviewer who was not in the demo	Citations point to wrong tab / generic URL / missing
MNPI isolation	Two mock deal rooms cannot see each other’s memory or files	Shared workspace “for the pilot” with no wall
Data wiring	At least one live licensed feed (FactSet/PitchBook/etc.) returns numbers that match the terminal	Filings-only sandbox sold as “full stack”
Edit loop	Annotator fixes three conflicting comments without regenerating the whole book	Full regen wipes prior accepted slides
Admin path	SSO + SCIM + audit export demonstrated on your IdP	PDF of certificates, no live admin console
Cost governance	Credit/usage view by team or deal code	“We’ll discuss pricing after you love it” only
Time-to-value	One custom agent for a real recurring workflow in the pilot window	Library tour with zero firm-specific agent

Score each gate 0/1. A product that wins the demo but scores <5/8 is still theater for your risk committee.

Security claims - what to verify, not what to trust on a slide

Rogo’s IB post asserts SOC 2, ISO 27001, GDPR, BYOK, SCIM, residency controls, EU AI Act-related work, and an autonomous security agent (“Sisyphus”) that runs frequent offensive tests. Those are serious claims. Your job is to:

1. Request current attestation letters and scope (which systems, which period).
2. Confirm residency options match your booking entity (US / EU / other).
3. Ask whether pen-test automation runs in your tenant or only in Rogo’s shared product infra.
4. Map MNPI memory tiers to your wall-crossing policy with Legal - vendor diagrams are not policies.

Watch out: “completed external-auditor-validated compliance with the EU AI Act” language (vendor) is not a substitute for your counsel’s reading of obligations that apply to *your* use case.

Demo theater vs hands-on limits {#demo}

What demos optimize for

Happy-path pitchbook from a clean prompt and a prepared data pack
Pretty citation clicks on pre-indexed docs
House-style templates the SE already loaded
Model Broker routing on a non-MNPI sandbox

What your pilot must force

1. Your templates, not the demo skin.
2. A messy live workbook with broken links and local overrides.
3. A VDR slice with conflicting exhibits.
4. An MNPI wall test between two mock deal teams.
5. Citation chase on a number the SE did not rehearse.
6. Time-to-first-useful-agent with your FDE calendar, not “we once did Baird.”
7. Admin: SSO, SCIM, residency, key management, audit export - ask for the compliance packet (Rogo cites SOC 2, ISO 27001, GDPR, EU AI Act work, BYOK, etc. on its IB post; verify certificates and scope dates yourself).

Hands-on limit of this review: public pages + product posts + investor note. That is enough to shortlist. It is not enough to declare win rates on your coverage group.

Watch out / when NOT {#when-not}

Watch out

Usage metrics ≠ quality. 40k users can still ship wrong EV / incorrect precedent sets. Reviewer time is the real unit of cost.
Content dependency. Without your FactSet/PitchBook/etc. entitlements wired, “rich financial data” collapses to filings + whatever you upload.
Lock-in via memory. Rogo markets firm-owned structured memory as the antidote to chat-vendor lock-in. Validate export/portability in writing.
FDE as product. Forward-deployed bankers are a feature and a cost center. Ask who pays for them after year one.
Security theater. Daily autonomous pen-test stories are interesting; your CISO still needs evidence scoped to your tenant.

When NOT to buy (or even pilot hard)

1. Boutique / lean advisory with <~15 bankers, no security reviewer, and no appetite for a sales-led enterprise cycle - start with Claude/ChatGPT Office + disciplined templates.
2. Quote-only research needs (earnings Q&A, theme search) - AlphaSense or a terminal AI feature may clear the bar cheaper.
3. Document-dump diligence as the only pain - Hebbia-class grid tools may fit tighter.
4. Procurement theater where the MD wants a logo slide for a board offsite but will not fund citation QA.
5. You need a published price today to pass finance - Rogo will not give you a SaaS landing-page number; that alone can be a dealbreaker for some mid-market firms.

FAQ {#faq}

Is Felix a proprietary foundation model?

No - on Rogo’s own framing and technical write-ups, Felix is an agent harness that orchestrates tools, templates, and frontier models (with a broker across labs). Model choice shows up in product updates (e.g., Claude Opus 5 in the selector, Jul 2026).

How much does Rogo cost per seat?

Undisclosed publicly. Ignore unaudited blog figures for budgeting. Run a pilot SOW with seat count, modules, data connectors, and FDE hours itemized.

Which banks actually use Rogo?

Public primary mentions include Rothschild & Co, Jefferies, Lazard, Moelis, Nomura (Kleiner) and Baird/Jefferies/Lazard (Rogo). That is not an exhaustive customer list.

Can Rogo replace associates?

The honest answer: it can compress drafting and packing cycles if citations and templates hold up. It does not replace client judgment, staffing for live negotiations, or accountability when a number is wrong. Any “3× output” claim you hear in a sales call is anecdote until you measure hours on your mandates.

Rogo vs ChatGPT for PowerPoint - which for decks?

ChatGPT for PowerPoint is a sidebar in your existing PPT with OpenAI credit metering - fast to install, weak on firm-wide agent libraries. Rogo is a platform + FDE bet for institutional workflows. Many banks will evaluate both layers; they are not 1:1 substitutes.

We recommend a pilot with hostile test cases, not a brand. Directory pages can rank tools; this review’s job is to keep you from confusing demo theater with production readiness. Weigh directory bias.

References {#references}

All accessed September 2026:

1. Rogo - Our $160M Series D and the Road Ahead (29 Apr 2026)
2. Kleiner Perkins - Rogo: The AI Platform for Global Finance (29 Apr 2026)
3. Rogo - AI for Investment Banking: Key Applications and Use Cases (20 Jul 2026)
4. Rogo - Meet Felix
5. Rogo - What’s New: May 2026
6. Rogo - Claude Opus 5 Now Available in Rogo (24 Jul 2026)
7. Rogo - Rogo’s Agent Library
8. Rogo - The Next Phase of AI Transformation in Finance (Felix live positioning)
9. Secondary architecture commentary (non-primary for customer claims): Felix Is a Harness, Not a Model
10. Pricing opacity context (no Rogo dollars claimed as fact): AllMind - Investment Research Software Costs: A 2026 Budget Framework

Rogo AI directory card - weigh ranking bias
Hebbia Matrix directory card
AlphaSense directory card
Install ChatGPT for Excel, Sheets and PowerPoint (2026) - horizontal Office layer
Claude for Excel / Word / PowerPoint / Outlook setup (2026)

---

One short email a week: which finance AI demos survive a hostile pilot, which credit cliffs hit Office add-ins, and which “named logo” claims actually have a primary source. Subscribe at vantaige.io.

Rogo AI Review (2026): Felix, Pitchbooks, Models - What’s Real vs Demo Theater

Table of contents