Hebbia: Coverage Update — 21 Oct 2026

GAII Intelligence defines the AI Research & Intelligence market as software that applies generative and agentic AI to the research process of professional investors: locating candidate securities, assembling and interrogating evidence, producing written analysis with traceable sources, and monitoring positions once held. Buyers are hedge funds, long-only managers, multi-manager platforms, family offices and allocators; the economic buyer is typically the chief technology officer, the head of research, or a portfolio manager with a technology budget.

Capabilities we treat as mandatory

• Evidence retrieval at scale: the ability to search and reason across regulatory filings, transcripts and news, and across documents the client supplies.
• Source traceability: every material claim in an output must be reachable back to the passage that supports it.
• Institutional data handling: a stated position on where client data goes, how long it is held, and whether it is used for model training.

Capabilities we treat as differentiating

• Persistent state: the system remembers a thesis, a coverage list or a position between sessions and acts on it without being re-prompted.
• Workflow surfaces: delivery inside the applications analysts already use — the spreadsheet, the research management system, email and messaging.
• Owned or exclusive content: material that a competitor cannot obtain by signing the same third-party contract.
• Published evaluation: a measured, disclosed accuracy methodology rather than an assertion of quality.

Hebbia competes most directly with Rogo across finance and with AlphaSense for licensed-content research. Its legal client base (Seyfarth, Latham & Watkins) is unusual in our coverage. Its API and MCP server let it run inside Claude and ChatGPT, making frontier labs partly partners and partly competitors.

The following are the capabilities and characteristics that, in our assessment, most distinguish Hebbia within the AI Research & Intelligence market. Each is drawn from publicly documented product capability, published interviews with management, or independent hands-on review.

• Retrieval architecture: Iterative Source Decomposition and a multi-agent research system with context distillation trace every claim to a source quote. In our assessment this is the most developed retrieval design disclosed in the cohort.
• Licensed content depth: FactSet, Capital IQ, PitchBook, Preqin, Fitch and CreditSights, Third Bridge, ICE and Daloopa, plus filings across the US, Europe, the UK and Australasia.
• Citation discipline: Sentence- and cell-level citations, Excel cell notes and a tick-and-tie agent suit audit and compliance review.
• Tier-one client base: BlackRock, Carlyle, KKR, Centerview, Morgan Stanley and MetLife are named, and Seyfarth confirms 7m+ pages processed across about 1,000 lawyers.
• Security posture: SOC 2 Type 2, ISO 27001:2022 and ISO/IEC 42001 badges, a no-training commitment covering model providers, US and EU residency, immutable audit logs and SCIM.
• Agent-ready delivery: An API and MCP server let Hebbia run underneath Claude or ChatGPT, alongside Excel and Word add-ins.
• Shipping cadence: Monthly release disclosures since August 2025, with Max, Matrix 2.0 and Headless Hebbia in 2026.

The following are factors that prospective clients should weigh, and areas where our public review did not find sufficient evidence to reach a settled view. Several are common across vendors at this stage of the market and are not specific to Hebbia. We encourage prospective clients to raise each of them directly with the company; we have done the same, and will update this note as answers arrive.

• Capitalisation: No priced round has been disclosed since the $130m Series B in July 2024, while Rogo raised at a reported ~$2bn in April 2026. Prospective clients planning multi-year commitments should discuss runway.
• Product maturity: Max, Matrix 2.0, Headless Hebbia and the Word plugin launched between July and October 2026. Buyers should validate stability before scaling.
• Deployment options: A dedicated tenant is the stated maximum; customer-VPC and on-premises deployment were not located.
• Governance detail: SSO/SAML, zero-data-retention terms with model providers, information-barrier controls and the subprocessor list were not located publicly; the trust centre was not readable at our review.
• Self-reported metrics: The share of large asset managers claimed has moved from 30% to over 40% without a defined denominator, and accuracy figures come from undisclosed benchmarks.
• Pricing: No price list is published; the reported $3,000–10,000 per seat comes from a single interview, and Max pricing is undisclosed.
• Competitive momentum: Press reports place Rogo's ARR ahead of Hebbia's, and OpenAI launched ChatGPT for Financial Services in September 2026 with licensed financial data built in.
• Leadership depth: Few executives beyond the CEO and CTO are named publicly.

Hebbia was founded in 2020 by George Sivulka, a Stanford mathematics and applied physics graduate who had begun a PhD in machine learning, with three other Stanford researchers. Its first product was a Chrome search plug-in. The company is headquartered in New York, opened a London office in autumn 2025 and a San Francisco engineering hub in October 2025.

Hebbia raised a $1.1m pre-seed (Floodgate, 2020), a $30m Series A (Index Ventures, 2022) and a $130m Series B led by Andreessen Horowitz in July 2024 at a reported ~$700m valuation, about $161m in total. It was reported profitable at $13m ARR at the Series B. Aabhas Sharma, formerly CTO of Found, became CTO in October 2025. Hebbia acquired FlashDocs in June 2025.

Company-reported figures are shown as such and are not independently verified. Verification is carried into the vendor questionnaire.

Hebbia's platform centres on Matrix, a grid that runs agents across many documents at once, with a rebuilt chat assistant and developer access. Figure 2 sets out our assessment of how it maps onto the investment workflow.

Matrix and Matrix 2.0

A grid that runs agents across documents and data; Matrix 2.0 carries work through to deliverables with human sign-off checkpoints.

Max and Deeper Research

An AI team member built on rebuilt chat, and a multi-agent deep-research system with sentence-level citations.

Agents and skills

A Global Agent Library, company screening and precedent-transaction skills, a tick-and-tie agent and an email Intern agent.

Office and developer access

Excel and Word add-ins, Excel and PowerPoint generation, and the Headless API and MCP server for Claude and ChatGPT.

Scope of the product today

We did not locate a PowerPoint add-in (as distinct from slide generation), persistent thesis monitoring or broker research. The company dates the Headless API and MCP server to 1 October 2026.

Hebbia's Iterative Source Decomposition breaks a question into sub-tasks and assembles an answer with an audit trail to source quotes, routed across OpenAI, Anthropic and Google models and a proprietary fine-tuned model. A router manages billions of tokens a day, and context distillation cuts context by more than 90%.

Hebbia publishes more evaluation material than most peers, including a Financial AI Benchmark and a paper on evaluator reliability. These are vendor-authored, and no independent test of the product was located.

Hebbia's licensed stack spans FactSet financials and estimates, Capital IQ, Daloopa, ICE equities and fixed income, PitchBook and Preqin for private markets, Fitch, CreditSights, LevFin Insights and Octus for credit, and Third Bridge expert interviews, alongside filings, transcripts and podcasts.

Client content connects through SharePoint, Box, Egnyte, Intralinks, Salesforce, DealCloud, Snowflake and Databricks. Broker research was not located, and buyers should confirm which premium sources require their own entitlements.

Hebbia fits the document-heavy core of buy-side work: diligence, credit-agreement review, screening and research across large document sets, with the asset managers and private-equity firms that make up its client base.

Persistent thesis monitoring is less developed; notifications and memory arrived in October 2026. A PowerPoint add-in was not located, although slides can be generated.

Hebbia shows SOC 2 Type 2, ISO 27001:2022 and ISO/IEC 42001 badges, a no-training commitment that extends to model providers, siloed per-customer storage, US and EU residency, immutable audit logs, just-in-time production access and SCIM. Figure 3 sets out what our public review located.

We would encourage any firm evaluating Hebbia to request the SOC 2 report and ISO certificates, the subprocessor list, zero-data-retention terms with model providers, SSO/SAML availability, information-barrier controls and whether a customer-VPC option exists.

The offsetting factors are substantial. ISO 42001 for AI management is rare in our coverage, the no-training commitment explicitly covers model providers, and a published DPA and EU residency suit European clients.

Hebbia does not publish pricing. An executive interview cited about $3,000–3,500 per seat per year for a lite tier and about $10,000 for the full product, sold as annual per-seat subscriptions. Max pricing is undisclosed.

The value case rests on hours saved in diligence and review — 30–40 hours per banking deal by the company's account. Buyers should confirm content-licence costs alongside seats.

Hebbia sits in the Established Solutions quadrant of the GAII Intelligence Matrix: Investment Utility of 7.63 rests on product, content and document workflow, while Market & Innovation Strength of 7.43 is held just below the boundary by governance evidence and capital. Peer positions shown are as currently published; the remaining tranche-1 companies are being recalibrated to the same standard.

GAII coverage universe, 21 October 2026. Peer scores as currently published on the GAII Intelligence page; the remaining tranche-1 companies are being recalibrated to the full sub-criteria standard. Preliminary and based on public information.

11.1 Where Hebbia is strongest relative to peers

• Against content platforms, on reasoning across the firm's own documents with claim-level citations.
• Against banking-centred peers, on buy-side and legal document workflows and licensed content depth.
• Against general-purpose assistants, on audit trails and finance-specific data — while running inside them through MCP.

11.2 Areas of narrower functional focus

Hebbia has built depth in retrieval and content rather than in persistent monitoring or deployment flexibility.

• Monitoring: persistent thesis monitoring not located.
• Content: broker research not located.
• Governance: information barriers, SSO and customer-VPC options not located.
• Capital: no priced round disclosed since July 2024.

Scores are on a 0–10 scale. Each dimension score is the mean of its sub-criteria, and the GAII Overall Score is the weighted mean of the six dimensions. The detailed sub-criteria scores that build each dimension are set out in Appendix A.

These are the areas where our public review did not find sufficient evidence to form a settled view. We have put them to the company and will update this note when they are answered. Prospective clients may find them a useful agenda for their own evaluation call.

• Capitalisation and runway, and whether capital has been raised since July 2024.
• The SOC 2 Type 2 report period and auditor, and the ISO certificates.
• The subprocessor list and zero-data-retention terms with each model provider.
• SSO/SAML availability.
• Information-barrier and MNPI controls.
• Whether a customer-VPC or on-premises option is available.
• Current ARR and the definition of "largest asset managers".
• Pricing for seats, Max and the API, and standard term.
• General-availability dates for Max, Matrix 2.0 and Headless Hebbia.
• Which content sources require the client's own licence.
• Roadmap for persistent thesis monitoring and broker research.
• Two buy-side references.

This note is an assessment of a vendor against a defined framework, not a recommendation to buy or to decline. The following are the considerations we would put at the centre of an evaluation of Hebbia specifically.

If you are a fundamental equity manager

• Test Matrix on a live diligence set and judge citation quality against your own review.
• Map which content partners your existing licences cover.
• Validate the 2026 previews before building process around them.

If you are responsible for technology or vendor risk

• Open with Figure 3 and request the trust-centre pack.
• Confirm SSO, information barriers and VPC needs early.
• Use the MCP server to evaluate Hebbia inside your existing assistant.

If you are assessing the category rather than the company

• Retrieval architecture is now a differentiator worth asking every vendor to explain.
• Licensed-content partnerships are consolidating around a few AI-native platforms.
• Capital is concentrating; viability belongs in every multi-year evaluation.

Each dimension score in Section 12 is the mean of the sub-criteria below.