Hebbia Is Impressive. It's Also Not Built for You.

You have seen the Hebbia demo. Matrix is impressive. The cross-document analysis works. You can see why financial analysts love it.

But something nags. You are three weeks into setup, writing extraction prompts, iterating on column templates, and you still do not have a workflow your associates can use on a live deal. The technology is not the problem. The fit is.

Platforms vs. Products

There is a meaningful difference between an AI platform and an AI product. A platform gives you powerful primitives and lets you build what you need. A product makes assumptions about your workflow and ships something that works on day one.

Hebbia is a platform. An excellent one. It is designed for analysts who want to build custom workflows across any document set. You define the columns. You write the extraction prompts. You iterate on the templates until the output is reliable. Then you share the workflow with your team.

That flexibility is Hebbia's best feature for finance teams. An analyst doing quarterly portfolio reviews on a recurring set of fund documents can invest weeks building the perfect template and then reuse it for years. The setup cost pays for itself many times over.

For M&A deal teams, that same flexibility becomes a structural limitation. Not because the technology is lacking, but because the workflow does not match.

The Setup Tax on Live Deals

Consider what platform setup actually means for an M&A attorney.

You need to decide which provisions to extract from each document type. You write prompts for each one. You test them, find edge cases, rewrite them. You build a column template that structures the output in a way your team can use. You train associates on the interface. You iterate until the workflow is reliable enough to trust on a live deal.

For an analyst doing the same analysis on the same document types every quarter, this investment makes sense. The documents are familiar. The analysis is recurring. The timeline is predictable. You build the workflow once, and it compounds in value.

For a deal attorney, the math is different. Every data room is new. The document mix changes. The signing deadline does not wait for your template to be ready. You need structured output on day one, not after weeks of configuration.

This is not a criticism of Hebbia's design. It is the structural reality of building a platform versus a product. Platforms are powerful because they make few assumptions. Products are fast because they make the right assumptions.

What "Purpose-Built" Looks Like in Practice

When the M&A knowledge is in the product instead of in your prompts, the experience changes fundamentally.

Upload a data room. Get structured extraction across every agreement in minutes. Not because you wrote the right prompts, but because the system already knows what provisions matter in an asset purchase agreement, what red flags to surface in employment agreements, what non-standard terms to flag in IP assignments.

Automatic document classification. Customer agreements, employment agreements, IP assignments, leases, NDAs, equity documents. The system categorizes every document because it was built to understand M&A document types, not because someone configured a classification workflow.

Risk flagging calibrated to deal context. An uncapped indemnity in an IP license gets flagged differently than one in a standard services agreement. The system understands that context because M&A risk assessment is engineered into the product, not layered on by the user.

Deal deliverables that work on day one. Tabular analysis, diligence memos, redlines, disclosure schedules, variance detection. These are not templates you build. They are workflows that ship with the product.

We wrote about this broader dynamic in The F1 Engine Problem: the most powerful AI engine in the world is not useful without the right chassis around it. In M&A diligence, that chassis is domain-specific infrastructure.

An Honest Side-by-Side

Consider a concrete scenario: reviewing 200 contracts in a new data room.

The Platform Approach

1. Upload documents
2. Write column prompts for each provision you want extracted
3. Iterate on prompt wording until outputs are reliable
4. Build a reusable template for your deal team
5. Train associates on the interface
6. Verify outputs against source documents
7. Manually structure findings into your deliverable format

Timeline to first useful output: weeks of setup before your team is productive.

The Purpose-Built Approach

1. Upload data room
2. Select your analysis type
3. Review pre-generated findings with source citations
4. Flag issues and add notes
5. Export directly to your memo or disclosure schedule

Timeline to first useful output: minutes.

The difference is not speed alone. It is architecture. A platform gives you a tool to build analysis. A purpose-built product gives you analysis.

How to Decide

This is a genuine decision, not a foregone conclusion. Both approaches serve real needs.

A platform like Hebbia might be right if:

• You have a dedicated AI or innovation team with time to build and maintain workflows
• You perform the same analysis on recurring document sets (quarterly reviews, portfolio monitoring)
• You want maximum flexibility to customize analysis across diverse use cases
• Your timeline allows for weeks of setup before needing production-quality output

A purpose-built product is right if:

• You are on live deals with signing deadlines that do not wait for configuration
• You need to onboard new associates quickly without training them on prompt engineering
• You want structured M&A deliverables (memos, disclosure schedules, tabular analysis) out of the box
• Your team's time is better spent reviewing analysis than building the system that generates it

Some firms genuinely should use Hebbia. If your workflow looks more like recurring financial analysis than live deal diligence, a platform's flexibility is the right trade-off.

For most M&A deal teams, it is not.

The Right Tool Matches How Your Team Works

The best AI tool is the one that fits the way your team actually operates. For M&A deal teams working new data rooms under deadline pressure, that usually means a product that already knows what to do, not a platform you have to teach.

That is the difference between a platform that can do legal work and a product that does legal work.

---