Plenty of advice firms spend months “looking into automation” without automating anything. They sit through demos, read white papers, talk to vendors and come away more confused than when they started.

The problem is rarely a shortage of options. It’s the order things are done in. Most firms start by asking which tools to buy, when the better question is which problem needs solving and what’s the smallest change that would help this week. Getting that order right separates the firm that’s running noticeably better in twelve months from the one that’s still evaluating.

Start by mapping what your people actually do

Spend a week writing down every repeating task in the firm before you touch a tool or talk to a vendor. Leave out the big strategic processes and look at the small ones: meeting follow-ups, annual review reminders, copying data from one system into another, chasing missing documents, drafting suitability letters.

Ask your advisers and paraplanners to keep a rough log for five working days. You’re looking for tasks that happen repeatedly, follow a predictable pattern, and don’t need professional judgement every single time.

The FCA has acknowledged that infrequent, ad hoc data requests are a real operational burden for firms, because they’re hard to plan and resource for[1]. The same applies inside your own firm: repeating work you can’t plan around is exactly where automation pays.

Most of what you find will fall into three groups: meeting admin, document handling and chasing. Those are the low-effort, high-reward places to start.

Decide whether you need Level 1, Level 2, or Level 3

Most firms overestimate how complex their needs are, so once you have the list, don’t jump straight to building infrastructure. Work through the levels in order.

Level 1 (education)

The fix already exists in a tool you pay for: a prompt template in your AI assistant, a meeting summary feature in your video platform, a scheduled report in your CRM. It’s free or close to it, and available today. A lot of what advice firms assume needs a build is sitting unused in software they already license.

Level 2 (integration)

Two or three tools that need to talk to each other. A new client fills in an onboarding form, the data flows into your back-office system, a task appears in your CRM and a welcome email goes out. That’s configuration work rather than engineering, and platforms such as Make or Power Automate handle it well. It typically takes days to weeks and costs a few hundred to a few thousand pounds.

Level 3 (custom build)

Real engineering: custom orchestration, complex logic, purpose-built pipelines. Four to twelve weeks, carefully scoped, and only worth it when Level 1 or 2 can’t solve the problem.

Always try Level 1 first. Quite often a firm that thinks it needs a custom build needs a prompt and a workflow instead of a development project.

Pick your first automation carefully

Start with the most painful repeating task that carries the least regulatory risk, not the most complex process. It gives you a quick win, builds confidence in the team, and shows you how people react to a changed workflow before the stakes are high.

Meeting notes and follow-up tasks are a strong first candidate for most advice firms. Tools that record, transcribe and summarise adviser meetings give time straight back to client conversations. Intelliflo’s research found that firms automating meeting recording and follow-up let advisers focus on the quality of the conversation, and one firm reported a 20% increase in average clients per adviser as a result. That change needed no custom engineering. The meeting notes to CRM playbook sets out how to build it.

Build your governance framework before you scale

Governance is the step most firms skip, and it’s the one that causes trouble later. It doesn’t need to be a long document, but it does need clear answers to four questions:

  • Who owns each automated workflow? Every automation needs a named person, not a team or a role. That person checks it keeps working and that its outputs are reviewed before they reach a client.
  • What needs human sign-off before it leaves the firm? AI and automation tools can draft suitability letters, produce reports and write client emails. None of those should reach a client until a qualified person has reviewed and approved them. In a regulated firm that’s not optional: the system drafts and the adviser signs off.
  • How will you know if something has gone wrong? Build a simple monitoring step into every workflow. For document generation, that might be spot-checking ten outputs a month. For a data-entry automation, a weekly reconciliation. The FCA’s focus on data integrity and Consumer Duty’s requirement for firms to take responsibility for client outcomes point the same way: you’re responsible for what your systems produce[1].
  • What happens if a tool fails or changes? Vendor instability is a real risk among AI application vendors at the moment. Tools firms have built workflows around can change direction, raise prices or be bought with little warning. Build each automation so the process can go back to a manual fallback within a day, and write that fallback down.

What to do in the first thirty days

This is a practical starting point, not a full project plan.

Map your repeating work

Ask everyone in the firm to log their repeating tasks for one week. Don’t filter or judge at this stage. Collect everything.

Sort by frequency and sensitivity

Start with the tasks that happen most often and carry the least risk. Anything touching advice, suitability or regulated recommendations needs more careful design and shouldn’t be your first automation.

Check your existing tools

Before looking at anything new, go through the tools you already pay for and find out what they can do. Most practice management systems, Intelliflo included, have automation and workflow features nobody uses. Most Microsoft 365 licences include Power Automate. The Level 1 fixes are often already in your stack.

Run a single pilot

Choose one workflow, automate it at the right level and run it for four weeks. Measure what changes: time saved, error rate, what the advisers say. Let that decide what comes next.

Document your governance decisions

Before the pilot goes live, write down who owns it, what the review step is and what the fallback looks like. A page is enough, and doing it from the start makes it a habit.

What to avoid

These four patterns slow firms down or create risk:

  • Automating before you understand the process. If you can’t describe a workflow in plain English, you can’t automate it safely. Automation doesn’t fix a broken process. It runs it faster.
  • Treating AI outputs as final. AI tools can draft suitability letters, meeting summaries and client emails. They can also be wrong in ways plausible enough to pass a quick read. In a regulated firm, AI output needs human review before it’s used in client documents or decisions, every time.
  • Buying a platform and then looking for problems to solve. Find the problem first, then the tool. Doing it the other way round leads to software nobody uses.
  • Scaling before the pilot is stable. Run one automation well before you run ten badly. A pilot that works builds more confidence than a rushed rollout that covers everything.

Keeping it going

Automation in a regulated advice firm isn’t a one-off project. It’s a routine: map the work, find where it sticks, try tools at the right level, review the outputs and update the governance as workflows change. Firms that keep at it aren’t doing it for love of technology. They’re doing it so advisers spend more of their time on the work that needs an adviser. If you’d like the mapping done with you, that’s what an Automation Audit is.

This article is for informational purposes only and does not constitute regulated financial advice or a compliance opinion. Consult a qualified compliance professional for advice specific to your firm.

Sources

[1] FCA, ‘PS26/8: Retail banking business models data’, Financial Conduct Authority, 30 May 2026. Available at: https://www.fca.org.uk/publications/policy-statements/ps26-8-retail-banking-business-models-data