Comparison
Legal AI under $100 per month is real, and there is more of it than there was a year ago. White Shoe AI’s Associate plan is $49 per month for 1 user and 40 billable hours; the Partner plan is $99 per month for 3 users and 130 shared hours, which works out to $33 per user per month, or $26.33 billed annually. Pricing is published, there is no sales call, and there is no seat minimum. What follows is every product we could find with a published price we could open and check, what you should expect to get at that price, what you should not, and what none of us can yet tell you about accuracy.
Last verified August 1, 2026
The table
Sorted ascending by the low end of per-user monthly cost. Each figure is quoted from the page cited beside it, as of the date shown. Nothing here is estimated or inferred.
| Product | Published price | Per user / month | Seat minimum | Published accuracy benchmark | Source & date |
|---|---|---|---|---|---|
| LegesGPTBasic tier; Premium is $69.99/month | $13.99/month (Basic) | $13.99 | None | No published benchmark | thelegalprompts.com pricing comparisonverified Aug 1, 2026 |
| White Shoe AI — Partner3 users and 130 shared billable hours per month | $99/month for the plan ($948/year) | $33 ($26.33 billed annually) | None | No published benchmark | whiteshoe.ai/pricingverified Aug 1, 2026 |
| Genie AIPro tier, 10 documents/month; free tier at 2 documents/month | $38/month (Pro) | $38 | None | No published benchmark | thelegalprompts.com pricing comparisonverified Aug 1, 2026 |
| White Shoe AI — Associate1 user and 40 billable hours per month | $49/month ($468/year) | $49 ($39 billed annually) | None | No published benchmark | whiteshoe.ai/pricingverified Aug 1, 2026 |
| The Legal PromptsEssential tier at 100 document generations; Strategic tier at $99/month | $49/month (Essential) | $49 | None | No published benchmark | thelegalprompts.com pricing comparisonverified Aug 1, 2026 |
| MyCase IQRange starts at the $100 line rather than below it. | $100–130/seat | $100–130 | None | No published benchmark | irys.ai price tableJun 2026 |
Where a figure comes from a third-party comparison page rather than the vendor’s own pricing page, the source column says so. Prices move. Confirm on the vendor’s site before you buy.
Two products people expect to see here are missing, and the reason is the same in both cases. Clio Duo is widely listed at roughly $49–59, but irys.ai labels its own figure “an independent 2026 estimate” on a gated price, so it does not meet the rule for this table. Vaquill describes its pricing as published and self-serve at one seat with a seven-day trial, but no per-seat figure resolved on its pricing page when we checked on August 1, 2026. A number we cannot point you to is a number we will not print.
The accuracy column reads the same in every row for a reason. No product at this price publishes an independent hallucination benchmark, White Shoe included. That is the subject of the next section.
General-purpose assistants such as ChatGPT and Claude are a separate category and sit below this range — free tiers with Pro plans at $20/month (thelegalprompts.com, verified August 1, 2026). They are excluded from the table because they are not legal-specific: no legal task structure, no matter context, and no confidentiality terms written for client work.
The part that should worry you
Stanford RegLab researchers ran a pre-registered evaluation of the leading paid legal research tools and published the results. General purpose models “hallucinated between 58% and 82% of the time on legal queries”. Among purpose-built legal tools, “the Lexis+ AI and Ask Practical Law AI systems produced incorrect information more than 17% of the time, while Westlaw’s AI-Assisted Research hallucinated more than 34% of the time” (Magesh, Surani, Dahl, Suzgun, Manning and Ho, “Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools”, Journal of Empirical Legal Studies 22 (2025) 216–242). Two things follow. The testing was run in 2024 and every product measured has shipped many versions since, so read those rates as a measurement of a moment rather than a current scorecard. And the tools measured are the expensive ones. Paying five to ten times the price of anything on this page did not buy immunity.
No product in the table above publishes an independent hallucination benchmark. Not LegesGPT, not Genie AI, not The Legal Prompts, not MyCase IQ, and not White Shoe. We are stating that as a fact about this price tier rather than as a defense of our own position in it: the accuracy column is the one column no vendor on this page can fill in for you, and ours is as empty as everyone else’s.
So verify instead of trusting. Four steps, and you can run all of them on any product on this page tonight. One, open every citation in the output against primary law before the work leaves your desk. Two, ask the vendor in writing whether output is grounded in a document you supplied or generated from model memory. Three, ask whether a passage-level source link is exposed for every assertion, so a reviewer can open the underlying text rather than take the summary on faith. Four, treat any vendor that will not answer the second or third question in writing as unverified, and price that into the decision. This applies to us as much as to anyone else listed here.
White Shoe AI is not a law firm and does not provide legal advice. Outputs are informational and should be reviewed by a qualified legal professional before use.
Be straight about the ceiling
The buyer at this price point is not afraid of spending $49. They are afraid of buying something that does not work. So here is the ceiling, stated plainly, including the part that costs us.
Nothing at this price licenses and maintains primary law with editorial treatment. The two legal AI products that carry a research corpus are sold with one: CoCounsel is bundled with Westlaw and Lexis+ AI with Lexis, and neither publishes a price. If your work requires cite-checking against a licensed citator, that is a separate subscription and a separate budget line. Treat it as an addition to an under-$100 tool, not a substitute for one.
At $49 to $99 a month there is no solutions architect, no scoped onboarding project, and no bespoke integration build. You configure the product yourself. That is the trade that makes the price possible, and it is the right trade for a team that would rather start on Tuesday than run a six-week procurement. Teams that need an implementation partner should expect a quote-only vendor.
Under-$100 plans meter something — documents, generations, or hours. White Shoe meters billable hours: 40 per month on Associate, 130 shared per month on Partner. Unused hours do not roll over. Read the meter before you buy, on any product on this page, and size the plan against the work you actually have.
White Shoe does not currently have a SOC 2 report. There is a readiness program in progress, and readiness work is not certification. If your procurement process requires a completed SOC 2 Type II before signature, White Shoe will not clear it today, and we would rather you learn that here than in week three of a security review.
On the SOC 2 point specifically: the security page separates controls that are implemented today from readiness work that is still in progress, item by item. A buyer who reads that and buys anyway is a buyer who stays.
For context
These products are built for a buyer with an implementation budget and a procurement function. That is a legitimate business, and for a large firm it is often the right one. It is simply a different buyer. Where a vendor does not publish a price, this table says so rather than reprinting a third-party estimate.
| Product | Published price | Published accuracy benchmark | Source & date |
|---|---|---|---|
| Irys | $299/seat/month | No published benchmark | irys.aiverified Jun 2026 |
| Paxton AI | $499/user/month | No published benchmark | vaquill.ai price table, listed as vendor-publishedverified Aug 1, 2026 |
| GC AI | $500/seat/month | No published benchmark | gc.ai, best legal AI tools for in-house counselupdated Jul 29, 2026 |
| LegalOngc.ai reports an individual plan published at $550/month billed annually in June 2026, and that LegalOn has since removed its public pricing page. The vendor page is no longer fetchable, so the figure is not carried into the price column. | Not published (quote-led as of July 2026) | No published benchmark | gc.ai, best legal AI tools for in-house counselverified Aug 1, 2026 |
| Harvey | Not published | No published benchmark | Vendor publishes no price; third-party estimates exist and are not repeated hereAug 1, 2026 |
| Legora | Not published | No published benchmark | Vendor publishes no price; third-party estimates exist and are not repeated hereAug 1, 2026 |
| Spellbook | Not published | No published benchmark | Vendor publishes no price; third-party estimates exist and are not repeated hereAug 1, 2026 |
| CoCounsel | Not published / bundled with Westlaw | No benchmark for CoCounsel itself. Stanford measured a different Thomson Reuters product, Westlaw AI-Assisted Research, at more than 34% (tested 2024). | Vendor publishes no price; third-party estimates exist and are not repeated hereAug 1, 2026 |
| Lexis+ AI | Not published / bundled with Lexis | Produced incorrect information on more than 17% of queries in Stanford RegLab testing (tested 2024). | Vendor publishes no price; third-party estimates exist and are not repeated hereAug 1, 2026 |
What you do get
The specification, so you can check it against the work you actually have.
Task-specific agents for the work that fills a day — contract review, diligence, policy drafting, research memos, correspondence. The same set on every plan.
The organization-context layer, in three parts: company profile, style rules, and knowledge base. It is what keeps output sounding like your team rather than like a chatbot.
The Associates run in Web, Email, Slack, Microsoft Word, Chrome, and iOS. The tool goes where the work already is instead of asking you to move.
Checkout is self-serve. No sales call, no seat minimum, no quote process, no annual commitment required to start.
One thing sits outside that list. White Shoe also runs an OAuth-protected MCP server that hands approved workspace context to supported third-party AI clients. It is not a surface the Associates run in, and unlike the surfaces above, the tools it exposes vary by plan, granted OAuth scope, and workspace permissions.
The listings are public, so you can verify the surfaces without taking our word for it: Chrome Web Store, Apple App Store, Microsoft Marketplace, and Legaltech Hub.
Why the price matters
Adoption is no longer the question. Active use of generative AI among in-house legal professionals “has jumped to 52% this year, more than double the 23% reported in 2024” (ACC / Everlaw 2025 GenAI Survey, N = 657 in-house legal professionals across 30 countries, published October 14, 2025).
Measurement is the question. In a related ACC / Everlaw survey of 284 chief legal officers, general counsel, and legal operations professionals worldwide, “only 12% of in-house teams track technology return on investment” (published November 17, 2025). That is the honest case for a published, low-commitment price: when you cannot yet prove the return, the rational move is to lower the cost of finding out rather than to commit to a number you cannot defend.
It is also why the meter matters more than the headline price. A plan you outgrow in week two is not a bargain. Size the hours against real work, then decide.
Direct answers
Among legal AI products with a published price you can open and check, the lowest per-user figures as of August 1, 2026 are LegesGPT at $13.99/month for its Basic tier, White Shoe AI's Partner plan at $33 per user per month ($99/month for 3 users), and Genie AI at $38/month for its Pro tier. The lowest price is not the same as the best fit: the narrower the tool, the fewer kinds of work it covers.
Every product built on a large language model can produce a fabricated citation, and price does not fix it. Stanford RegLab researchers found that general-purpose models “hallucinated between 58% and 82% of the time on legal queries,” and that among purpose-built tools “the Lexis+ AI and Ask Practical Law AI systems produced incorrect information more than 17% of the time, while Westlaw's AI-Assisted Research hallucinated more than 34% of the time” (Magesh, Surani, Dahl, Suzgun, Manning and Ho, Journal of Empirical Legal Studies 22 (2025) 216–242, tested in 2024). Those are the expensive tools; paying more did not buy immunity. No product on this page publishes a third-party hallucination benchmark, and White Shoe does not either. Until one exists, verify rather than trust: open every citation against primary law before the work leaves your desk, ask the vendor in writing whether output is grounded in a document you supplied or generated from model memory, and treat a vendor that will not answer that in writing as unverified.
Yes. White Shoe AI's Associate plan is $49/month for 1 user and 40 billable hours ($468/year, a $39/month equivalent). The Legal Prompts publishes an Essential tier at $49/month. LegesGPT and Genie AI publish tiers below $40. Clio Duo is often listed around $49–59, but that figure is an independent estimate rather than a published price, so this page does not put it in the table.
None of the products in the table above. White Shoe has no seat minimum and no quote process; you can buy one seat. The Legal Prompts, LegesGPT and Genie AI are each listed with no seat minimum on the pricing comparison cited, and MyCase IQ is listed with none on the irys.ai table. Some third-party roundups print seat-minimum figures for vendors that sell by quote — those are third-party reports rather than vendor-published terms, so this page does not repeat them. Ask the vendor directly and get it in writing.
Four things. No primary-law research corpus with a licensed citator — that is sold separately and the products that bundle one do not publish a price. No dedicated implementation team; you configure it yourself. No unlimited capacity; every plan at this price meters documents, generations, or hours. And in White Shoe's case, no SOC 2 report yet.
No. White Shoe does not currently have a SOC 2 report. A readiness program is in progress, but readiness is not certification and does not establish Type II operating effectiveness. Customer data is encrypted in transit over TLS and encrypted at rest through managed-provider protections, and designated third-party integration credentials receive an additional Google Cloud KMS layer. The production application targets Google Cloud Run in us-east1. The security page states what is implemented and what is not.
31 AI Associates, Firm IQ for organization context, and the Associates running in Web, Email, Slack, Microsoft Word, Chrome, and iOS — 1 user and 40 billable hours per month. There is also an OAuth-protected MCP server that hands approved workspace context to supported third-party AI clients; it is a separate connection rather than a surface the Associates run in, and the tools it exposes vary by plan, granted OAuth scope, and workspace permissions — see whiteshoe.ai/platform/mcp. The Partner plan at $99/month covers 3 users and 130 shared hours. The same Associates are on both plans; the plans differ in seats and capacity.
They sell through a quote process, which is a normal way to sell software configured per customer and to a buyer who expects an implementation team. It is a real product decision, not a failing. It does mean a buyer comparing options cannot price them without entering a sales cycle, which is why every cell for those vendors on this page reads Not published.
White Shoe does not train its own foundation model on customer content. Third-party model-provider handling depends on the applicable provider contract and configuration, which is why subprocessors are disclosed in the DPA rather than covered by a single blanket statement.
White Shoe AI is not a law firm and does not provide legal advice. Outputs are informational and should be reviewed by a qualified legal professional before use.
Built for the working lawyer
Associate is $49 a month. Partner is $99 a month for 3 users. Every number on this page came from a page you can open yourself.