Law firm intake and conflicts automation: what to buy, and what to build.
New-matter intake is two bottlenecks, not one. Missing the after-hours call is a marketing problem, and a legal answering service fixes it for a few hundred dollars a month. Conflicts clearance is a compliance problem no answering service touches, and it is where the matter actually stalls. Buy the first; commission the second. ColabContent LLC publishes this page: a boutique AI consulting house in Boston that builds custom systems for law firms of 10 to 150 attorneys.
The $499 AI-Ready Audit comes first; after it, the system that matters most is proven as a working prototype on your own data before any build fee. Production builds are one fixed fee from $10,000, one time, with the code owned by the firm at handoff and no per-seat licence. The $499 AI-Ready Audit is ordered at colabcontent.com/ai-ready-audit/.
Written for managing partners, operating partners and firm administrators at practices of 20 to 150 attorneys. What legal intake services genuinely cover, where clearance swallows the days, and what must be built against your own matter data.
What the numbers say and where each path fits.
Every vendor sells one product called intake. Inside the firm the word covers two different failures, with two owners, two price tags and two buying decisions.
The first is that nobody picked up. A prospective client calls at 6:40pm, or during a deposition, and the call goes to voicemail. That is a marketing loss, it is solved by a staffed or AI answering service, and that market is mature and cheap relative to a build. If this is your problem, buy the service.
The second is that the caller was captured perfectly and the matter still did not open for four days. Somebody had to search the party names against everything the firm has ever touched, chase two partners in trial, decide whether an old representation was substantially related, get an engagement letter out, and only then issue a matter number. No answering service touches any of that, because every step is either a judgment reserved to lawyers or a query against data that exists only inside your firm.
Firms conflate the two because the client experiences one delay. Buy an answering service to fix a clearance bottleneck and you will capture bad matters faster. Commission a build to fix an after-hours gap and you will pay $15,000 (an example entry-level fee, not a quote) for something sold monthly.
Key Terms
Data residency: the physical location where client data is stored and processed; a compliance requirement for firms handling matters subject to GDPR or state privacy laws. Billable hour recapture: the revenue recovered when AI captures time entries that attorneys would otherwise forget to log; industry surveys commonly cited in legal trade press put unrecorded billable time in a 10 to 30 percent range, treat it as an estimate rather than a figure specific to any firm. Practice management integration: connecting AI tools with the firm's existing case management, billing, and document systems so data flows without rekeying. Model hallucination rate: the frequency at which an AI system generates plausible but factually wrong legal citations or case holdings; the metric that separates usable legal AI from a liability.
The calls nobody answered, and why you should buy that fix.
Somebody ran the experiment. For the 2024 Legal Trends Report, a research company contacted 500 US law firms as a prospective client, by phone and email. Only 40 percent answered the phone, down from 56 percent in 2019. Just 33 percent replied to the email, down from 40 percent. Forty-eight percent were essentially unreachable by phone, neither picking up nor calling back.
Those firms were not short of demand, they were short of an answering function, and the loss lands on the highest-intent contact a firm ever receives. Somebody who dials a law firm has already decided they need one.
The correct response is almost never a build. Legal answering is a competitive, published-price market. Smith.ai lists Starter at $300 a month for 30 calls with overage at $11.50 per call, and Pro at $2,100 for 300 calls. Ruby's published receptionist tiers run from $250 a month for 50 minutes to $1,725 for 500. Against a fixed fee starting at $10,000 the arithmetic is not close if coverage is your only problem, and our AI receptionist versus custom build page runs the crossover on Smith.ai's published per-call pricing, and Smith.ai alternatives for law firms compares six vendors that publish real rates, in the three different units they bill by.
The second reason to buy has nothing to do with money. Answering is the one part of intake with no firm-specific logic in it. Greeting a caller, taking a name and booking a slot is the same job at every firm in the country, which is why a product does it well. The moment a step depends on your matter history the product stops fitting, and that moment is the conflicts search.
Conflicts clearance, and why no answering service can help.
Conflicts clearance is not paperwork. It is the firm deciding whether it is permitted to act, under rules that impute one lawyer's disqualification to everyone in the building. Model Rule 1.10(a) provides that while lawyers are associated in a firm, none of them shall knowingly represent a client when any one of them practicing alone would be prohibited from doing so by Rules 1.7 or 1.9. That is why the search cannot be scoped to the originating attorney's book.
Rule 1.18 extends the exposure to people who never became clients. A person who consults a lawyer about the possibility of forming a client-lawyer relationship is a prospective client; the firm owes confidentiality on what it learns whether or not a relationship follows; and information that could be significantly harmful to that person can bar the firm from acting adversely in the same or a substantially related matter. ABA (American Bar Association) Formal Opinion 492, issued in June 2020, frames the test as information which could be significantly harmful rather than harm the person can prove.
Now put that next to an intake script. Most scripts are optimized for conversion, which means they invite the caller to tell their story, and under Rule 1.18 the story is the disqualifying information. The rule's own escape routes are design decisions made before the call: 1.18(d) lets the rest of the firm proceed where the consulted lawyer took reasonable measures to avoid exposure to more disqualifying information than was reasonably necessary, is timely screened, and written notice reaches the prospective client promptly. Parties first, facts later.
The mechanics are equally unfriendly to off-the-shelf tools. Intapp, writing in February 2025 on conflicts checks and client intake, notes that professionals checking conflicts in the firm's finance system are not necessarily seeing the corporate trees of existing or prospective clients, so the subsidiary that creates the conflict is invisible to the search. It flags the common workaround of emailing every lawyer about a prospective client, which risks sharing sensitive information with lawyers who should not have it. And it names the failure that surfaces years later: manual conflict review does not automatically create an auditable record.
How do you automate a law firm conflicts check?
Automating a conflicts check means automating the search and the record, never the decision. In practice a working system does five things. It captures every party the caller names during intake, including opposing parties, insurers, spouses and corporate parents, rather than only the person on the phone. It normalises those names, so that Acme Corp, Acme Corporation and ACME all resolve to one entity. It queries the firm's own matter history, client list and prior prospective contacts in a single pass rather than making a person search three systems. It returns ranked possible hits with the matter, the date and the relationship that triggered each one. Then it writes the whole run, hits and clears alike, into a timestamped record attached to the prospective matter. A lawyer still clears the hit, judges whether a matter is substantially related and decides whether a waiver is available, because Model Rule 1.10(a) makes that a legal judgment and no vendor can take it. What the automation removes is the two hours of searching and the undocumented gap between the caller hanging up and somebody remembering to check.
This is also why law firm intake automation and conflicts check automation are two different purchases that get sold as one. The intake half, answering the call and capturing the details, is a competitive market with real published prices and you should buy it. The conflicts half runs entirely on your own party data, your own matter history, your own corporate-tree knowledge and your own waiver conventions, which is exactly the material no off-the-shelf product has. That asymmetry, not a feature gap, is what decides which half you buy and which half you build.
The six steps between "we want to hire you" and a matter number.
Buy the intake product, or commission the layer underneath it?
The sections below walk through the buy or commission question for intake: what the packaged intake products actually cover and where they stop, the situations where commissioning the layer underneath makes more sense than buying another product, and what a governance clean build looks like architecturally once conflicts data is involved.
What the intake products cover, and where they stop.
There is a real product category here and it should be evaluated before anything is commissioned. Legal Technology Hub describes new business intake as the complex set of administrative processes required to onboard a new client or matter, and names Intapp Intake and Clio Grow as two of the leading solutions: one for firms with a dedicated intake function and a risk team, the other for firms whose intake is one person and a spreadsheet.
These products own the workflow around the decision well: the form, the queue, the approval chain, the status. What they represent less well is your firm's idea of what a hit is. Party matching against your naming conventions, the corporate trees your industries create, and a search crossing a document system and a finance system nobody designed to be queried together.
So the ask in a demo is narrow: run this on my last twenty declined matters and show which ones you would have caught. Not their data, yours. A product that clears that bar is the cheaper answer. Our build, buy, or commission framework is the three-way version, and off-the-shelf AI SaaS cost at scale runs the arithmetic on what a rented product costs as more seats are added.
When commissioning is the right call.
Commission when the search must span systems no single vendor owns. A firm running Clio or Litify for matters, iManage or NetDocuments for documents, and Aderant or Elite 3E for billing holds party data in three products with three ideas of what a client record is; stack detail is in the Clio, Litify, iManage and NetDocuments playbooks.
Commission when conflicts knowledge lives in people rather than fields. Most mid-market firms have an administrator who knows this client is the parent of that entity, and that this insurer sits behind half the defense book. That is a retrieval problem with a real answer, the same shape as knowledge retrieval over a firm's own documents.
Commission when elapsed time rather than search time is the cost. If clearance takes four days because two partners are in trial, the build that repays itself is routing and escalation, which is workflow automation rather than legal tech.
Do not commission if your party data is a mess; a build on top inherits it. We say so on the first call and decline the work. Longer version: data readiness for mid-market AI and what we do not build.
What a governance-clean build looks like architecturally.
Four decisions separate a system a general counsel signs off on from one that becomes a finding later, and all precede any code.
Where the data comes to rest. Party names and matter history are confidential client information under Rule 1.6. Keep them in the firm's own tenant (a private cloud account), send the minimum to any model provider, and get that provider's retention and training terms in writing.
The human decision gate. The system searches, ranks and presents. A lawyer clears. Enforce that boundary in software, not in a policy document, because a policy document is not what an auditor examines.
The audit log as a first-class output. Every search writes an immutable record of inputs, sources queried, results, decision, decider and timestamp. That is what answers a disqualification motion, and the largest gap between an automated process and a manual one.
Walls applied at creation. If clearance produced a screen, apply it when the matter and workspace are created, not by a follow-up ticket. Retrofitted walls are the ones that leak.
The wider vendor question list is security questions to ask before an AI build, and the ownership argument is why code handoff matters.
What it costs, and what we have and have not built.
Below are the published fee bands for this kind of work, rather than a quote on request process, alongside a plain accounting of what has actually been built here and what has not, so a firm can weigh the cost against a realistic, not aspirational, track record.
The fee bands, published rather than quoted on request.
Fixed fee against a written scope, in three bands. A focused build over one clearly defined system, for example a party-matching and conflicts search layer over a single matter database, runs from $10,000 over 4 to 5 weeks. An operations rebuild covering capture through matter opening, with routing, escalation and engagement letter assembly, is quoted after the $499 AI-Ready Audit over 6 to 8 weeks. A multi-system platform with a custom interface for the intake team is quoted after the $499 AI-Ready Audit over 10 to 14 weeks.
Set that against the answering line, a few hundred dollars a month at the entry tiers Smith.ai and Ruby publish. Most firms in this band should be paying both bills, for different problems. Breakdown by scope: AI consulting cost for law firms and the pricing page.
Source code, prompts, models and the architecture document transfer to the firm at handoff, which is what makes the data path auditable permanently rather than for the length of a subscription. The five phases from first call to handoff are on the process page.
What we have actually built, and what we have not.
This matters more than usual, because half of this guide tells you to buy from somebody else.
What we have shipped in law firms is the capture and routing half. The nameable reference is Jim Glaser Law, where we built five channel-specific voice agents covering PPC, Organic, TV, Meta and LSA, giving per-channel attribution on answered calls rather than form fills. Those agents have handled 3,787 calls and 5,514 minutes. Jimmy will take a reference call and does refer. Across all clients our systems have handled more than 6,000 live calls. To hear one, our own receptionist answers at (617) 675-9067.
The most relevant engagement is one we can describe but not name: a litigation firm whose matter, client, invoice and IOLTA (the client trust account a law firm must keep separate) trust accounting platform we commissioned and which the firm now owns. That is a regulated financial-record system running under money-handling rules a state bar audits, the closest analogue we have to the standard a conflicts system would be held to.
What we have not built is a production conflicts clearance engine. We have built against matter and party data, and the architecture above is the design we would bring, but no firm has yet run our conflicts search in production and we will not imply otherwise. If a shipped, referenceable conflicts build is a requirement, say so on the first call and we will tell you where we stand. More on the vertical: AI for mid-market law firms and what we would commission first at a $30M law firm.
Start with the $499 audit.
No slides. We walk your intake from the first ring through the matter number, name the step costing the most elapsed days, and say whether it is a subscription problem or a build. If an answering service would serve you better, we say so.
Frequently Asked Questions
Short answers first, detail underneath. Every answer here matches the FAQ schema on this page word for word, and each one is the answer we give on the audit call. Where a question depends on your own numbers, the $499 AI-Ready Audit report replaces the general answer with your figures.
What do legal intake services actually cover?
Almost always the conversation and the record of it: answering the call, taking details, running the firm's qualification script, booking the consultation, and dropping a lead record into practice management. That is the marketing half of intake and it is worth buying. What they do not cover is clearance, because whether the firm may act is a judgment reserved to lawyers.
Should a law firm outsource legal intake or automate it?
Outsource the answering, automate the clearance. Staffed and AI answering is a competitive, cheap, solved market and building your own rarely repays the cost. Conflicts clearance runs on your own party data, matter history, corporate-tree knowledge and waiver conventions, so no outside service can run it for you and a generic product only partly represents it.
Can an AI receptionist run conflicts checks for a law firm?
It can run the search. It cannot make the call. A well-built intake agent captures every party the caller names, normalizes them, queries your matter records, and surfaces ranked possible hits. Clearing a hit, judging whether a matter is substantially related, and deciding whether a waiver is available belong to a lawyer. The system makes that judgment fast and documented, not automatic.
How long does a manual conflicts check take at a mid-market firm?
Nobody publishes a defensible average, and the per-matter minute figures on vendor blogs do not trace to a real study. Clio's own guidance says a manual check can take hours against minutes or seconds with tooling. The number that matters is elapsed time from the caller hanging up to the matter opening, because that window includes waiting on a partner.
What does AI intake for law firms cost to build?
Fixed fee against a written scope. A focused single-system build, for example a conflicts search and intake record layer over one matter database, runs from $10,000 over 4 to 5 weeks. An end-to-end rebuild is quoted after the $499 audit over 6 to 8 weeks, and a multi-system platform quoted after the $499 audit over 10 to 14 weeks. The answering subscription is separate.
Does an AI answering service create a bar rules problem for the firm?
It creates an exposure to manage deliberately. Under Model Rule 1.18 a person who consults a lawyer about a possible representation is a prospective client, the firm owes confidentiality on what it learns, and information that could be significantly harmful to that person can bar the firm from acting adversely in a substantially related matter. Parties first, facts later.
How do we know the intake and conflicts build actually worked?
Pick the measurement before the build starts and take the baseline first. Three numbers most firms already hold: inbound calls answered live, elapsed hours from first contact to matter number issued, and matters per month needing a billing correction because a party or rate was wrong at setup. If none exist, instrument intake first.
How do you automate a law firm conflicts check?
There is a real product category here and it should be evaluated before anything is commissioned. Legal Technology Hub describes new business intake as the complex set of administrative processes required to onboard a new client or matter, and names Intapp Intake and Clio Grow as two of the leading solutions: one for firms with a dedicated intake function and a risk team, the other for firms whose intake is one person and a spreadsheet.
What is the difference between law firm intake automation and conflicts check automation?
They are two purchases that get marketed as one, and they have opposite answers. Intake automation is answering the call, capturing details, running the qualification script and booking the consultation. That market is competitive, cheap and solved, several vendors publish real prices, and building your own rarely repays the cost. Conflicts check automation runs on your own party data, matter history, corporate-tree knowledge and waiver conventions, which is material no off-the-shelf product holds, so a generic tool can only partly represent it. Buy the first, commission the second. Firms that treat them as a single product usually end up with good answering and a clearance process that still lives in somebody's head.
What happens if the conflicts search does not work as expected?
Every build starts with human review of every search result during a break-in period, so a party-matching miss is caught by the reviewing lawyer rather than shipped silently. If the system does not clear the agreed accuracy bar (tested against the firm's own last twenty declined matters, not a vendor demo) during the prototype stage, we rescope or stop before the fixed fee is invoiced in full.
Who owns the conflicts and intake system at handoff?
The firm does. Source code, prompts, models and the architecture document transfer to the firm at handoff, with no per-seat licence and no vendor relationship the firm depends on afterward. That ownership is what makes the audit log auditable permanently rather than for the length of a subscription, and it is the same handoff standard used on every ColabContent commission.
What is expected of the firm during the build?
The managing partner, operating partner or firm administrator sits in on the $499 AI-Ready Audit call and names the workflow. During the build, the firm provides read access to the relevant matter, client and document data, reviews the ranked hits the prototype returns against real declined matters, and names the lawyer who will keep clearing hits and the internal owner who will keep the system running after handoff.
Does this replace intake staff or the conflicts-clearing lawyer?
No. The system searches, ranks and documents; it never clears a hit. A lawyer still decides whether a matter is substantially related and whether a waiver is available, because Model Rule 1.10(a) makes that a legal judgment no vendor can take. The goal is to remove the two hours of manual searching and the undocumented gap after a call ends, not to remove the people making the call.
Where to look next.
If the answering half is your problem, start with AI receptionist versus a custom AI build for the cost crossover on published per-call pricing. If the clearance half is your problem, read the playbook for whatever you actually run: Clio, Litify, iManage, NetDocuments. Before either, set the baseline: how to measure ROI on a mid-market AI engagement and why rollouts stall in month four.
For firm-level context, the law firm practice page lists the workflows most often commissioned, AI consulting cost for law firms breaks the fee bands down by scope, and the 2027 law firm AI benchmark is the not-yet-fielded study of adoption at firms of 20 to 150 attorneys, with its method published ahead of any number. On headcount, see internal AI hire versus commissioned build.
If a product is already on the shortlist, read the comparison before the demo: Clio Duo, Filevine AI alternatives, Spellbook versus custom, Harvey alternatives. Still choosing who to talk to? How to choose an AI consultant for a law firm is the checklist, and the best AI consultants for mid-market law firms is the ranked comparison.