Our review
Hebbia deserves a specialist look when financial or deal research needs to be checked against its source documents. Evaluate it on the research your team actually produces, including the conclusions a reviewer may challenge.7,2This is our reading of the cited material for a buyer with this need. We have not measured the product’s performance.
Hebbia introduces itself with a single line: "AI built for the rigor of finance"8. Its home page describes the product as purpose built and trusted by leading investors, bankers, lawyers, consultants and Fortune 500 companies9,10, and its about page carries the headline "Transforming the way finance works"11. Those three statements set the frame for this profile. Hebbia is assessed here as an AI platform aimed at financial and advisory work, and each section below separates what the company publishes from what a buyer would have to obtain in a sales conversation.
The company says it has spent more than six years building for that audience12, and it sums up the outcome as an intelligence platform built for how finance operates13. Its about page also records that Hebbia was honoured on the IA 40 list in 202314. The same page shows usage figures with labels such as tokens processed by investors, but in the copy we saved those values do not clearly pair with their labels, so this profile does not repeat them15. A buyer who wants those numbers should ask Hebbia for them directly, together with the period they cover.
The company's public material is concentrated on two named products, Max and Matrix, and a blog post records that a single month brought two major launches, Max and Matrix 2.016. That post is undated in the copy we saved, so the launch month cannot be given here16. The remainder of this assessment follows the same order used for every platform on this site: the product, implementation, integrations, security, governance, pricing, typical buyers and the points the public pages leave open.
Max is described on the product page as working the way a firm works, researching and analysing any deal to produce client ready spreadsheets, slides and reports17. The output formats matter to the intended reader, because the deliverables named are the ones a deal team hands to a client or an investment committee. The same page says Max can reason over any volume of documents or companies with precise analysis and complete traceability back to every finding2. Traceability is the claim a reviewer in a regulated firm will want to test first, since it determines whether a number in a finished model can be followed back to the passage it came from.
The product page shows how a user starts. There is a chat entry point labelled "Ask anything"18, a control labelled "Auto-source" for gathering inputs19, and a set of prompt chips. Three of those chips name a task directly: "Build an Excel model"20, "Draft a presentation"21 and "Research a topic"22. A fourth reads "Use a skill"23, which indicates that a stored way of working can be called from the same box. Read together, these labels describe a single prompt surface from which a user can ask for a spreadsheet, a deck or a piece of research.
Matrix is presented differently. Its page runs under the headline "Manage the INFINITE WORKFORCE"24, and the interface shown there includes a label reading "Source Documents"25 and a control reading "Add row"26. The home page closes its Matrix section with an invitation to get in touch from anyone ready to start building their own matrices27. The vocabulary points to a working grid that a user assembles from source material, one row at a time.
Hebbia also describes the collaborative side of the product. The home page invites users to collaborate with the entire deal team28 and to turn individual AI work into firm wide productivity29. On data scope, it promises documents and financial data, all in one place30. The practical reading is a product intended for shared use across a team working on one transaction, with internal files and market information held together.
Two recent product notes extend where the software can be reached. A blog post dated 1 October 2026 says Hebbia now works wherever a firm does, with a new API and an MCP server31. A later release note, dated 3 November 2026, lists team dashboards, Hebbia in PowerPoint and Copilot Cowork, Projects on mobile and other additions as live32. For technical buyers the API and MCP server are the significant items, because they describe routes for connecting Hebbia to other software. A buyer planning to use either should ask for the current documentation and the access terms that apply to it.
Implementation is an area where Hebbia publishes little. The main call to action on the pricing page is a request for a demo33, and the route into Matrix is also a conversation with the company27. An implementation timeline and a list of setup steps are not publicly stated on the pages we read.
That leaves several practical questions for the sales process. A buyer should ask how long a first deployment normally takes, who at Hebbia carries it out, and what the customer has to prepare. Connection work deserves its own question, because the integrations described below reach into data warehouses, content stores and virtual data rooms, and each of those belongs to a different owner inside a typical firm. A buyer should also ask how a team's own methods are captured, given the company's statement that Max works the way a firm works17 and the presence of a "Use a skill" control on the product page23. The answers will determine the real time to first use more than any headline figure could.
The integrations page sets out the purpose in one sentence: bring market feeds and a firm's internal data into the Hebbia workflow34. The entries on that page fall into three groups, and it helps to read them separately.
The first group is the firm's own information. Snowflake is listed for querying an organisation's warehouse in plain language35, and Databricks is listed for querying a firm's data warehouse in the same way36. SharePoint is listed for search and chat over documents across an organisation37. A separate entry covers search and chat over files stored in an enterprise content cloud38. Outlook is listed as a source of emails, calendar events and contacts from a user's account39.
The second group is deal material. Datasite is listed for search and chat over deal documents held in its virtual data rooms40. For a banking or private equity team this is the entry closest to live transaction work, and it is also one of the relationships Hebbia has announced formally, as described below.
The third group is market and regulatory content. Fitch Solutions is listed for credit ratings, rating history and rating actions, searchable and monitored continuously41. Expert call transcripts are listed as available with service access42, which is a condition a buyer should clarify, since it suggests that some content depends on a separate entitlement. Regulatory filings are covered for European public companies43 and, for US public companies, the list runs to 10-Ks, 10-Qs, 8-Ks, proxy statements and other filings44. Quarterly earnings call transcripts of public companies complete this group45.
The blog adds three dated partnership announcements in the weeks before this check. Octus was announced on 13 October 202646, Datasite on 20 October 202647 and Databricks on 23 October 202648. The announcements show recent activity on the partner side, and the integrations page has a separate route for companies that wish to become a data, research or technology partner49. The page's search box has an empty state reading "No integrations match your search"50, so a buyer can test a required system by name directly on the page before raising it with the company.
Hebbia's security page lists ISO/IEC 42001:202351, SOC 2 Type 252, ISO 27001:202253 and CCPA54. GDPR appears on the home page55. These are the company's own listings. The security page links to a Trust Center56, which is the place to request the underlying reports, and also links a Data Processing Agreement57 and a Fair Use Policy58.
On encryption, the security page says data is encrypted at rest and in transit59. The home page gives the detail: AES 256 at rest and TLS 1.2 or later in transit60.
Three further statements concern how customer data is treated. Hebbia says documents, prompts and outputs are never used to train its models or those of any model provider it works with61. It says customer data is stored in siloed environments, fully isolated from every other customer62. And it says Hebbia runs across the US and EU, with regional processing available for storage and inference63,64. For a European buyer the mention of inference is the useful detail, because storage location and processing location are separate questions in a data protection review.
Customers that need a stricter arrangement have one further option on the same page: Hebbia can deploy on a dedicated tenant if needed65. The wording is conditional66, and the conditions, the price and the lead time are matters to settle with the company during procurement.
The governance statements on the security page are short and specific. Hebbia says it has no direct access to customer data and that production access is restricted through just in time access management67. It also says customers can see exactly who accessed their data, when they accessed it and what they did3.
These two statements address different risks. The first concerns the vendor's own staff, and the second gives the customer a record of activity on its data. In a procurement review, both should be followed up with a request for evidence: the access procedure behind the first statement, and a sample of the log behind the second, including its export format and retention period. Questions about administrator roles, approval steps and identity provider setup also belong in that review, and they are best put to the company alongside a request for the reports held in its Trust Center56.
Pricing is not publicly stated. Hebbia has a pricing page, and it offers a demo request33, but when we saved it the heading loaded and the content did not, returning the message "Failed to load page data from Contentful"68,69. A plan, a price or a per seat figure did not appear on any of the other pages we read.
The practical consequence is that a buyer will have no list price to work from at the start of a conversation. A commercial discussion should cover the unit of charge, the treatment of the content sources described above, including those marked as available with service access42, and any additional cost for regional processing63 or a dedicated tenant65.
The company names its segments consistently. The home page lists investing, covering public and private markets70; banking, covering M&A, coverage and capital markets71; legal, covering transaction advisory72; and corporate, covering in house finance and strategy teams73. The Matrix page navigation uses the same four labels74.
Its customer page shows named people from each kind of firm it describes. Sonja Renander, Managing Director at OHA, appears for investors75. Dan Durocher, Partner at Ice Miller, appears for law firms76, and a customer quotation on the same page says that onboarding Hebbia is reshaping the approach to corporate due diligence and other legal work77. Rebecca Leiba, Senior Managing Director at Provident Healthcare Partners, represents healthcare investing78. A customer story is titled "Okapi Partners Sharpens Campaign Precision with Hebbia"79, which extends the list beyond deal work. On the home page, the chief technology officer of Apogem Capital is quoted as saying "It used to take our team 12 hours a day to go through diligence docs"80. These are customer statements published by the vendor.
Taken together, the published evidence places Hebbia with teams whose daily material is financial documents, data room contents and market data, and whose output is a model, a presentation or a report17,30. A buyer outside those four segments would need to ask for a comparable reference.
Two gaps are recorded in our notes. Implementation timelines and setup steps are not publicly stated on the pages we read, and pricing content did not load on the pricing page69. Any commitment on either point has to come from Hebbia in writing.
Two limits are stated by the company's own wording. Regional processing is named only for the US and EU64, so a buyer with a requirement in another jurisdiction, the UK included, should raise it early. The dedicated tenant is offered on an if needed basis66. The about page's usage figures could not be matched reliably to their labels in our saved copy, which is why none appears in this profile15.
Hebbia's documentation site at docs.hebbia.ai did not resolve when we tried it, so this assessment rests on the public pages read (the list is in the source notes): the home page and the pages for pricing, security, product, about, customers, integrations, careers, blog, Matrix and resources.
Hebbia describes a platform for investors, bankers, lawyers and in house finance teams9,70,71,72,73. Its public material is detailed on content sources34 and on security, where the company lists SOC 2 Type 2, ISO 27001:2022 and ISO/IEC 42001:202351,52,53 and states that customer data is never used for model training61. A buyer should open the Trust Center early56, test a real deal folder for traceability back to source2, and obtain the implementation plan and the commercial terms in writing, since neither is published. Include a difficult document set in that evaluation, and check whether a reviewer can trace each material finding back to its source. Confirm which market-data entitlements and regional-processing options the quote includes.
We distinguish dated supplier documents, direct company confirmations and our own analysis. Documentation tells us what a supplier says; it does not prove how the product performs. Where the evidence leaves a question unanswered, we explain what your buying team still needs to ask.
Who will set up the research process, connect its sources and teach your team to review the findings?Ask the supplier to answer this for your purchase. Our research has not established the answer.
Which launch milestones and support terms will Hebbia agree to in writing?
Ask who handles discovery, setup, testing, support and changes, and what your team must have ready first.
Hebbia lists financial-data and document connections including FactSet, PitchBook and SharePoint.1This is the supplier’s statement in material checked November 8, 2026. We have not tested it ourselves.
Does the quote cover every market feed, data room and internal source your team needs?
Ask to see the records and actions you need. Check sign-in, permissions, limits, custom work and who maintains each connection.
Hebbia describes tracing findings to their sources and seeing who accessed data and what they did.2,3This is the supplier’s statement in material checked November 8, 2026. We have not tested it ourselves.
Can a reviewer follow a disputed conclusion back to its source and examine the access history?
Request evidence for the exact product. Ask to see access refused, approval requested, permission removed and the resulting action traced.
Hebbia states that regional storage and inference are available in the US and EU. It describes separate customer environments and a dedicated tenant if needed.4,5This is the supplier’s statement in material checked November 8, 2026. We have not tested it ourselves.
Which region and tenant arrangement will the contract cover, including content sources and model processing?
Request a diagram of the data’s route, with hosting regions, model providers, retention rules and who gets access.
What is the full price, including content licenses, setup, regional processing and any dedicated tenant?6,4Ask the supplier to answer this for your purchase. Our research has not established the answer.
What can your team export before the agreement ends, and which source licenses will it still need?
Ask for an itemized proposal covering setup, software, usage, support, your own staff’s work, changes and what you receive when the service ends.