James ran a four-person IFA practice in Edinburgh. Four good advisers, 60 clients, and a filing system that had evolved over twelve years into something he described as "organised but slow." Every month the same bottleneck: suitability letters took four hours each, annual review packs needed half a day to assemble, and meeting notes were either rushed or happened at 9pm. The practice was well-regarded and growing — but every new client added to the operational weight that was already maxing out the team. Here's how three AI agents changed those numbers, and what it took to get there.
The Starting Point: Four Good Advisers Buried in Paper
Cairngorm Financial Planning is a four-adviser independent practice based in Edinburgh, working primarily with professionals and business owners in the 40–65 age bracket. Pension planning, investment management, protection, and estate planning. Sixty active clients, a solid reputation, and a steady flow of referrals.
James Crawford — director and the firm's longest-serving adviser — had one consistent complaint: the ratio of advice work to admin work was wrong. His team were good at what they did, building financial plans, managing client relationships, navigating complex pension situations. The operational drag around that work was eating it alive.
The breakdown across the team's working week:
- Suitability letters — 3–4 hours per letter, 10–12 per month for the practice. At peak review season, one adviser could spend an entire week writing letters that documented decisions already made in the client meeting
- Annual review packs — compiling fund performance data, portfolio valuations, cash flow model updates, and agenda documents for each review. Roughly 3.5 hours per client per year, across 60 clients
- Meeting notes and action logs — post-meeting write-ups, action tracking, and CRM updates. Averaging 45 minutes per meeting, across 12–15 meetings per adviser per week
- New client onboarding — collecting fact-find information, verifying documents, setting up client portals, running AML checks, and creating the initial financial plan record. 10–14 hours per new client, spread across two or three staff members
James estimated roughly 40% of each adviser's working week was going to documentation and process work. Work that wasn't advice. Work that none of them had imagined spending their careers doing when they qualified.
The FCA's Consumer Duty requirements added another layer. The firm needed to evidence ongoing service quality, demonstrate fair value, and produce retrievable records showing advice was in each client's best interest. All legitimate requirements — but each one added to the documentation load rather than reducing it.
"We were compliant. We were profitable. But 40% of four advisers' time on paperwork means 40% less capacity for clients. When I worked out what that cost us in advisory fees not earned, it wasn't comfortable reading."
— James Crawford, Director, Cairngorm Financial Planning
What the Audit Found
James booked a £500 AI Audit. We spent a day mapping every recurring process: time cost, decision complexity, data structure, and automation suitability. The output is a prioritised matrix — what to automate, what to leave alone, and in what order.
Four processes scored above 4.0 on our five-point suitability scale:
- Suitability letter generation — structured meeting notes in, structured letter out. The regulatory language, risk profile alignment, and rationale sections are heavily templated. The adviser's judgment sits in the meeting; the letter documents it. Automation score: 4.6/5
- Annual review pack assembly — data in from portfolio platforms, performance benchmarks, and cash flow models; branded review document out. Automation score: 4.8/5
- Meeting notes and CRM updates — structured notes and action items from meeting records, synced to back-office system. Automation score: 4.4/5
- New client onboarding — templated document request sequence with AML verification triggers and CRM record creation. Automation score: 4.3/5
One process we recommended leaving entirely alone in phase one: financial planning itself. The cash flow modelling, tax planning, investment selection, and retirement income structuring that define the firm's value — none of that is in scope. Automating the documentation of those decisions is what creates capacity to make more of them.
Projected weekly hours recoverable across the four priority processes: 32–38 hours across the team. At a blended adviser cost of £35 per hour, that is £1,120–£1,330 of recovered professional capacity per week.
Building the AI Operating System
Same infrastructure as every deployment: OpenClaw running on Google Cloud Run, connected to the firm's back-office system (Intelliflo Office), Microsoft 365, and the portfolio platform (Transact). Monthly infrastructure cost: £47.
Agent 1: Suitability Letter Generation
The meeting preparation workflow at Cairngorm involves pre-completing a structured meeting agenda in Intelliflo with the client's current position, recommended changes, and risk profile. After each advice meeting, the adviser updates this record with what was discussed, what was recommended, and what the client agreed to.
This agent connects to that record. When the adviser marks a meeting as "advice given" in Intelliflo, the agent generates a full suitability letter — the client circumstances and objectives section, the recommendation rationale, the risk warning disclosures appropriate to the recommendation type, the cost illustration, and the next steps — and posts the draft back to the client record for the adviser to review.
Average letter generation time: under 4 minutes. Average adviser review and edit time: 20–25 minutes, down from 3–4 hours. Time saving per letter: approximately 3 hours. Across the firm's 10–12 monthly letters, that recovers 30–36 hours per month from the most time-intensive document in financial advice.
Agent 2: Annual Review Pack Assembly
Four weeks before each scheduled annual review, this agent triggers automatically. It pulls the client's current portfolio valuation from Transact, retrieves previous year benchmark performance data, updates the cash flow model with current values, and compiles a branded review pack — agenda, portfolio summary, performance commentary, and discussion prompts for the adviser to customise.
What previously took 3.5 hours of adviser or paraplanner time now takes 8 minutes of automated preparation and 30–40 minutes of adviser review and personalisation. The reviews are also better: the adviser arrives more prepared, the client receives a more structured pack, and Consumer Duty evidencing is built into every document automatically.
Agent 3: Client Onboarding and Fact-Find Processing
New client onboarding at an IFA practice is document-intensive. Anti-money laundering checks, fact-find completion, identity verification, portal setup, initial data entry into the back-office system, and the first preliminary assessment all have to happen before any advice work can begin. Previously this required 10–14 hours spread across two or three staff members over two to three weeks.
The onboarding agent triggers when a new client engagement is confirmed. It sends a structured welcome sequence — one clear step at a time, with automatic 48-hour reminders for outstanding items — and routes each completed document to the correct location in Intelliflo and the firm's SharePoint. AML verification is handled via integrated identity check APIs. Once all required items are received and verified, the agent creates the client record pre-populated with the fact-find data and sends a "ready to advise" notification to the lead adviser.
Average onboarding time: reduced from three weeks to eight days. Admin hours per new client: from 10–14 hours across the team to under two hours of oversight.
The Numbers at 90 Days
At the 90-day mark, every metric was measured against the pre-build baseline:
| Metric | Before | After 90 Days |
|---|---|---|
| Suitability letter (time each) | 3–4 hrs | 20–25 mins |
| Annual review pack (time each) | 3.5 hrs | 30–40 mins |
| New client onboarding (admin hours) | 10–14 hrs | Under 2 hrs |
| Active client accounts | 60 | 95 |
| Monthly infrastructure cost | £0 | £47 |
| Weekly team hours recovered | 0 | 34+ |
Active client accounts grew from 60 to 95 — a 58% increase — in 90 days, without adding headcount. The 34 recovered hours per week went directly into new client work and deepening the service for existing ones: the only two activities that grow an advisory practice.
Total build cost — audit, build, and 90 days of operation — came to £3,600. At approximately £1,190 per week of recovered capacity at adviser rates, the system paid for itself in three weeks.
"The suitability letter agent alone justified the entire build cost inside a fortnight. The rest was upside. We now have capacity to do the work we were turning away — not because we weren't good enough, but because we didn't have the hours."
— James Crawford, Cairngorm Financial Planning
The Compliance Dividend
The time savings were expected. What James didn't anticipate was the compliance benefit.
Under the FCA's Consumer Duty, firms must demonstrate ongoing service value, evidence that advice was suitable at the time it was given, and maintain retrievable records of their decision-making process. The compliance overhead of doing this well — across 60+ clients, every year — had been growing steadily since Consumer Duty came into force in 2023.
When suitability letters are generated by an AI agent from structured meeting data, every letter follows the same format, contains the same mandatory sections, and is filed automatically against the correct client record. The evidencing Consumer Duty requires is built into the process rather than bolted on afterwards. The same applies to annual review packs: the agent's output creates a consistent, dated, structured record of every review, automatically.
At the firm's next compliance review, this consistency was noted positively. The reviewer commented that the standardisation of documentation was above the typical level for a firm of this size. James's response: it costs £47 a month to achieve it.
This is the pattern we described in our piece on AI regulation in the UK: the compliance overhead that was previously a cost of serving clients has, with the right architecture, become a by-product of serving them well. The admin doesn't disappear — it just happens automatically.
What This Means for UK Financial Planning Firms
The IFA sector faces a specific structural problem. Consumer Duty compliance, ongoing service requirements, and the documentation demands of regulated advice mean the non-advice workload per client has been growing year on year. The logical response — charge more, serve fewer clients, or hire more staff — has limits that many practices are already hitting.
An AI operating system changes the denominator. The administrative work that currently scales linearly with client numbers stops doing so. One agent handling suitability letters costs the same whether the firm writes 10 letters a month or 40. One agent assembling review packs runs the same process for 60 clients or 120. The fixed cost of the infrastructure — typically £40–60 per month — does not move when you take on new clients. Your headcount bill does.
The pattern at Cairngorm mirrors what we saw with the Manchester recruitment agency and the Yorkshire accountancy firm: a professional services firm with good people, real clients, and a 35–40% operational overhead that was capping growth. The same 35–40% that an AI operating system absorbs — reliably, consistently, without burning out or taking annual leave.
If you run an IFA practice, financial planning firm, or wealth management business and the bottleneck isn't advice quality but adviser capacity, the conversation worth having is the same one James had. Get in touch — we scope it in a single 90-minute session and tell you exactly where the hours are going.