You are evaluating a new legal AI tool for your law practice. You open G2. You open Capterra. You expect to find dozens of detailed user reviews, just as you would for any general CRM or billing software. Instead, you find almost nothing.
This is not a glitch in your search process. It is a defining characteristic of the legal technology market. Out of the 43 legal AI tools tracked by LegalTechMag, exactly 2 carry a verifiable public aggregate star rating.
Those two exceptions are Gavel, which holds a 4.9 out of 5 rating on Capterra across 51 reviews, and WealthCounsel (Wealth Docx), which has a 4.4 out of 5 rating on G2 across 59 reviews. Both figures are current as of June 23, 2026. The other 41 tools have no verifiable aggregate score on any major review platform.
This review vacuum makes traditional software due diligence impossible. However, the silence is not a sign of poor product quality. It is a structural feature of the legal technology market. Understanding why this vacuum exists will help you vet these tools safely.
Why Most Legal AI Tools Have No Public Ratings
The absence of reviews in this sector is driven by five distinct structural causes. These causes are built into the business models of legal software developers and the professional realities of legal practice.
Reason 1: Legal AI Software is Genuinely New
The vast majority of systems in this category are very young. Core platforms like CoCounsel Legal, EvenUp, Supio, Paxton AI, Spellbook in its generative AI format, and Everlaw with its advanced AI features either launched or underwent complete technical rebuilds between 2023 and 2025.
Software review directories typically require 12 to 18 months of active use by a broad customer base before meaningful aggregate scores accumulate. The legal AI category has simply not had the time to build this volume.
Reason 2: Confidentiality and Attorney Risk-Aversion
Attorneys face strict professional and ethical standards regarding client files. A marketing manager can easily write a public review of a database tool. A lawyer cannot easily write a public review detailing how an AI performed on sensitive discovery documents in an active litigation matter.
Discussing real matters raises client confidentiality issues. It also invites scrutiny from malpractice carriers. Because of these risks, the practitioners who are best positioned to write detailed reviews are the least likely to post them online.
Reason 3: Direct Sales Distribution Lacks Review Triggers
General business software platforms rely on self-serve sign-ups. They use automated email campaigns to request reviews right after onboarding or at key user milestones.
The legal AI market operates differently. It is run by direct sales representatives and customized demos rather than self-serve sign-up buttons. Without a self-serve onboarding funnel, vendors have no automated mechanism to ask users for reviews. This direct-sales approach matches the dynamic we analyzed in our article on why so many legal AI vendors hide their pricing.
Reason 4: Platform Pay-to-Play Dynamics
Directories like G2 and Capterra operate as advertising businesses. Vendors must pay to be featured, to run automated review outreach campaigns, and to access full platform analytics.
A young legal AI startup spending its capital on expensive compute costs and sales hiring often chooses not to buy into these vendor programs. A missing listing is usually a budgeting decision, not a sign of a poor product.
Reason 5: The Over-Reliance on Curated Vendor Case Studies
Because independent reviews are rare, vendor-designed case studies carry too much weight in buyer research. These documents are selected and written by the vendor's marketing department.
While they feature real law firms and actual litigation, they showcase only the most successful reference clients under ideal conditions. You must treat these case studies as promotional materials rather than objective peer reviews.
The Trap: What the Two Rated Exceptions Reveal
To understand the review gap, look closely at the only two exceptions. Gavel and WealthCounsel share specific operational patterns. Both focus on document automation. Gavel is a general document automation platform, while WealthCounsel offers a highly specialized estate-planning library.
Both tools have been in the market much longer than newer generative AI systems. They also built their user bases through self-serve options. Gavel offers a 7-day free trial with no credit card required. WealthCounsel has a subscription model that has run long enough to accumulate 59 G2 reviews organically.
The data shows a clear rule. Public ratings only appear when a tool has both self-serve access and years of market presence. They are virtually absent when a tool relies on direct sales and is highly recent.
Because systems like Kira, CoCounsel Legal, Everlaw, Spellbook, EvenUp, Supio, and Paxton AI lack statistical review volume, buyers fall into a trap. They rely on curated demos, vendor-sponsored analyst reports, and conference buzz. The buyer lets the vendor control the entire flow of information. The lack of reviews does not mean these tools are poor quality, but it does mean you must change your evaluation process.
The Actionable Method: How to Vet Without Reviews
To evaluate a legal AI system safely, you must abandon standard review directories. Use these seven concrete tactics to perform independent due diligence.
1. Request Practice-Specific Reference Customers
Do not accept generic reference calls. Ask the vendor to connect you with an active user who matches your firm size, your practice area, and your jurisdiction.
For example, a large personal injury firm using EvenUp to draft demand letters is not a relevant reference for a three-lawyer workers' compensation practice evaluating the same tool. If a vendor cannot produce any matching reference, treat it as a warning sign.
2. Run a Time-Boxed Paid Pilot
This is the single most important action. Request a 30 to 60 day pilot using active client files. You must secure proper data-handling agreements first.
Do not rely on vendor-curated demo data, which is designed to hide software limitations. Real files will show you where the AI fails. Even if the vendor normally requires annual contracts, ask for a paid trial period anyway. Respectable vendors will usually accommodate the request.
3. Check for a Free Trial
Some vendors like Gavel offer a 7-day trial with no credit card. Most enterprise tools do not offer self-serve trials.
A missing trial is not a disqualification, but it means you must negotiate a custom pilot to see the software. You can check which tools offer open access in our legal AI for solo and small law firms buyer's guide.
4. Verify Security Posture and Data Handling in Writing
Demand to see a SOC 2 Type II certification report and a formal Data Processing Agreement (DPA). Secure a written commitment detailing how your matter documents are used.
Specifically, verify whether your data is used to train proprietary machine learning models. The Stanford Law School guide on navigating AI vendor contracts highlights this data ownership issue as a critical negotiation point. Never accept verbal security promises.
5. Test Failure Modes with Independent Benchmarks
Do not rely only on vendor claims. Look at independent scientific tests.
The Vals AI legal benchmark, first published in February 2025, evaluated several tools on document extraction, summarization, and redlining. The follow-up study in October 2025 found that leading legal AI tools scored 78% to 81% accuracy on legal research tasks compared to an 80% ChatGPT baseline. Use these third-party benchmarks to calibrate your expectations.
6. Search Bar Association Listservs and Forums
Peer networks are highly valuable. Check American Bar Association listservs, local bar tech committees, and active online forums like r/Lawyertalk.
Post direct questions about the specific tool and your exact workflow. A single candid answer from an active practitioner is worth more than dozens of anonymous reviews on general software directories.
7. Evaluate the Vendor's Reference Volume
A reliable vendor should provide five to ten active reference customers willing to speak with you. If a vendor offers only one or two corporate accounts and some website quotes, their reference list is too thin. Ask directly how many paying customers they support in your practice area.
FAQ
Why doesn't CoCounsel (or Harvey, or Kira) have any reviews?
These tools are enterprise-grade systems that launched or underwent complete AI rebuilds between 2023 and 2025. They rely on direct sales representatives rather than self-serve online sign-ups, which means they lack automated email funnels to solicit reviews. Furthermore, their users are attorneys who work with confidential client files. These attorneys are highly risk-averse and rarely write public reviews of tools that process proprietary case documents. The lack of reviews is entirely structural and does not mean the software is untested.
Are the G2 and Capterra ratings that do exist trustworthy?
Yes. The 4.9 out of 5 rating for Gavel on Capterra and the 4.4 out of 5 rating for WealthCounsel on G2 are based on verified, organic reviews from real users. However, you must understand the context of these reviews. Both Gavel and WealthCounsel are mature document-automation tools that offer self-serve onboarding. Because of this, their reviewers are mostly solo attorneys and small-firm practitioners. The ratings are highly reliable, but they may not reflect how the tools perform in massive enterprise environments.
How many real users should a vendor's reference list have?
If you are evaluating an early-stage tool for a pilot, the vendor should have at least five to ten reference clients. If you are preparing to sign a major annual contract, ask to speak with at least three reference customers who match your firm size and practice area. A vendor who cannot connect you with three comparable users is likely shielding a very small install base. Understanding the size of their reference pool is a crucial part of your software due diligence.
Is a tool with no public reviews automatically risky?
No. It is the standard state of the market. Out of 43 legal AI tools we track, 41 have no public aggregate rating. This includes market leaders used by major law firms, such as Kira and CoCounsel. The absence of ratings is a product of market youth, enterprise sales structures, and strict attorney confidentiality rules. It does not indicate poor quality. It simply means you must abandon review platforms and do the active vetting work yourself through reference calls and paid pilots.
The Bottom Line
The review vacuum in legal AI is not going away soon. With 41 out of 43 tools lacking a public rating, you cannot use G2 or Capterra as a shortcut for your technology decisions. This vacuum is a structural reality of the market, not a sign of bad software.
To protect your firm and your clients, you must run a disciplined vetting process. Demand relevant customer references, negotiate a time-boxed paid pilot with real matter files, and secure written data-handling commitments.
Vetting a tool manually requires more effort, but it ultimately works in your favor. By running real documents through these platforms yourself, you will learn far more about their practical strengths and failure modes than any five-star rating could ever tell you.