Quick answer: Build a test set of 30-50 documents with known correct answers, run each candidate tool against it, and measure citation accuracy, recall of material clauses, and false-confidence rate. Demo accuracy rarely survives contact with real files.

This is the framework I would use before signing an annual contract. It takes two weeks and it is the cheapest part of the decision.

What should your test set contain?

Pull from matters you already closed, where you know the outcome. Include the messy ones: scans of scans, documents with unusual formatting, contracts in a second language if you handle them. A test set of only clean PDFs will flatter every tool equally and tell you nothing.

Which metrics actually matter?

Citation accuracy is non-negotiable. A tool that cites real, correct passages but misses some clauses is usable with review. A tool that invents authorities is not usable at all, at any speed.

How do you design the test?

1. Pick 10 contracts with a partner’s annotated review as ground truth.
2. Run each tool blind — no vendor engineer sitting in.
3. Have a mid-level associate score outputs against the annotations.
4. Repeat on 5 documents outside your usual practice area.
5. Compare not just accuracy but how long verification took.

That last step is where honest answers live. If verifying the AI’s output takes nearly as long as reading the contract, the tool has not saved you anything.

What about data handling and confidentiality?

Ask for the deployment model in writing: where does data live, is it used for training, what happens on deletion, and can it run in your environment. Tools like Harvey offer enterprise controls, and general assistants usually do not. If a vendor is vague here, treat it as a red flag rather than a gap to be filled later.

How do you compare against doing nothing?

Run the manual baseline. Take the same 10 contracts, have an associate review them unaided, and record the time and the clauses found. Without this number you cannot tell whether the tool helped or just moved work around. It is often the most illuminating part of the whole exercise.

What are the warning signs?

• Demos run only on the vendor’s cherry-picked documents.
• No citation, or citations you cannot click through to a source.
• Accuracy claims without a defined test set behind them.
• Vague answers on training data usage.
• A pilot that quietly excludes your hardest documents.

How should you roll out?

Even after a good pilot, start narrow: one workflow, one team, mandatory human review on every output for the first month. Track error types, not just error counts. If the same class of mistake keeps appearing — dates, defined terms, cross-references — that tells you what to constrain in the workflow rather than abandoning the tool.

Who this framework is for: legal teams evaluating any AI vendor, whether Harvey or a general LLM with careful prompting. Who can skip it: nobody buying an enterprise legal tool should skip it, but individual practitioners doing occasional drafting have less to lose.

Quick Summary

• Test on 30-50 of your own documents, including messy ones.
• Citation accuracy and false confidence matter more than raw speed.
• Measure verification time, not just drafting time.
• Establish a manual baseline to know if the tool helps at all.
• Roll out one workflow at a time with mandatory human review.

Quick answer: Harvey is an enterprise product, sold to firms and legal teams, not a self-serve app with a free tier. It is trained and tuned for legal work and runs inside a controlled environment with your firm’s data.

The honest framing: Harvey is a serious tool for a specific buyer. If you are an individual or small firm looking for a $20 subscription, this is not the product.

What does Harvey do?

Its core workflows are legal research, document review, contract analysis, and drafting. You can ask it to summarize a merger agreement, flag unusual clauses against your playbook, or draft a first-pass memo. It also supports uploads of your own documents so answers are grounded in your files rather than general training data.

How does a contract review workflow look?

1. Upload the contract, or point Harvey at a repository.
2. Ask for a summary and a list of clauses that deviate from standard positions.
3. Review the flagged items and drill into the source passages.
4. Export a memo for the partner or client.

Harvey cites the passages it used, which is the feature that matters. Uncited legal AI is unusable, because you cannot verify a claim you cannot see the basis for.

How does it compare to general chatbots?

The difference is not raw model quality. The same underlying models are available to a general chatbot. The difference is grounding, provenance, and the contractual terms a firm can actually accept.

What does it cost?

Harvey does not publish public pricing. It is quoted per firm, typically on an annual contract with seat-based tiers. Expect it in the enterprise budget line, not the tool-expense line. If you need a number, you need a sales conversation.

Where does it fall short?

• It is not a substitute for a lawyer’s judgment. It speeds review; it does not assume responsibility.
• Coverage of unusual jurisdictions may be thinner than a specialist’s knowledge.
• Output still needs verification against primary sources.
• Small firms may find the pricing model and minimums out of reach.

How should a firm adopt it?

Start with one workflow — contract summarization or first-pass review — on a small set of matters. Have a senior associate check every output for a month. Track time saved honestly, not by optimism. Roll out wider only after the accuracy is proven on your own document types, which vary far more between firms than vendors imply.

Who this suits: firms and in-house teams with high document volume and the budget for a managed platform. Who should skip it: solo practitioners and small firms — general LLMs with careful prompting and mandatory verification cover a surprising share of the same ground for far less.

Quick Summary

• Harvey is an enterprise AI platform for legal research, review, and drafting.
• It grounds answers in your uploaded documents and cites source passages.
• Pricing is quoted per firm; no public or free tier.
• Verification remains mandatory — it assists, it does not advise.
• Pilot one workflow on real matters before a firm-wide rollout.