Fracas Digital • Jul 17, 2026 • 9 min read
AI Client Intake for Law Firms: What to Automate, What to Check with the SRA
Most of the new-enquiry work at a law firm can now run through an AI agent without a fee earner on the line. The screening call, the conflict check, the engagement letter, the terms of business. The SRA's compliance guidance, updated 9 February 2026, permits it. Clients must be told they are speaking with AI. A named human must review the output before it reaches them. The firm's COLP owns the decision to introduce the technology.
That is the full compliance picture. What the guidance does not do is restrict what the intake workflow can look like, only who is accountable for it.
Almost all the content you will find on this subject is written by US SaaS vendors with no reference to SRA obligations. This is the UK version.
What does manual client intake actually cost a law firm?
A first call with a new enquirer takes 15 to 30 minutes for a paralegal who is experienced enough to ask the right questions. Add the follow-up email, the conflicts check in the practice management system, the engagement letter, and the terms of business, and a single new matter can absorb 90 minutes of fee earner time before a single billable second accumulates.
Personal injury, residential conveyancing, and employment firms handle 50 to 150 new enquiries a month. The maths is not kind.
The other cost is response time. Enquiries that arrive on a Friday afternoon and do not get called back until Monday morning convert at a fraction of the rate of same-day responses. Clio's 2026 UK Legal Insights Report found that 89% of legal professionals now use AI in some capacity, and response speed is one of the primary drivers. Firms using AI for after-hours intake capture enquiries that previously went to competitors with shorter callback windows. Staffing for a round-the-clock response window is not practical. An AI agent is.
What can an AI intake agent handle end to end?
The answer differs between what US SaaS vendors describe and what makes sense for a UK practice operating under SRA obligations.
Five workflows come up in every conversation we have with law firms considering this.
Initial qualification. The agent asks what kind of matter the caller has, which jurisdiction it falls under, and whether they have spoken with anyone else. It captures contact details and the key facts needed to identify the correct practice area. Conversational intake, compared with a web form, typically pushes lead-to-consultation conversion from 8-12% to 20-30%. The caller feels heard rather than processed.
Conflict screening. The agent searches your existing client and matter database for the names involved and flags any match for human review. It cannot make the conflict determination; a fee earner has to. But the check happens before the consultation is booked, not after.
Engagement letter draft. For matters that clear the conflict screen and fall within your accepted work types, the agent drafts the engagement letter from a template, pre-filled with the matter details captured during the call. A fee earner reviews and sends. Writing the letter from scratch is rarely what the fee earner should be doing with that time.
Enquiry routing. Based on matter type and complexity, the agent routes the qualified enquiry to the right fee earner or team, with a summary of the call attached.
After-hours follow-up. The agent holds the enquiry, confirms next steps, and flags it for first-thing call-back. No enquiry falls through the gap between Friday afternoon and Monday morning.
The agent does not give legal advice, make risk assessments, or put anything in front of a client without a human review step. That line is non-negotiable under current SRA guidance, and any system that blurs it is not built correctly.
What the SRA actually says about AI in client intake
The SRA is not hostile to AI in client intake. Garfield.Law, the first fully AI-driven law firm, received SRA authorisation in May 2025. The regulator's position is that existing rules apply to AI and are sufficient as they stand.
The SRA's compliance guidance updated February 2026 sets out four things the firm must do. Clients must be told they are interfacing with AI. The COLP is responsible for compliance when new technology is introduced. Confidentiality and security obligations apply to AI tools the same way they apply to any other system. And the firm's AI policy must cover the approved tools, the verification obligations, and the record-keeping requirements.
The Law Society's generative AI guidance takes the same position: a solicitor's professional obligations apply to AI-assisted output exactly as they apply to any other output, and the supervising solicitor is responsible for verifying what the system produces before it reaches a client.
There is also the post-Mazur supervision guidance, issued in June 2026, which requires human review and professional judgement before AI-assisted output reaches a client. For intake, this means the engagement letter and matter summary need a fee earner review. Most well-designed intake agents build this gate in by default.
One point worth clearing up: in June 2025, the High Court in Ayinde v Haringey LBC [2025] EWHC 1383 referred both legal teams to their regulators after AI-generated submissions included fabricated case citations. That ruling is a warning about AI used for legal research and advocacy. It does not apply to intake qualification, which involves structured question-and-answer flows, not legal reasoning. The two workflows are different and should not be conflated.
AML, conflict checks, and source-of-funds packs
Solicitors have anti-money laundering obligations that do not change because the first contact is an AI agent. Under the Money Laundering Regulations 2017 and the SRA's AML guidance, client due diligence is required before a firm acts for a new client.
An intake agent handles the administrative side. It requests the required identity documents, chases missing items, logs what has been received, and assembles the source-of-funds pack for review. The actual risk assessment, whether to proceed with instruction, is a decision for a qualified person.
Firms doing high volumes of property or corporate transactions find this useful for a specific reason. The agent requests the same documents from every new client, chases non-responders on a schedule, and presents the COLP with a complete pack rather than a list of gaps. The COLP still makes the call, but their time goes on judgement rather than administration.
SaaS intake tools vs a custom intake agent: the real difference
Clio Grow, Lawmatics, and MyCase are the tools recommended by most US content on this topic. They are intake workflow software with configurable forms, automated follow-up sequences, and basic integration with their respective case management systems.
They are useful. They are also built for the general market and do not know your firm, your practice areas, your conflict-check logic, or your engagement letter templates. Configuration takes you some of the way; custom logic takes you the rest.
A custom intake agent built on your actual workflows does things the SaaS platforms do not. It checks conflicts against your specific case management system, not a separate database. It handles practice-area-specific routing. Property screening asks different questions than personal injury or employment matters, and a generic questionnaire will miss things a fee earner has to chase up later. Engagement letters come out in your house style, from your own precedents. And it integrates with your existing GDPR compliance setup rather than adding a new data processor to your ICO register.
We see a similar pattern in UK accounting practices, where off-the-shelf tools handle common workflows and a custom build takes over where the firm's processes diverge from the generic product. It is the same reason our AI agent services almost always start with a process audit rather than a tool selection.
How to implement AI client intake in a UK law firm
Three things need to happen before a system goes live.
A DPIA. Processing personal data from enquiries at scale, using automated triage, almost always meets the ICO threshold for a Data Protection Impact Assessment. The COLP is exposed if the firm skips this step. A DPIA does not take long to prepare, but it has to happen before the system touches live data.
A scoped intake policy. The SRA expects the firm to have documented what AI is doing in the workflow and what the human review gates are. A one-page policy that names the tool, the workflows it handles, and the fee earner responsible for reviewing output is enough to show proportionate governance.
Calibration on your matter types. A generic intake agent asks generic questions. Map the specific qualifying questions for each practice area before build. Personal injury screening is different from commercial property due diligence, and the agent needs to know which it is dealing with from the first message.
Our AI automation service for law firms covers the full build. Intake agent, AML pack assembly, conflict screening, engagement letter generation, and a DPIA prepared alongside rather than left to the firm after the fact. If you want to see what this looks like for your practice, we are talking to firms this month at /book-a-call.
One thing to do this week: audit your current new-enquiry handling. Track how many enquiries arrive outside office hours over the next 30 days, how long the average first response takes, and where in the intake process qualified staff time is being spent. That audit is the starting point for every specification we write, and it will tell you exactly where an intake agent pays for itself.