Hebbia vs Shortcut (2026)
The decision is whether an investment team needs AI to establish what its sources say, or AI to build the model those facts feed. Hebbia does the first. Banks, investment firms and law firms use it to put questions to filings, data rooms and market data feeds. Each answer links to the passage behind it. Shortcut does the second. An analyst asks for a three statement model, a DCF or an LBO, and its agent writes live formulas into the workbook, waiting for approval on any tab set to manual. The two sit side by side more than they compete. Hebbia's Max agents can produce an Excel model once the research is done, and Shortcut lifts figures from filings with the page shown. A firm could run both, Hebbia for diligence reading and Shortcut for the build. Shortcut posts its rates and has a free plan, while Hebbia sells named user licenses with no published price. Hebbia can keep data in EU regions, and Shortcut's standard service runs only in the US. Pick Hebbia when the hard part is reading widely and proving each claim. Pick Shortcut when the hard part is building the model.
- The same question has to go to hundreds of documents. Hebbia's Matrix splits a hard question into steps, sends agents through each document in full, and returns a cited answer for every one.
- The facts sit in paid data and deal systems as well as in filings. Hebbia reads FactSet, PitchBook and Fitch data, along with DealCloud and the files in Datasite and Intralinks rooms.
- Part of the work falls under European data rules. Hebbia can run storage and inference in EU regions on AWS, and its data processing addendum names the two EU regions it uses.
- Every seat should carry the same training terms. Hebbia rules out training on customer content for its own general models and for its model providers, and zero data retention is its default for most model calls.
- Research should carry on into deliverables. Hebbia's Max assistant, rebuilt in July 2026, has agents that produce Excel models and decks from work done on the platform, and they can run on a schedule.
- Bankers need the model itself, not notes toward it. Shortcut puts together the DCF, the LBO with its debt schedule and the three statement model inside Excel, and it also refreshes comps and updates models each quarter from filings.
- Every figure should lead back to its arithmetic. Shortcut leaves formulas in the cells instead of fixed values, so a reviewer can follow any number to the calculation behind it.
- A model should get a second look before it goes out. Shortcut's Auto Review checks the agent's workbook changes, and a financial model audit skill added in September 2026 reviews the finished model.
- A trial should not wait on a contract. Shortcut's free tier carries 20 credits a week, and Pro adds 1,000 monthly credits and Platform API access for $100 a month billed annually.
- Not every analyst lives in Excel. Shortcut also works in Google Sheets and a web app, and since August 2026 it handles Word and PowerPoint files as well.
This comparison is published by AI FinTech Index, an independent research platform that publishes independent ratings of AI vendors for financial services. Hebbia and Shortcut are each graded against the same capability taxonomy, from each vendor's own public materials and the regulatory record, under the AI FinTech Index verification standard. No vendor pays for placement, and no vendor has reviewed this page. How this evidence is graded
Plain facts
| Hebbia | Shortcut | |
|---|---|---|
| Primary category | Capital Markets & Research AI | Capital Markets & Research AI |
| Founded | 2020 | 2023 |
| Headquarters | New York, New York, United States | San Francisco, California, United States |
| Website | www.hebbia.com | shortcut.ai |
Side by Side
Select any grade to read the note behind it.
| Axis | H Hebbia |
S Shortcut |
|---|---|---|
| AI Centrality | ||
| Autonomy and Oversight Model | ||
| Model Risk Management and Transparency | ||
| Operational and Outcome Evidence | ||
| AI Safety and Data Stewardship | ||
| GLBA and Data Privacy Posture | ||
| Security Certifications and Trust Center | ||
| Regulatory Status and Licensure | ||
| AI Governance and Bias Disclosure | ||
| AI Liability and Recourse | ||
| Model Supply Chain Disclosure | ||
| Core Systems and Integration Depth | ||
| Deployment Model and Data Residency | ||
| Commercial Transparency | ||
| Institution and Segment Coverage |
Product updates
The latest logged changes from each vendor, newest first, each linked to its source.
Hebbia
-
Integration / interoperability
Hebbia added a Databricks integration that lets users ask plain language questions of a firm's own Databricks data, such as portfolio metrics, positions and fund performance, alongside documents, filings and market data.
Source -
Integration / interoperability
Hebbia now integrates with Octus through the Octus MCP Connector, which links every Octus product a firm subscribes to into Hebbia at once.
Source -
Product / capability
Hebbia's October 2026 release adds memory to Max, its agent, which stores and recalls context across conversations under user controls and an org level admin switch, version history with rollback for decks and docs made in Max, and a Microsoft Word plugin for running workflows and drafting inside Word.
Source
Shortcut
-
Product / capability
Shortcut 1.4.2 made recurring workflows available, so users can set its Excel agent's tasks to run on a schedule instead of starting them by hand each time.
Source -
Product / capability
Shortcut release 1.2.1 introduced Sho, an autonomous agent with its own cloud computer that takes work by voice or text, carries it out in the background and remembers conversations and notes between sessions, in early alpha for organizations with voice mode enabled.
Source -
Model / architecture
Shortcut release 1.1.0 added Claude Sonnet 5.5, GPT-6 Sol and GPT-6 Luna to its model picker and made background tasks follow the user's chosen model by default.
Source
The short version of each
Hebbia
Hebbia is an AI research platform that banks, investment firms and law firms use to question large sets of documents and data. Morgan Stanley and Centerview Partners are among the banks that use it, and OHA, New Mountain Capital and 26North among the investment firms. It reads filings, data rooms and document stores such as SharePoint and Box. It also reads market data from FactSet, PitchBook and S&P Global. Its agents work through each document in full and return answers with a preview of the source passage. Max agents can then carry that research into an Excel model, a deck or a document. Hebbia is hosted on AWS with a choice of US or EU regions, and its data processing addendum names Google, OpenAI, Microsoft and Anthropic as model suppliers. The caveat is accuracy on hard material. Hebbia publishes nothing on how its answers hold up on scanned exhibits or poorly structured filings, and it posts no price.
Source: AI FinTech Index, 2026
Shortcut
Shortcut is an AI agent that builds and edits financial models in Excel. Its maker, Fundamental Research Labs, began as an MIT spinout in late 2023. Hedge funds, private equity firms and banks lead the buyer list on its site, ahead of asset managers, FP&A teams and accounting firms. An analyst or associate describes the job in plain words. The agent plans the steps, then writes formulas, formatting and charts into the workbook. Typical jobs include a DCF valuation, an LBO with its debt schedule and a three statement model. Any tab can be set so each change waits for approval, and saved checkpoints let a team roll the workbook back. Rates are public, with a free plan and Pro at $100 a month billed annually. Enterprise adds the firm's own encryption keys, its own storage bucket and a private model gateway. The caveat is judgment. Shortcut does not describe which assumptions its agent picks, which figure it trusts when sources disagree, or how it fills gaps.
Source: AI FinTech Index, 2026
Common questions
What is the difference between Hebbia and Shortcut for an investment banking team?
Hebbia finds and sources the facts, and Shortcut builds the model. A banking team uses Hebbia to read across filings, data room files and market data, and each answer comes with a preview of the passage behind it. Morgan Stanley and Centerview Partners are among Hebbia's bank customers. A team uses Shortcut to build and update the model itself, from three statement models to LBOs with debt schedules. Shortcut's case pages, one of them written for investment bankers, run from accretion and dilution work to month end close. The inputs differ too. Hebbia reads FactSet, PitchBook and Fitch data and deal systems such as DealCloud. Shortcut names no link to a market data provider or deal system. It draws on filings, the files a team gives it, web search and connectors such as Gmail. A bank could run both, with Hebbia for diligence and Shortcut for the model. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified November 4, 2026. No vendor pays for placement.
Can Hebbia build a financial model the way Shortcut does?
Hebbia comes at the model from the research side. Its Max agents can turn analysis done in Hebbia into an Excel model, a deck or a document, and they can run on their own or on a schedule. Hebbia also offers Office add ins through Microsoft's store. Shortcut is built around the workbook. An analyst describes the job, and the agent plans it, then writes formulas, formatting and charts into Excel or Google Sheets. Its usual work is a three statement model, a DCF, an LBO with debt schedules or a comps refresh. Neither Hebbia nor Shortcut says how often a model its agents build contains an error. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified November 4, 2026. No vendor pays for placement.
How does an analyst verify a figure from Hebbia or from Shortcut?
At Hebbia, the check runs back to the source. Each answer carries a preview of the passage it came from, and version history can roll a change back. Workflows built in Matrix 2.0 also wait for a person's approval. Hebbia describes no confidence score per answer and no sampling guidance for large batches of answers. At Shortcut, the check covers the arithmetic as well as the source. Any tab can be set to hold each change for approval. Figures sit in live formulas a reviewer can follow, and each extracted value shows the filing, date and page. Auto Review checks the agent's edits, and an audit skill reviews a finished model. Shortcut says an internal benchmark, ShortcutBench, gates every production release, and it has not published that benchmark or its method. Hebbia's 92 percent accuracy figure, from OpenAI's account of its work, has no published test set either. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified November 4, 2026. No vendor pays for placement.
Can a bank keep its data in Europe, or inside its own systems, with Hebbia or Shortcut?
Hebbia answers the first need, and Shortcut the second. Hebbia hosts the work on AWS, with storage and inference available in either US or EU regions. A firm that wants a dedicated tenant can have one. It makes the case for SaaS over on premises and offers no customer cloud option. Shortcut's standard service is hosted in US data centers, and no other region is offered. Its Enterprise plan can store files in a Google Cloud Storage or Amazon S3 bucket the firm owns, encrypt stored data with keys the firm holds, and send model traffic through the firm's own gateway. Shortcut's homepage mentions an on premises version too, though no guide to setting it up is published. Both keep their SOC 2 reports behind a request. Hebbia commits that customer content trains neither its general models nor its providers' models. Shortcut keeps Enterprise data out of training, while Pro and Teams accounts start with a training setting switched on. The AI FinTech Index traces Pivot, Shortcut's own finance model, back to Qwen 3.5, an open model from Alibaba. Shortcut post trained it on free tier usage only and serves it in house.
How are Hebbia and Shortcut sold, and what do they cost?
Shortcut sells from a public rate card, and Hebbia sells named user licenses with no published rate. Shortcut's free plan comes with 20 credits each week. Pro runs $100 a month when billed annually and $125 month to month, with 1,000 credits a month. On annual billing, one analyst pays $1,200 for the year. Teams is $320 a month plus $100 per seat when billed annually, and groups under 15 seats can sign up without a sales call. A message usually takes between 2 and 15 credits. Enterprise is quoted. Hebbia lists no plan, rate or seat price. Whether a feed such as FactSet needs a license of the firm's own is not addressed, so put that in the written quote beside the seat price. Graded by AI FinTech Index against the same capability axes from each vendor's own published materials, verified November 4, 2026. No vendor pays for placement.
Related comparisons
Other published head to head assessments involving these vendors or their closest peers. The full set for this category is on the Capital Markets & Research AI page.
Hebbia and Shortcut leave the same three gaps. Neither publishes how often its agents get an answer or a figure wrong, measured by a stated method. Neither says how agent work should be documented for a firm's examiners. Neither commits to anything when a figure is wrong. Hebbia's master agreement excludes damages from inaccurate data, and Shortcut's terms supply output as is. Both limit what they owe to 12 months of fees.
For Hebbia, the points to put in writing are how long customer data is kept while the contract runs, what Max agents may change when they run on a schedule, and what accuracy it sees on scanned exhibits. For Shortcut, they are what ShortcutBench measures, whether an Enterprise service level can cover the accuracy of a model, and whether conversations kept for training before the setting was switched off can be deleted.