AI for small UK IFA firms and wealth managers: what actually works in 2026

It's the last Friday of the month in a six-adviser IFA firm somewhere in Birmingham or Bristol. The senior paraplanner is four hours deep into preparing next week's annual review packs. She has eight reviews to run by Wednesday. Each pack means pulling platform statements from Aviva, Quilter, and Transact, dropping numbers into the house template, writing the covering letter, chasing the adviser for the commentary paragraph, and queuing for compliance review.

The back-office has a new "AI summary" button. She clicked it twice last month. It summarised the client's activity nicely. It did not pull the Aviva statement, did not populate the template, did not draft the letter, and did not help with the one client whose income on Transact didn't reconcile to the statement on record.

This is the firm I build for. Two to twenty advisers, UK-based, running on Intelliflo or Iress Xplan or True Potential or Nucleus Adviser, losing paraplanner and adviser hours every week to admin the software was supposed to eat. This is what I'd want a principal of one of those firms to read before buying another "AI for IFAs" pitch.

The short answer

AI is genuinely useful in UK IFA firms in 2026, but most of the value is in narrow workflows the big vendors aren't selling: suitability report drafting, annual review pack preparation, platform reconciliation, fact-find summarisation, call note capture. The work sits outside the back-office, not inside it. For a small UK firm (two to twenty advisers), expect to save 4-10 hours per adviser per week once built properly, and to pay between £10-£100 per user per month for bundled or specialist SaaS AI or £12,500 upwards for a custom automation layer that targets one or two high-volume workflows end to end.

Why AI is stalled in most small UK IFA firms

The UK IFA software market has been selling "adviser efficiency" since the early 2000s. Back-office systems, planning tools, platform wraps, document portals. Most of it is genuinely useful. None of it solved the actual problem, which is that a six-adviser firm in Bristol still spends most of the month on the same admin loops it ran in 2018. Three reasons.

The Consumer Duty framework forces careful adoption. Since 2023, advisers have had to evidence good outcomes for retail clients across four dimensions, with the bar noticeably raised on supervision, file notes, and adviser competence. Any AI workflow that touches client-facing output has to survive a Consumer Duty audit. The default response from most principals is to do very little, very carefully, and wait for the FCA to speak more clearly on gen-AI use.

The back-office and platform vendors are drip-feeding AI features. Intelliflo, Iress Xplan, True Potential, Nucleus Adviser, Curo. Each is rolling out AI features (drafting client communications, summarising fact-find data, suggesting next actions). The features are real but deliberately narrow. The vendor is not going to build the workflow that pulls statements from Aviva plus Quilter plus Transact into the house review pack template, or the workflow that reconciles unusual platform income flows, or the workflow that automates the Letter of Authority chase across three providers. That is the firm's problem, and most firms do not have anyone who can solve it.

The Letter of Authority and platform reconciliation chain is fragmented. UK platforms each have their own adviser portal, their own reporting format, and their own integration surface. Some have public APIs and most do not. An LoA chase involves chasing the platform, the provider, and occasionally the legacy policy administrator. Paraplanners spend significant time on this every month. None of the bundled AI features touch it.

The firms that get this right stop waiting for the back-office roadmap and build a narrow automation layer alongside their existing stack, with Consumer Duty constraints and file-note discipline baked into the architecture from day one.

What's actually worth automating in a 2-20 adviser UK IFA firm

Every adviser publication will tell you to automate everything. That's vendor copy. In practice, some automations compound and some waste build hours. Here's what I've seen actually work in small UK IFA firms, in rough order of ROI per workflow.

1. Client onboarding, fact-find, and AML

A new client agrees to engage. The current process: adviser sends a fact-find form, client completes and emails back, paraplanner types into the back-office, AML provider run separately, Letter of Authority sent by post or email, risk questionnaire completed separately. 90-180 minutes of admin across two weeks of chasing.

A workflow that collapses this into a structured digital fact-find, AML submission to your provider's API (Smartsearch, Veriphy, Credas), LoA document generation and DocuSign, risk profiling questionnaire captured digitally with Oxford Risk or equivalent output, back-office record created automatically on completion. Human review on the AML risk decision and the final suitability consent only.

Time saved: 60-120 minutes per new client onboarding. Multiply by 30-60 new clients a small firm onboards per year.

Build complexity: medium-high. AML provider integration is the hardest. Back-office API integration is the second-hardest. The digital fact-find and document generation are easier.

2. Suitability report drafting from fact-find data

Every new investment recommendation needs a suitability report (SR). Advisers currently draft the rationale, paraplanners format, compliance reviews, adviser signs off. A good SR runs 3,000-6,000 words and takes 60-120 minutes of adviser plus paraplanner time to produce.

A workflow that takes the fact-find structured data, the recommended model portfolio or fund selection, and firm-specific SR template, and produces a draft that the adviser reviews and edits rather than writes from scratch.

Time saved: 30-60 minutes per SR. A typical adviser writes 4-10 SRs a month.

Build complexity: medium. The LLM drafting is well-understood in 2026. The interesting work is in template structure, compliance red-flagging, and ensuring citations (to the fact-find, to the research basis) are correct.

3. Annual review pack preparation

Every client gets an annual review. Paraplanners pull platform statements, populate the house template, write the performance commentary, queue for adviser review. 60-120 minutes per review. Larger firms have dedicated annual review teams.

A workflow that pulls statements via platform API or scraped PDF, aggregates performance, drafts the commentary paragraph from the year's activity and market context, produces a review-ready pack.

Time saved: 45-75 minutes per review. Multiply by the 200-600 annual reviews a small firm runs per year.

Build complexity: medium-high. The pull from platforms is the variable part: Aviva and Quilter have adviser APIs; some smaller platforms need PDF scraping or CSV exports. Template logic is straightforward once inputs are structured.

4. Platform reconciliation and income tracking

Adviser charges and platform fees flow from providers to the back-office via statements and reports. Paraplanners or operations manually reconcile, flag exceptions, post to the income ledger. Chronic admin for firms handling 200+ clients across multiple platforms.

A workflow that reads platform income reports, matches to back-office records, flags mismatches, produces a weekly exception report for the operations manager.

Time saved: 3-6 hours per week of operations time.

Build complexity: medium. Platform integration is the work. The matching logic is simple once data is structured.

5. Call and meeting note capture

Advisers finish a client meeting and spend 10-20 minutes writing a file note. A workflow that takes the call recording (voice memo, Otter, Fathom, or a softphone integration), transcribes, summarises to the firm's house file-note format, writes back to the back-office record.

Time saved: 8-15 minutes per meeting. An active adviser has 4-8 client meetings per week.

Build complexity: low to medium. Transcription is commoditised. The summarisation style and the back-office write-back are the work.

What I wouldn't automate first

Investment advice outputs. LLMs are improving but still confidently wrong on UK product specifics, wrapper mechanics, and pension rules often enough that any client-facing advice content needs senior adviser review. The review wipes the time savings.

AML risk scoring. The FCA and supervisory bodies expect a qualified person making the risk decision. Automate the document collection and the chase; do not automate the risk decision.

Portfolio construction decisions. Regulated advice. Use AI to prepare comparisons and summaries, not to choose the portfolio.

Client-facing email without a human gate. The tone and content risk is higher than the time savings, and Consumer Duty audits look at this closely.

The typical stack in 2026

A working AI and automation stack in a small UK IFA firm has five layers.

Layer 1: the back-office. Intelliflo, Iress Xplan, True Potential, Nucleus Adviser, Curo, or similar. System of record for clients, plans, platforms, fees. Intelliflo has the most usable REST API at the SME end. Iress and True Potential are more variable in integration surface. Curo and newer entrants have modern APIs.

Layer 2: the planning tools. Voyant, Cashcalc, FE Analytics, Dynamic Planner, Oxford Risk. Each has its own integration surface. Voyant has a usable public API; others are more closed.

Layer 3: the platforms. Aviva, Quilter, Transact, Nucleus, M&G Wealth, Fidelity Adviser, Embark, Abrdn. Integration quality varies widely; Aviva and Quilter have adviser APIs, some smaller platforms still rely on CSV exports or PDF statements.

Layer 4: the data and compliance services. Companies House, HMRC APIs, your AML provider (Smartsearch, Veriphy, Credas, ID-Pal), Oxford Risk for questionnaires, DocuSign for client signatures.

Layer 5: the LLM decision layer. Claude (Anthropic) or GPT (OpenAI) called over API from the orchestration layer. One or two LLM calls per workflow is typical. Use API tier, not consumer apps. Confirm data is not used for training (it isn't, by default, on Anthropic and OpenAI API tiers).

The interesting question is what you use to wire layers 1-5 together. This is the orchestration layer, and it's where agency archetype matters.

Who builds this for you (and who pretends to)

I wrote a longer version of this framing in my AI automation agency buyer's guide. The short version, applied to IFA firms specifically:

• The Zapier or n8n reseller will quote £800-£5,000 per workflow built in their Zapier account. Usable for the simplest jobs (intake form to back-office, missing-document email reminders) but breaks on anything involving platform API integration, AML provider integration, or Consumer Duty audit-trail requirements.
• The AI-labelled consultancy will sell you a £12,000-£25,000 "AI for advice" strategy engagement. The deck cites FCA guidance and Consumer Duty expectations. The implementation is either out of scope or subcontracted.
• The specialist IFA AI vendor (AdvisoryAI, PlannerPal, Aveni, Nexus AI) sells pre-built AI tooling targeted at advisers. Often excellent for the specific workflow they target (Aveni for file notes, PlannerPal for workflow, AdvisoryAI for suitability drafting). Less useful for workflows they don't cover.
• The engineer-led custom shop will quote £12,500-£30,000 to build a narrow automation layer that reads and writes to your back-office and platforms over their APIs, runs on your own AWS or Render account, and is documented well enough that a future engineer can maintain it.

Laires Labs sits in the engineer-led category. The build floor is £12,500 because smaller builds don't cover scoping, FCA-aware design, build, handoff, and the post-launch support a real IFA automation needs. Smaller jobs are better served by the reseller tier or by a specialist IFA AI vendor.

Real costs: SaaS stack vs custom build over 2 years

Same 6-adviser UK IFA firm, same desired automation (three of the workflows above: onboarding and AML, suitability report drafting, annual review preparation).

SaaS-and-specialist-vendor path

• Intelliflo Office (6 advisers): £95/seat/month × 6 × 24 = £13,680
• AdvisoryAI suitability drafting: £80/seat/month × 6 × 24 = £11,520
• Aveni for file notes: £50/seat/month × 6 × 24 = £7,200
• AML provider (Smartsearch): £100/month × 24 = £2,400
• DocuSign for client signatures: £40/month × 24 = £960
• Annual review spreadsheet templates and misc tooling: £2,000 across 2 years

2-year total: £37,760. The workflows the specialist vendors cover are well-served. Platform reconciliation and ad-hoc cross-system work remain manual.

Engineer-led custom build path

• Intelliflo Office (6 advisers): £95/seat/month × 6 × 24 = £13,680 (same)
• AML provider (API direct to Smartsearch): £100/month × 24 = £2,400 (same)
• DocuSign: £40/month × 24 = £960 (same)
• Render or AWS hosting for the automation layer: £40/month × 24 = £960
• LLM API costs: £200/month × 24 = £4,800
• Custom build (three workflows): £22,000 one-off
• Retained maintenance (ad-hoc): £3,000-£6,000 across 2 years

2-year total: £47,800-£50,800. Costs £10,000-£13,000 more over 2 years than the SaaS-and-vendor path. In return, the workflows are integrated with the back-office and platforms (not just running adjacent to them), and the firm owns the code.

The honest argument for SaaS-plus-vendor at this firm size: the specialist vendors (AdvisoryAI, Aveni) are genuinely good at what they do, and replicating that specific capability in custom code costs more than the subscription. The honest argument for custom: at 12+ advisers the vendor seat fees compound, the workflows that need to be truly integrated with your specific stack multiply, and the value of ownership grows. The switch typically happens around 10-15 advisers.

FCA, Consumer Duty, and the compliance questions you'll actually face

Every automation in a UK IFA firm has to survive four questions. A good build answers all four in writing before the principal signs off.

Where does the data live and who can see it? Client-confidential information must not land on a US consumer AI service. On the API tier, Anthropic and OpenAI default to not training on your data, but the contract details differ. Your data protection record needs to reflect the actual data flow: where the prompt goes, where the response is stored, retention, deletion. UK or EU regions where possible.

Who is accountable for the output? A qualified adviser signs off anything reaching a client or the FCA. An automation can draft, summarise, prepare, and chase, but it cannot send a suitability report or an annual review pack to a client unsupervised. Build the human review gate in.

How do you evidence Consumer Duty outcomes? The FCA expects advisers to evidence that their process delivers good outcomes for retail clients. A well-designed automation improves this (consistent documentation, faster turnaround, reduced error rates) but needs to record how AI contributed to the output in the matter file. The automation log is Consumer Duty evidence if designed as such.

What would your PI insurer ask? At 2026 renewal many PI insurers add AI-specific questions: are you using generative AI on regulated advice, which workflows, what supervision, what data handling. A firm that can answer those questions clearly from day one tends to get better renewal terms than one that cannot.

Book a call

If you're running a 2-20 adviser UK IFA firm, bleeding paraplanner hours on the admin side of regulated advice, and trying to work out whether AI is a real lever or another subscription, that's the call to book. Twenty minutes, no pitch. Tell me your back-office, adviser headcount, and the two workflows your team complains about most. I'll come back with a scoped build plan: what I'd build, how I'd phase it, what it would cost, and which of the workflows above I'd park for your specific firm. Yours to keep whether you hire me or not.

Don't book if you're a sole adviser, if you're mid-networks-transition, or if your compliance stance is "wait for clearer FCA guidance." The cultural gap is the biggest blocker and the one I can't fix.

Book a 20-minute call or read the full AI automation agency buyer's guide if you're still deciding between archetypes.