Matter intake is one of those processes that every firm thinks it has under control — right up until you look at it closely. Incomplete conflict checks, client details entered differently across three systems, fee arrangements that don't match the engagement letter, AML documentation outstanding two weeks after the matter opened. These aren't exceptional failures. They're structural. They happen on every matter, across every practice area, and they compound quietly until a COLP review or a client complaint forces the issue.
We were brought into one situation — a multi-office UK firm, mid-size, predominantly private client and residential conveyancing — where the intake backlog had grown to the point that matters were being opened on the practice management system before AML checks were complete. Not because anyone was cutting corners deliberately, but because the fee earner pressure to open a file and start billing was outrunning the support team's capacity to process documentation. The firm's COLP knew it was a problem. They didn't know how widespread it was until we ran an extract from their case management data and found that, across a six-month period, roughly 340 matters had been opened with at least one mandatory compliance field blank at the point of file opening.
That number — 340 matters — is what focused the conversation. It shifted the discussion from "we should look at this" to "we need to fix this now." And the fix was not, as several people assumed, a better checklist. It was also not, as the IT manager initially proposed, a better intake form.
The Wrong Starting Point: Automating the Form
Most intake automation projects we see proposed — including several sold to firms by legal tech vendors — start by building a better intake form. A client-facing web form, often branded, that collects name, address, matter type, and occasionally uploads an ID document. This is presented as "automating intake." It isn't. It automates data collection at one point in the process. The underlying coordination problem — who checks what, when, and how the system enforces completeness before a matter can proceed — is entirely untouched. That misconception is worth naming directly, because it's where most intake projects stall or fail.
We've seen Clio's intake forms used this way. Clio does provide genuinely useful client-facing intake tooling, and their intake pipeline integrates with their matter management reasonably well. But the limitation is that Clio's intake logic is essentially linear: it collects, it notifies, it creates a draft matter. It does not enforce compliance gates. A matter can progress without AML documentation attached. There's no mechanism within the standard Clio workflow to prevent a fee earner from opening a file until a compliance step is complete — that requires either customisation or a separate tool sitting alongside it. Firms discover this after they've sold the board on "automated intake."
Similarly, Actionstep — which has better native workflow logic than most practice management systems in its price range — allows conditional workflow steps and task dependencies. This is closer to real intake automation. But Actionstep's workflow builder, while powerful, has a meaningful learning curve, and the configuration required to build a genuinely compliant intake gate (one that, for instance, blocks matter opening until an AML check result is recorded and reviewed) typically requires either a dedicated implementation partner or significant internal technical resource. Out of the box, it won't do this for you. We've found that firms underestimate the configuration time by roughly three to one.
At the conveyancing firm, both of these paths had already been explored. Clio had been used for client-facing collection for about eighteen months. The 340 non-compliant matter openings happened on Clio's watch — not because Clio failed, but because no one had built the gate that stopped a matter progressing without the compliance fields populated. The tool was doing what it was designed to do. The problem was that nobody had designed what came next.
What the Gate Actually Needs to Do
The coordination burden in matter intake is real and multi-directional. A new matter in a UK firm requires: conflict of interest checks against existing clients and matters; source of funds and AML verification under the Criminal Finances Act 2017 and the Money Laundering Regulations 2017 (as amended in 2019); a client care letter issued within a defined period; fee arrangement recorded and agreed; and the matter opened on the PMS with the correct supervising partner, matter type, and billing codes. In conveyancing, add protocol documentation and SDLT considerations on top. Some of this falls to the fee earner, some to support staff, some to compliance. The handoffs are where things go wrong.
The value of intake automation — properly built — is gate-keeping, not data collection. The system should prevent a matter from reaching "open" status until a defined set of conditions is satisfied. A sensible baseline for a UK firm includes:
- Conflict search completed and cleared, or escalated with a recorded decision
- Client identity verified: document type, reference, verification method, and date recorded
- Source of funds declaration completed for matters above a defined transaction value threshold
- Engagement letter issued date recorded
- Supervising fee earner confirmed
- Fee arrangement type and estimated value recorded
When this is built properly — using an AI agent or a workflow automation layer sitting between the intake form and the practice management system — the matter simply cannot open without those fields populated. The fee earner sees a dashboard showing what's outstanding. The support team gets task assignments. The COLP gets a daily exception report showing anything that has been open in intake for more than 48 hours.
This is not technically complex. But it requires someone to design the gate logic correctly for your firm's matter types, and to build the integration between your intake tooling and your PMS. That integration is usually where vendor promises meet operational reality. If you're weighing up whether to build this yourself or commission it, the build vs buy question is worth working through carefully before you commit to either path.
Where AI Adds Genuine Value
Beyond workflow gating, there are two places where AI earns its place in intake specifically.
The first is document classification and data extraction. When a client emails in a bundle of documents — passport, utility bill, bank statement, sometimes a trust deed or power of attorney — a trained extraction model can identify document type, pull the relevant fields (name, address, document number, expiry date), and populate the intake record without a human re-keying anything. The accuracy on well-structured identity documents is high: we typically see above 94% on clean scans with a model tuned for UK document formats. The failure cases are low-quality mobile photos and non-standard international documents, which still need human review — but that's a much smaller set than the whole.
The second is conflict checking. Traditional conflict checking is a keyword search against matter and client names. AI-assisted conflict checking can surface semantic similarities — related entities, known aliases, group structures — that a keyword search misses. This matters particularly in M&A and corporate work where the counterparty might be disclosed at the parent level but the matter involves a subsidiary. We've written about the broader picture separately at AML and KYC automation for UK law firms if that's a parallel concern.
At the conveyancing firm, the document extraction piece had the most immediate impact. Identity document processing had been a manual re-keying task. Moving it to an extraction model cut the average time from document receipt to intake record completion from just over two working days to under four hours. That compression alone reduced the pressure on support staff that had been causing the compliance shortcuts in the first place.
The Manual Cost You're Probably Underestimating
One assumption we push back on consistently: that intake is "low-cost admin." The fully-loaded cost of a paralegal or legal secretary role in London is typically between £35,000 and £50,000 per year in salary alone, before NI, pension, management time, and error-correction costs. If that person spends 30–40% of their time on intake processing and chasing — which is not unusual in high-volume conveyancing or private client practices — you're spending £12,000 to £20,000 per year per head on a process that is largely automatable. The real cost of manual data work is consistently higher than firms estimate when they account for the full picture.
Automation projects of the kind described here — intake gating, document extraction, conflict check enhancement — are typically scoped and delivered at fixed price in the £15,000–£35,000 range depending on PMS complexity and matter type breadth. That's usually a payback period of under 18 months, often well under, when the compliance risk reduction is factored in alongside the staffing cost. At the conveyancing firm, the COLP's view was that the reduction in exposed matters alone — from 340 across six months to effectively zero in the following quarter — justified the project independently of the efficiency argument.
Where to Start
The single most useful exercise before any intake automation project is to pull your own data. How many matters opened in the last six months had mandatory compliance fields blank at file-open date? How many engagement letters were issued more than five working days after the matter opened? If your PMS can't answer those questions directly, that tells you something important about where the work needs to start.
Take those numbers to your COLP and your practice manager before you look at any tooling. The business case — and the priority sequencing — almost always becomes clearer once you're looking at your own operational reality rather than a vendor demonstration.
If you'd like us to run that diagnostic extract against your case management data and map it to a specific intake gate design for your matter types, that's a defined piece of work we can scope in a single conversation. It's also how we started at the firm described above — not with a proposal, but with their own numbers.