§ Case Studies

Three AI Agents, One IFA Practice, 68% More Client Reviews Completed

— Luke Needham··8 min read
Three AI Agents, One IFA Practice, 68% More Client Reviews Completed

In August 2026, a four-person independent financial adviser practice in Cheltenham was carrying 47 active clients and turning away enquiries because the paraplanning backlog had become unmanageable. The problem was not the advice. The advice took three to four hours per client per year and was where the team's expertise actually lived. The problem was the documentation surrounding that advice. Suitability reports, annual review letters, protection recommendation reports, and FCA Consumer Duty fair value assessments were consuming 30 hours of professional time every week before a single piece of advice had been given. Three AI agents changed the maths. Within ten weeks, the practice was completing reviews for 79 clients a month — 68% more than before — had recovered 25 hours of adviser and paraplanner time weekly, and was running the entire agent stack for £85 per month.

AI agents transforming an independent financial adviser IFA practice UK — three AI agents automate suitability report drafting, annual review letters, and compliance pre-checks to enable 68% more client reviews completed per month

The 30-Hour-a-Week Paraplanning Problem

Paraplanning admin burden at a UK IFA practice — suitability reports taking 4 to 5 hours each, annual review letters 90 minutes, protection recommendation reports 2 to 3 hours, creating 30 hours weekly of document production before any advice is given

The practice — two chartered financial planners, one paraplanner, and a client services manager — handled pension consolidations, investment portfolio management, protection planning, and inheritance tax work for professionals and business owners across Gloucestershire and the Cotswolds. Their average client had been with the firm for six years. The relationship work was excellent. The documentation work was breaking the business.

A suitability report for a pension transfer recommendation took between four and five hours to produce. FCA rules under COBS 9 require suitability reports to explain why a recommendation is suitable for the specific client, addressing their stated objectives, risk profile, financial circumstances, and tax position. Consumer Duty obligations, introduced in 2023 and actively enforced through 2026, added further requirements around demonstrating good outcomes and fair value. Each report had to be individual, evidenced, and defensible.

The paraplanner — who held the Diploma in Regulated Financial Planning and the CII Paraplanning qualification — was skilled and experienced. The problem was not quality. It was volume. A suitability report that should take four hours still took four hours, whether it was the first of the week or the sixth. By Thursday afternoon, the week's capacity was gone. New client enquiries were queuing. Annual review deadlines were slipping. The adviser team was filling weekend hours editing paraplanning drafts.

Annual review letters added another 90 minutes per client — summarising the year's performance, confirming ongoing suitability, and recording the client's updated circumstances and objectives. Protection recommendation reports for new protection cases took two to three hours each. Quarterly compliance reviews of client files, required since the Consumer Duty assessment obligations came into force, added another three to four hours per week across the book.

The arithmetic was brutal. The practice had the advisory expertise to serve 80 clients well. It had the administrative capacity to document 47 of them adequately.

The practice had the advisory expertise to serve 80 clients well. It had the administrative capacity to document 47 of them adequately. That gap — not a skills gap, not a strategy gap — was where the revenue ceiling lived.

The Three-Agent Architecture

Three AI agent architecture for UK IFA practice — Suitability Report Writer agent, Annual Review Letter agent, and Compliance Pre-Check agent working together to automate the paraplanning and documentation workflow end to end

The build took ten weeks. The principal adviser had one non-negotiable requirement from the start: every output had to meet FCA standards without the paraplanner having to rewrite it from scratch. A draft that saved 30 minutes but required 90 minutes of correction was not a saving. The target was first drafts that needed editing, not first drafts that needed replacing. That requirement shaped the entire architecture.

Agent 1: The Suitability Report Writer. This agent produces first drafts of suitability reports for pension transfers, investment recommendations, and protection advice cases. The paraplanner completes a structured intake form — client name, recommendation type, client objectives, risk profile, recommended product or portfolio, key reasons for the recommendation, alternatives considered, costs and charges comparison, and any relevant tax considerations. The agent uses that intake, combined with a knowledge base containing FCA COBS 9 guidance, Consumer Duty outcome documentation requirements, the firm's own suitability methodology, and a library of previously approved reports, to produce a complete first draft in under six minutes.

The drafts are structured reports covering every FCA-required section: client information, objectives and circumstances, attitude to risk, the recommendation and its rationale, the basis for dismissing alternatives, charges and costs, and the ongoing service proposition. The paraplanner reviews, adjusts where the specific client's circumstances require it, and adds any nuance the intake form did not capture. Average review and editing time: 55 minutes, down from the previous 4.5 hours. The saving per report is approximately 80%.

Agent 2: The Annual Review Letter Agent. The annual review letter is the document that confirms ongoing suitability for every client each year — a requirement under FCA rules and a Consumer Duty obligation since July 2023. For a practice with 47 active clients, producing annual review letters is a significant recurring admin burden, with no natural scaling mechanism: each letter must reflect the specific client's portfolio performance, their updated circumstances, and any changes to the advice.

This agent takes three inputs: the previous year's review letter (for context), the notes from the annual review meeting (in any format — typed notes, a bullet list, or a voice-to-text transcript), and the current portfolio summary from the practice's back-office system. It produces a complete, personalised annual review letter that covers performance against benchmarks, any changes to the client's circumstances, ongoing suitability confirmation, and next steps. The paraplanner reviews and approves. Average time: 20 minutes, down from 90. The letters read as though they were written individually for each client — because the inputs are individual. The agent handles the structure and the regulatory language. The paraplanner handles the judgment calls.

Agent 3: The Compliance Pre-Check Agent. FCA Consumer Duty ongoing monitoring obligations require firms to demonstrate that they are actively assessing fair value and good client outcomes across their book. In practice, this means periodically reviewing client files to confirm that the advice on record remains suitable given the client's current circumstances, that the charges being paid are justified by the services received, and that no client has been disadvantaged by an omission in their documentation.

This agent runs a weekly review across the practice's client files. It checks each file against a compliance checklist — suitability report present and dated within the required window, annual review letter issued, ongoing service agreement in place and confirmed, charges disclosed and documented, Consumer Duty outcome assessment recorded. For any file with a gap, it creates a task for the paraplanner or client services manager, categorised by urgency and flagged with the specific compliance requirement being missed.

The agent does not make compliance judgments. It identifies gaps and flags them. The human team resolves them. What previously required the principal adviser to spend Friday afternoons auditing files now runs automatically every Monday morning, with a summary in the practice's project management system before the team arrives.

The Numbers After Ten Weeks

Results from AI implementation at Cheltenham IFA practice — 68 percent more client reviews completed monthly, 25 hours of professional time recovered weekly, £85 per month running cost, suitability report draft time cut from 4.5 hours to 55 minutes

Ten weeks after the agents went live, the principal ran the numbers against the baseline from the previous quarter.

  • Client reviews completed per month: 25 → 42 (68% increase)
  • Professional time recovered per week: 25 hours
  • Monthly running cost: £85 (n8n hosting + Claude API + back-office integrations)
  • Suitability report first-draft time: 4.5 hours → 55 minutes
  • Annual review letter production time: 90 minutes → 20 minutes
  • Compliance file gap rate: 11% of files flagged at first run → 2% after eight weeks
  • New enquiries accepted: 0 per month (at capacity) → 3–4 per month

The 25 recovered hours were redistributed in two directions. The paraplanner used approximately ten hours per week on the backlog of overdue reviews that had accumulated before the build — clearing it within six weeks. The remaining fifteen hours went into onboarding new clients. The practice accepted twelve new clients in the ten weeks following the build, generating substantially more than the £85 monthly running cost within the first billing cycle.

One outcome the team had not expected was the consistency improvement in Consumer Duty documentation. Before the agents were in place, the thoroughness of compliance documentation varied depending on workload. Under pressure, sections were shortened. Cross-references to the Consumer Duty outcome categories were sometimes omitted. After ten weeks with the Compliance Pre-Check Agent flagging gaps weekly, the practice's file quality had measurably improved — not because the human team got better, but because the agent made gaps visible before they became regulatory issues.

Before the agents, Consumer Duty documentation varied depending on workload pressure. The agent made gaps visible before they became regulatory issues — not because the team got better, but because nothing was allowed to slip through unnoticed.

What Made It Work

Three success factors for AI agent implementation at a UK IFA practice — FCA-compliant knowledge base built before agents were deployed, human review at every output stage with adviser sign-off, and sequential build starting with the highest-time-cost task

Three decisions separated this build from the AI experiments that most IFA practices have run and quietly abandoned.

First, the knowledge base was built before the agents were deployed. The paraplanner spent three weeks before the build started assembling a structured library: FCA COBS 9 suitability rules, Consumer Duty outcome documentation guidance, the firm's own suitability assessment methodology, a library of anonymised previously approved suitability reports across the main advice categories, and the firm's house view on risk profiling. Agents without that context produce generic, compliance-light drafts that require complete rewriting. Agents with it produce drafts that reflect the firm's actual standards. The three weeks of preparation work was what made the first real report draft usable rather than a demonstration.

Second, every agent output has a mandatory human review stage. No suitability report leaves the practice without the paraplanner reading it. No annual review letter is sent without adviser approval. The Compliance Pre-Check Agent flags gaps but takes no action — resolution is always a human task. This is not a workaround for AI limitations. It is the correct design for a regulated professional services environment where the adviser carries personal regulatory responsibility for every piece of advice. The human-in-the-loop architecture is what makes the system both FCA-compliant and commercially trustworthy.

Third, the build sequence was disciplined. The Suitability Report Writer was built first and proved in production before the Annual Review Letter Agent was started. The Compliance Pre-Check Agent was the last to be built, once the team had confidence in the first two. The temptation in any professional services AI build is to automate everything simultaneously. That temptation produces pilots that never reach production. Starting with the single highest-cost task — suitability reports — and running it in production for two weeks before building anything else is what avoided the pilot trap that most IFA AI projects fall into.

The practice is now scoping a fourth agent to handle protection renewal reminder letters — a task that currently requires the client services manager to spend several hours each quarter drafting individual renewal reminders for protection policies approaching their annual review. Early prototypes show the same pattern: strong first drafts, 75% time reduction per letter, and no reduction in the personalisation that makes clients feel genuinely looked after rather than processed.

Financial advice is a knowledge-intensive profession built on trust, expertise, and regulatory accountability. None of that changes with AI. What changes is the time cost of the documentation that surrounds that expertise — and for this Cheltenham practice, that time cost has been cut by 80% without touching the quality of the advice itself. If you want to understand whether the same architecture applies to your IFA or financial planning firm, book a free 30-minute call and we will work through it with you.

L

Written by Luke Needham

Founder at Quantum Flow Automation — building AI systems that work.

§ 99Subscribe

More field notes, in your inbox.

One email per week. What we shipped, what broke, what's worth paying attention to in AI.

BOOK CALL