I looked at nineteen tools marketed as "AI for lawyers" this year. Most were a chat box bolted onto a document viewer, priced like enterprise software and sold only after a demo call. Small firms don't have a knowledge-management team or a $40,000 annual research budget; a two-lawyer practice needs the AI to live inside the software it already pays for, or it never gets used. So I split the field into two groups: assistants built into the practice-management system you run your firm on, and specialist tools that do one legal task — drafting, research, intake — better than a generalist. Six earned a place. Here is what each actually does, and where each one stops.
| Tool | Best for | Pricing | Free trial | Standout |
|---|---|---|---|---|
| Clio Duo | All-in-one firm running on Clio | From ~$39/user/mo (Clio Manage, billed annually); Duo on higher tiers | Yes (7-day) | Answers questions across your live matters, contacts, and billing |
| Smokeball | Document-heavy small firms | [Not publicly disclosed at time of writing] | Free trial via demo | Automatic time capture plus 20,000+ court-form library |
| Spellbook | Transactional and contract work | [Not publicly disclosed at time of writing] | Demo/trial on request | Drafts and redlines contracts inside Word |
| CoCounsel | Legal research and document review | [Not publicly disclosed at time of writing] | No (sales demo) | Retrieval-grounded research over Westlaw content |
| Lexis+ AI | Firms already on LexisNexis | [Not publicly disclosed at time of writing] | Demo only | Linked citations back to Shepard's |
| Kirava | Automating firm admin work | [Check current site pricing] | Check site | Turns a sentence into a runnable automation |
Best for: All-in-one firm running on Clio Pricing: Clio Manage from about $39/user/month (EasyStart, billed annually); Duo is bundled into the higher Advanced and Complete tiers, not the entry plan Free trial: Yes (7-day) Standout: Answers questions using your live matters, contacts, time entries, and documents
Clio is the practice-management system a large share of solo and small firms already run on, and Duo is the assistant built on top of that data. That is the whole point. Instead of pasting a matter into a separate chat window, you ask Duo things like "which of my open matters had no activity in the last 30 days" or "summarize the correspondence on the Rossi file," and it answers from what is already in Clio. It also flags time you probably forgot to bill, which for a firm that lives on the billable hour is the feature that pays for itself.
The limit is that Duo is a firm-operations assistant, not a legal-research engine. It will not brief you on case law or draft a motion grounded in authority — for that you still need CoCounsel or Lexis+ AI alongside it. And Duo only helps if you have committed to Clio as your system of record; if your firm's data is scattered across email, Word, and a shared drive, Duo has little to read. You are also paying for the higher Clio Manage tiers to get it, so the real cost is the platform, not the AI line item.
Pros: - Answers questions against your actual matters and billing, not a generic legal corpus - Surfaces unbilled time and stale matters automatically - No new login or workflow — it lives inside software staff already open daily
Cons: - Not built for case-law research or authority-grounded drafting - Only useful if your firm keeps its data disciplined inside Clio
Best for: Document-heavy small firms Pricing: [Pricing not publicly disclosed at time of writing]; quoted per seat after a demo Free trial: Free trial available via demo request Standout: Automatic time capture plus a court-form library reported at over 20,000 documents
Smokeball is the practice-management platform for firms whose day is dominated by documents — family law, estate planning, immigration, small litigation. Its two long-standing strengths are automatic time tracking, which records what you work on without a running timer, and a large library of pre-built court forms and document templates you can auto-populate from matter data. The newer layer is Smokeball AI and its Archie matter assistant, which summarizes matters and drafts routine documents from the information already in the file. For a firm that produces the same twelve document types over and over, that automation is the draw.
The trade-off is pricing opacity and platform lock-in. Smokeball does not list prices publicly, so you cannot compare it against Clio without sitting through a sales call, and its tiers step up in cost as you add features. It is also Windows-first in feel and heavier to adopt than a lightweight assistant — you are buying a whole operating system for the firm, not a bolt-on. If you do not generate a high volume of standardized documents, you are paying for capacity you will not use. Firms that live in contracts rather than court forms will get more from Spellbook.
Pros: - Automatic time capture reduces revenue leaked to unlogged work - Large court-form and template library populated from matter data - Archie AI drafts and summarizes from existing file information
Cons: - No public pricing; requires a demo to get a number - Heavier platform commitment than a standalone AI assistant
Best for: Transactional and contract work Pricing: [Pricing not publicly disclosed at time of writing]; historically quoted per seat on annual contracts Free trial: Demo or limited trial on request Standout: Drafts, reviews, and redlines contracts directly inside Microsoft Word
Spellbook is a Word add-in built for the work transactional lawyers actually do: drafting clauses, reviewing an incoming contract, suggesting redlines, and flagging missing or aggressive terms. Because it runs inside Word rather than a separate web app, it fits the existing drafting workflow instead of asking a lawyer to change how they work. You can ask it to draft a mutual NDA from a plain description, or point it at a counterparty's agreement and ask what is unusual or one-sided. For a small firm doing corporate, commercial, or real-estate paper, that is the daily grind, and Spellbook targets it precisely.
Two honest limits. First, it is a drafting assistant, not a lawyer — every suggested clause needs review, and it can propose language that reads clean but is wrong for the deal. Second, pricing is not public and has generally been sold on annual contracts, which is a harder commitment for a solo than a month-to-month tool. Spellbook is narrower than the practice-management platforms; it does nothing for your calendar, billing, or research. It is a specialist you buy on top of your firm's system, not a replacement for it. If your work is litigation rather than contracts, most of Spellbook's value does not apply.
Pros: - Works inside Word, where contract drafting already happens - Flags unusual or one-sided terms in a counterparty's draft - Generates first-draft clauses and whole agreements from a description
Cons: - No public pricing; typically an annual commitment - Contract-focused — little value for litigation or research
Best for: Legal research and document review Pricing: [Pricing not publicly disclosed at time of writing]; sold through Thomson Reuters sales Free trial: No — sales demo only Standout: Research and review grounded in Thomson Reuters and Westlaw content
CoCounsel, now part of Thomson Reuters after the Casetext acquisition, is the research and document-review assistant of the group. It handles the tasks a junior associate would: legal research questions, reviewing a document set for relevant passages, summarizing depositions, and preparing a first pass at a memo. Its argument for trustworthiness is that answers are grounded in Thomson Reuters' own legal content rather than the open web, which reduces — though does not eliminate — the risk of invented authority. For a firm that does real research and cannot justify a full Westlaw-plus-associate budget, it compresses hours of reading.
The friction is cost and access. There is no self-serve trial and no public price; you go through sales, and the effective cost tends to sit at the high end for a small firm, often bundled with a Westlaw subscription. It is also overkill for a firm that rarely does original research — a family-law or straightforward transactional practice may never touch its strongest features. And despite the retrieval grounding, I would still verify every citation before it goes into a filing; "grounded" is not the same as "correct," and courts have sanctioned lawyers for AI-fabricated cites. CoCounsel earns its place for research-heavy litigation firms, not for everyone.
Pros: - Answers grounded in Westlaw and Thomson Reuters content, not the open web - Reviews and summarizes large document sets faster than manual reading - Handles associate-level tasks: memos, deposition summaries, research
Cons: - No public pricing or self-serve trial; enterprise-style sales process - Cost is hard to justify for firms that rarely do original research
Best for: Firms already on LexisNexis Pricing: [Pricing not publicly disclosed at time of writing]; quoted through LexisNexis Free trial: Demo only Standout: Answers link back to Shepard's citations for verification
Lexis+ AI is LexisNexis' answer to CoCounsel, and the calculus is similar: it does conversational legal research, drafting, and summarization grounded in the Lexis content library. Its most useful trait for a careful lawyer is that answers link back to the underlying authority and to Shepard's, so you can check whether a cited case is still good law without leaving the tool. If your firm already pays for Lexis, adding the AI layer keeps research inside one ecosystem rather than introducing a second research vendor, which is a real workflow benefit.
The obvious limitation is that the value depends heavily on whether you are already a Lexis firm. If you run on Westlaw, CoCounsel is the more natural fit, and running both research ecosystems makes no sense for a small firm's budget. Pricing, again, is quote-based and not public, so comparison shopping means two sales calls. And like every research assistant here, it produces confident prose that still needs citation-checking before filing — the Shepard's linking helps, but it does not remove your duty to verify. Pick Lexis+ AI if you are already inside the LexisNexis world; otherwise the switching cost outweighs it.
Pros: - Citations link back to source authority and Shepard's for fast verification - Keeps research in one ecosystem for existing Lexis subscribers - Handles research, drafting, and summarization in one interface
Cons: - Value collapses if you are not already a Lexis subscriber - Quote-based pricing with no public number or self-serve trial
Best for: Automating firm admin work Pricing: [Check current site pricing at time of writing] Free trial: Check site Standout: Turns a plain-English task description into a runnable automation
Kirava is not a legal tool, and I am including it deliberately. It is a workflow-automation product for small businesses that takes a sentence — "send a reminder when an invoice is 7 days overdue," "email new intake contacts a scheduling link" — and builds the automation for you, instead of making you assemble and maintain a flow on a canvas the way Zapier, Make, or n8n do. For a solo or two-person firm with no operations staff, the value is the boring middle of practice management: intake follow-ups, payment reminders, appointment confirmations, document-request nudges. That work is pure friction, and it is exactly what falls through the cracks when the lawyers are billing.
The caveats are real. Kirava is a general small-business tool, so it has no built-in understanding of legal ethics rules, trust-accounting constraints, or client-confidentiality requirements — you are responsible for making sure an automation does not, say, email privileged information to the wrong list. It also overlaps with automation that Clio or Smokeball may already offer natively, so if you run one of those platforms, check what you can do before adding a separate tool. And because it is a younger product, I have not seen independent reliability benchmarks. Treat it as glue for admin tasks, not for anything touching client matters directly.
Pros: - Builds automations from a plain-English description, no flow-building - Targets the admin friction solos have no staff to absorb - Cheaper and simpler than wiring up Zapier or Make by hand
Cons: - No legal-specific awareness of ethics, trust accounting, or confidentiality - Overlaps with automation already inside Clio or Smokeball
Start with one question: what does your firm do most hours of the day?
If most of your day is running matters, billing, and documents, buy the assistant inside your practice-management system before you buy anything specialist. On Clio, that is Clio Duo; if you are document-heavy and generate the same court forms repeatedly, Smokeball's automation and form library fit better. Do not buy both platforms — pick the system of record first, then use its native AI.
If most of your day is contracts, Spellbook is the highest-leverage single purchase, because it works where you draft — inside Word — and needs no platform migration. A transactional solo can bolt it onto Clio and cover both operations and drafting.
If most of your day is research and litigation, you need a grounded research assistant: CoCounsel if you are a Westlaw firm, Lexis+ AI if you are already on LexisNexis. Choose by the research ecosystem you already pay for; running both makes no financial sense at small-firm scale.
If budget is the binding constraint and you want a published, self-serve price under a demo call, Clio is effectively the only option here — every other tool routes you through sales. Add Kirava only to automate admin friction, and only after checking what your practice-management platform already automates for free. In practice most small firms end up with one platform plus one specialist, not five subscriptions.
Four of the six — Smokeball, Spellbook, CoCounsel, and Lexis+ AI — sell through a demo-and-quote process rather than listing per-seat pricing. That usually signals value-based pricing and annual contracts. Clio is the exception with public per-user rates. Budget for a sales call before you can compare real numbers.
No tool here is safe to file unchecked. CoCounsel and Lexis+ AI ground answers in licensed legal content, which lowers the hallucination rate versus a general chatbot, and Lexis links to Shepard's for verification. Courts have already sanctioned lawyers for AI-fabricated citations, so verify every cite yourself before it reaches a filing.
The established legal vendors — Clio, Thomson Reuters, LexisNexis — publish security and confidentiality commitments aimed at the profession's duties, but you should read the data-processing terms yourself, especially around whether prompts train the vendor's models. For a general tool like Kirava, be extra careful never to route privileged content through it.
No. Each tool shifts a slice of a specific task — drafting, research, time capture, admin follow-up — but vendor claims of large productivity gains are not backed by independent benchmarks I could find. Treat them as leverage on existing staff, not a headcount replacement.
Most small firms do. A practice-management assistant (Clio Duo or Smokeball) covers operations; a specialist (Spellbook for contracts, CoCounsel or Lexis+ AI for research) covers the deep work. Two subscriptions is a realistic setup; five is overspending.
If I were opening a small firm tomorrow, I would run on Clio and turn on Clio Duo as the operations assistant, because it is the one tool here with a published price and no separate login for staff to ignore. From there I would add exactly one specialist based on the actual work: Spellbook if the firm is transactional, CoCounsel if it is research-heavy litigation on Westlaw. I would hold off on Kirava until I saw which admin tasks Clio could not already automate. What would change my mind is a document-first practice — heavy family law or estate planning — where Smokeball's form library and automatic time capture would earn back its opaque price faster than Clio would.
BestAIFor