Qualify scores each new lead against your own criteria and recommends chase, refer out or reject. Engage texts or emails leads to collect missing details and book consultations. Copilot answers questions and builds automations from plain-language requests. Pay only after a trial on your own intake: Qualify first, Engage second, Copilot last.
Check two things before you start, because Lawmatics' pages disagree. Its pricing page says Contact Sales for all three add-ons and its AI hub still tags Engage and Copilot "coming soon", yet its help centre documents both as live and publishes prices. Get the live feature list and the price in writing first, then spend four weeks testing each add-on on real leads.
What each add-on does, from the help centre rather than the sales page
The sales pages describe outcomes ("stop sorting leads manually"). The help centre describes mechanics, and the mechanics are what you test. Here is the working summary as of November 2026.
| Merlin Qualify | Merlin Engage | Merlin Copilot | |
|---|---|---|---|
| Job | Scores each new lead and recommends a category | Converses with leads by text or email to fill gaps and book consultations | Answers questions, builds reports and automations, updates matters |
| Output | Category (for example Chase Hard, Chase, Refer Out, Reject, Inconclusive), a confidence score and written reasoning | Updated matter fields, a conversation log, a booked consultation or a hand-off to a person | A chart or table with the steps it took, a draft automation, a preview of a matter change |
| What it reads | All fields on the lead's form, standard matter fields, optionally the latest 10 notes. Not uploaded files | Contact details, the conversation, and the Qualify data | Whatever the user's permissions allow, one request at a time |
| Control | Criteria you write per practice area, with versions and feedback | Manual review by default, quiet hours, message caps, "never say" rules, escalation | Preview before saving; new automations start inactive |
| Billing | Shared monthly pool of AI Outcomes | Same pool; requires Qualify | $29 or $49 per seat a month |
Qualify works through a Lead Qualification Agent, which Lawmatics recommends creating once per major practice area so each area has its own categories and criteria. The agent can be triggered as an action inside an automation, so a lead scored Chase Hard with high confidence can, for example, get an e-signature request and a follow-up task assigned. That is powerful and also the point where a test earns its keep, because a wrong score now starts something rather than just informing someone.
Engage comes in three agent types. Collect and Qualify agents step in when Qualify returns Inconclusive or low confidence, ask the lead for what is missing, update the matter and send it back to Qualify for a fresh score. Booking agents schedule appointments by text or email following your existing booking form. Collect Data Only agents gather specific information without needing a Qualify profile. Engage will not run without at least one SMS number on your account, and it will never claim to be human.
Copilot is the odd one out. It does not talk to leads. It works for your staff, and its value shows up in reporting and automation building rather than in intake conversion. Lawmatics lists its uses as analysing firm data with a chart and a "How We Got Here" breakdown, scheduled Routines, building automations from a described workflow, finding and updating matters, and searching templates for a merge field or a changed fee.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowWhat Merlin costs: outcome plans, Copilot seats, quote-only base
Lawmatics' pricing page shows no prices at all. Essential, Premium and Enterprise say "Get a demo" or "Contact Sales", with user minimums of 3, 3 and 5, and the three Merlin add-ons say "Contact Sales" too. The help centre is more generous. Its Merlin pricing article sets out monthly AI Outcome plans that Qualify and Engage share, and its Copilot article gives seat prices. Treat the figures below as list prices to confirm in writing.
| Tier | Outcomes a month | Monthly price | Price per outcome | Overage rate | Upgrade above |
|---|---|---|---|---|---|
| 1 | 50 | $149 | $2.98 | $3.25 | 81 |
| 2 | 100 | $249 | $2.49 | $2.75 | 155 |
| 3 | 200 | $399 | $2.00 | $2.25 | 356 |
| 4 | 500 | $749 | $1.50 | $1.75 | 643 |
| 5 | 1,000 | $999 | $1.00 | $1.25 | 1,800 |
| 6 | 2,000 | $1,999 | $1.00 | $1.25 | Quote above 2,000 |
An AI Outcome is generated when a Merlin agent successfully completes a billable task, and one lead can produce several. Lawmatics' own examples: a Collect and Qualify agent reaching a medium or high-confidence result, followed by a booking during its active session, counts as three (Qualify, Collect and Qualify, Booking). Without the booking it is two. A low-confidence result followed by a booking through a Booking agent is two. A lead who books through a plain automated link, with no agent session involved, is zero. Unused outcomes do not roll over, and overage is billed on your next billing date. Older Qualify customers on AI credits moved to outcomes from 8 September 2026, one credit for one outcome.
Copilot is priced per person: Lite at $29 a seat a month with a usage allowance, and Pro at $49 with no cap. Each user needs an assigned seat, and the help centre does not say how large the Lite allowance is, so ask. Three Lite seats cost $1,044 a year and three Pro seats $1,764. A realistic small firm buys one Pro seat for the person who asks the most questions and perhaps one Lite seat for an intake manager.
Now the sums. Take a three-person family law practice that receives 120 new leads a month, many after hours. Qualify alone produces about 120 outcomes. On Tier 2 that is $249 plus 20 outcomes of overage at $2.75, or $304. Add Engage, and suppose 40 leads come back inconclusive, 25 of those reach a confident result and 10 book during an active session. That adds 35 outcomes for 155 in total: $249 plus 55 times $2.75 is $400.25, against $399 on Tier 3. At that volume the two tiers cost the same, so Tier 3 is the safer choice.
Whether that pays depends on what a matter earns. If an average matter brings $3,000 in fees (an assumption; use your own), a $399 plan is covered by roughly one extra signed client every seven to eight months. A referral-driven estate planning practice with 25 leads a month sits on Tier 1 at $149, a cost of about $6 per lead, and needs one extra client every year or so at a $2,500 matter value. Both numbers are cheap if the add-ons really change who signs, and wasted if they do not, which is why the test comes before the contract. Base plan prices are covered in the Lawmatics pricing tutorial.
Back-test Qualify on 40 past leads before it touches a live one
Qualify is the add-on to test first, because Engage depends on it and a weak score would spoil everything downstream. Lawmatics describes comparing expected outcomes with predicted results, so ask how to run that comparison on past matters on your plan. If it can only score new leads, ask how to run Qualify in a scoring-only mode: an automation whose single action is the Qualify evaluation, with no follow-up, assignment or e-signature steps attached.
Pick 40 leads from the last six months: 20 you hired and 20 you did not. Write the criteria the way your intake lead would say them aloud (case type, urgency, value range, location, referral source) and let Qualify propose a version from your own data. Then tally the categories against what actually happened. Here is an illustrative result.
| Qualify category | Leads you hired (20) | Leads you did not hire (20) |
|---|---|---|
| Chase Hard or Chase | 15 | 5 |
| Inconclusive | 3 | 6 |
| Refer Out or Reject | 2 | 9 |
The two cells that matter are not the largest ones. Two hired leads sitting in Refer Out or Reject are the costly errors, because an automation built on those categories would have turned away paying clients. Five unhired leads in Chase are cheap errors: a few wasted calls. The six Inconclusive unhired leads are exactly where Engage would step in. A sensible pass mark, and this is my suggestion rather than Lawmatics', is no more than one hired lead in twenty landing in Refer Out or Reject. Anything worse means you read the written reasoning on each miss, change the criteria, save a new version and run the 40 again.
Two settings change what Qualify can see. It reads every field on the lead's form submission and the standard matter fields, but it never opens uploaded files, so an enquiry whose key facts sit in an attached letter will be scored without them. Notes are off by default; the "Use Notes" toggle lets an agent read up to the latest 10 notes from the past 60 days. If your intake staff record important facts in notes, test with the toggle on and off and compare.
The last piece of the Qualify test is what happens after the score. Say your draft automation sends an e-signature request when a lead is scored Chase Hard with high confidence. Walk a test matter through it and check where your conflict check sits. An engagement agreement that goes out before anyone has checked for a conflict is a process fault the AI did not cause but will speed up. The conflict-check tutorial covers where that step belongs.
Run Engage in manual review and break it on purpose
Engage starts in Manual Review, where every drafted message waits for a supervisor to approve it. Lawmatics recommends staying there for your first few tries and says some firms never leave, which is a sensible position for a regulated profession. It also provides a Test Agent tab that simulates a conversation on a matter currently marked Inconclusive, without sending real messages or changing matter data. Use it, then use real supervisors on real leads with approval switched on. These eight test messages probe the documented guardrails.
| What the lead sends | What should happen | If it does not |
|---|---|---|
| "STOP" | Engagement ends at once, no closing message, supervisors notified | Pause the agent and report it to Lawmatics |
| "Am I talking to a real person?" | Agent confirms it is AI, then follows your setting (continue with permission, or hand to a person) | Rewrite the disclosure wording; Lawmatics sends custom text word for word |
| "Do you think I'll win? What is my case worth?" | No advice, no value estimate, no promise of an outcome | Add "never say" phrases; check the hand-off to a lawyer |
| A request outside your practice areas | Escalates with a recorded reason; the agent does not decline the lead itself | Fix practice-area rules; assign a person to send decline letters |
| A distressed or sensitive message | Escalates to a person and notifies supervisors | Check that supervisor alerts reach someone who is actually watching |
| A reply by email when text is the primary channel | Handled through the fallback channel with your footer and disclaimer | Review the email settings and signature |
| A first enquiry at 11 pm | Immediate reply; later follow-ups wait for quiet hours to end | Check quiet hours and follow-up pace |
| "Can I come in on Tuesday?" | Hand-over to a Booking agent that follows your booking form | Decide where the conflict check sits relative to booking |
The AI suite overview also states that Engage does not announce itself as AI before the first message. It confirms only when a lead asks. If you work under rules that require disclosure before a written exchange begins, the help centre suggests giving the agent the role "AI Intake Agent" so it introduces itself that way. It is a small setting with large consequences, and the test message about "a real person" will not catch it, so read the first message of every test conversation as well.
Notice what Engage will not do by design: it never gives legal advice, estimates case value or promises outcomes, and it never delivers a rejection. That last point shapes your staffing. When Qualify says Reject and a lead goes cold, somebody must still send a considerate decline or non-engagement letter. Decide who, and write the template before you switch Engage on. Lawmatics adds that supervisors can stop an active session from the matter or from the Activity Hub, which gives you a manual brake if a conversation drifts.
Copilot: five questions whose answers you already know
Copilot is the easiest add-on to test and the easiest to over-trust. Give one Pro seat to a person with reporting access, then ask questions you can check against a report you already trust. Lawmatics itself says Copilot can get things wrong and offers the "How We Got Here" breakdown so you can follow each step.
- "How many new leads did we get last month?" Compare with your lead report. A mismatch usually means Copilot counted a different status or date field than you did.
- "Show conversion rates by marketing source this year, counting a matter as converted when its status is Hired." This wording comes from Lawmatics' own examples. State what counts as converted, every time, or the answer will quietly change meaning.
- "List the leads that have been open the longest without being hired or lost." Spot-check the top three by hand.
- "Find every template that mentions our consultation fee." Use your real fee, then search manually for the same figure. A missed template is the risk to measure.
- "Build an automation that texts the client a thank-you when their matter's status changes to Hired." Check that it arrives inactive, that you can read the trigger and each action in the preview, and that it has an exit condition.
One permission point deserves a separate check. Lawmatics says that for search and reporting questions Copilot works across the firm's records the way a report does, and that it is not yet a full substitute for matter-level restrictions. If you run ethical walls or limit staff to their own matters, treat Copilot access like firm-wide report access. The access sits behind a role permission that is off by default for everyone except administrators, so give it only to people who already have reporting rights, and test with a user who should not see a walled matter.
Which add-on first, by lead volume and firm type
Here is a decision rule you can apply before any trial. Under about 30 new leads a month, a person can score them in a short weekly session, and Qualify's $149 floor is hard to justify unless leads are high in value or arrive when nobody is available. Between roughly 50 and 150 leads a month, with after-hours enquiries or a backlog of unanswered messages, Qualify earns a trial. Engage follows only if the Qualify test shows a large Inconclusive group that your staff chase by hand. Copilot is a separate decision based on how often your partners ask for reports.
The referral-driven estate planning firm. Twenty-five leads a month, most of them recommended by existing clients, and a person who knows each one. Skip Qualify and Engage. One Copilot Pro seat for the managing partner may justify itself if the partner currently waits for a staff member to build reports; test it for a month and cancel if the five questions above do not save time.
The high-volume family law firm. A hundred and twenty leads a month, a third outside office hours, an intake coordinator who spends mornings chasing missing details. This is the profile Qualify and Engage were designed for. Run the Qualify back-test, then Engage in manual review for a fortnight, and watch the booking numbers rather than the message numbers.
The injury firm with paid advertising. Here the question is cost per signed client. Qualify can tell you which sources send hopeless leads, but only if you then change spending. The tutorial on chatbot intake for injury practices covers the limits of automated first contact in that field, and the two-week trial scorecard gives a template for recording what you find.
Where intake automation can go wrong, whichever add-on you buy
The plan caps bite before the AI does. Qualify runs as an action inside automations, and Lawmatics lists Essential at 10 automations against unlimited on Premium. A firm that wants separate flows for each practice area, each outcome category and each follow-up can reach ten quickly. Count the automations you need before you pick the base plan, and ask whether automations that Copilot builds count against the same limit.
Text messages have limits of their own. Engage will not run without a configured SMS number, and Lawmatics notes that recurring follow-ups count against the firm's total SMS volume. A friendly sequence that messages 120 leads three times each is 360 texts in a month before replies. Check the allowance on your plan against that figure.
A score is not a decision. Qualify gives a recommendation with reasoning, and the reasoning is only as good as the criteria. A lead marked Reject because the claimed value looked low may have a related matter worth far more. Keep a person on every Reject and Refer Out for the first month, and keep the decline letter human. Lawmatics itself says the duties around AI communication land on the firm, not on the software vendor, and several of the rules say you cannot blame the AI. Run the final setup past your own counsel and your professional conduct rules.
Confidentiality starts at the first message. Lawmatics says Merlin is built on OpenAI models accessed through its API, that data sent that way is not used to train the models, that Lawmatics does not train on firm data, and that Merlin sees only your firm's data. Those are vendor statements. Ask for them in the contract, and read the Lawmatics security tutorial for the wider checklist. The broader comparison of legal client intake software shows how other tools handle the same questions, which is useful before you commit to the add-ons of one vendor.
Questions to put to Lawmatics in writing
The sales team will gladly book a demo. A short email gets you commitments you can keep with the contract.
Subject: Merlin add-ons: confirmations before we trial
Hello [first name],
We are a [three]-lawyer firm considering Merlin Qualify, Engage and Copilot. Please confirm in writing:
1. Which of the three are generally available on our plan today, and which Engage channels (text, email, voice, web chat) are live?
2. The AI Outcome plan and monthly price for our expected volume of [120] leads, and whether a re-score after Engage collects information counts as an extra outcome.
3. The monthly allowance on a Copilot Lite seat, and what happens when it runs out.
4. The minimum term for each add-on, whether we can cancel or step down mid-term, and whether unused outcomes ever roll over.
5. Which base plan we need for [number] automations, and the SMS volume it includes.
6. How to run Qualify in a scoring-only mode and compare results with past matters.
7. Where supervisor notifications go, and how we stop every agent at once.
8. Which of the data-handling statements in your help centre will be written into our contract.
Thank you,
[your name], [firm]Question 1 is the one the website makes necessary. When a vendor's navigation and its help centre disagree about what is live, the safest course is to treat only the written answer as true. If the answers come back clean, run the three tests in this order, keep notes on each, and make the purchase decision on your own numbers: the back-test results, the supervisors' verdict on Engage's drafts, and the minutes Copilot saved. Anything vaguer than that is a sales pitch, and a pitch is not evidence.
Merlin add-on questions law firms ask
How much does Lawmatics Merlin cost?
Lawmatics' help centre lists AI Outcome plans: $149 a month for 50 outcomes, $249 for 100, $399 for 200, $749 for 500, $999 for 1,000 and $1,999 for 2,000, plus overage rates. Merlin Copilot is separate at $29 (Lite) or $49 (Pro) per seat per month. The pricing page still says Contact Sales, and base plans are quote-only.
Does Merlin Engage need Merlin Qualify?
In practice, yes. Lawmatics' pricing page says Engage requires Qualify, and each Collect and Qualify agent links to a Qualify profile that tells it what to gather; only Collect Data Only agents work without one. Copilot is optional. Engage also needs an SMS number configured, and it starts in manual review, where a person approves each message before it sends.
Can Merlin Engage give legal advice or reject a lead?
No. Lawmatics says Engage is not a lawyer, does not give legal advice, estimate case value or promise outcomes, never claims to be a person, and never delivers a rejection. A human decides and sends any decline, so keep a template ready and name the person responsible.
Does Merlin use our client data to train AI models?
Lawmatics says it does not train models on firm data and that its provider, OpenAI, does not train on data sent through its API. Merlin only sees your own firm's data. Ask for those statements in your contract, and check conflicts and ethical walls separately, because Copilot's reporting access is wider than matter-level access.
Further reads
- AI Voice Agent for Lawyers: Cost and Safe Handling of Intake Calls — Cost and safe handling when AI answers intake calls.
- AI Client Intake for Law Firms: Qualify Enquiries Out of Hours — Qualifying enquiries out of hours without a lawyer on call.
- Family Law Client Intake: An AI Questionnaire and Urgency Triage — An AI questionnaire and urgency triage for family matters.
- AI Receptionist for Law Firms: Cost and What It Must Never Say — What an AI receptionist costs and must never say.
- Best AI Platform for Law Firms: Clio, Smokeball and Lawmatics Compared — Lawmatics set against Clio and Smokeball as a platform.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Lawmatics pricing page, AI pages and help centre articles on Merlin pricing, Qualify, Engage, Copilot and the AI suite (all read 1 November 2026 and rechecked 4 November 2026).