Dacuity — Data & AI advisory for wealth, asset management and fintech

Plenty of people will tell you what AI could do. Fewer will tell you what your organisation is actually ready to do — and what to leave alone.

Wealth, asset management and fintech firms are under real pressure to show progress on AI. That pressure produces a lot of pilots, a lot of vendor decks, and not much that reaches production or survives a risk review.

The bottleneck is rarely the model. It's data that isn't fit for the purpose being claimed, governance that can't evidence a decision, and delivery capability that was sized for something else entirely.

I work on those three things — and I'll tell you when the honest answer is that the case doesn't hold up.

80%+

of AI projects fail to deliver intended business value — around twice the rate of IT projects without AI

RAND Corporation, 2024

60%

of AI projects unsupported by AI-ready data are forecast to be abandoned through 2026

Gartner

42%

of companies abandoned most of their AI initiatives in 2025, up from 17% a year earlier

S&P Global Market Intelligence, 2025

RAND interviewed 65 data scientists and engineers to find out why projects fail. The leading root cause wasn't the technology — it was misunderstanding or miscommunication about what problem was being solved in the first place.

Services

Five ways in, from a conversation to a working solution

Engagements are deliberately sized so you can start small. Most begin at the top of this list and only move down once there's a reason to.

01

Advice

A sounding board for whoever owns data and AI in your business. Bring a decision you're weighing, a strategy you're drafting, or a problem you can't put a name to yet.

Half day — 1 day

02

Second opinions

You have a vendor proposal, an internal business case, or a recommendation from a consultancy, and you want an independent read before you commit budget to it. What's sound, what's optimistic, what's missing.

2 — 5 days

03

Reviews

An assessment of where you actually stand — data estate, AI readiness, governance, delivery capability, or a programme already in flight that isn't delivering what was promised. Findings you can act on, not a maturity score.

2 — 4 weeks

04

Proposals

The strategy, commercial case, target architecture and roadmap — built to survive a board, an investment committee and a risk function. Prioritised by value and feasibility, with the sequencing and ownership spelled out.

3 — 6 weeks

05

MVP and incremental delivery

Building the first working version end to end — data foundations, the model or pipeline, and the governance around it — deployed and in real use. Then extending it in defined increments, each one scoped and priced separately, so you can stop after any of them.

8 — 16 weeks, then by increment

Approach

How this works, and why it's different

Four commitments that shape every engagement.

01

Independent

No vendor partnerships, no reseller margin, no platform of my own to sell you. If the right answer is the tool you already own, or no tool at all, that's the answer you'll get.

02

Practitioner-led

I've built and run these functions, not just advised on them. You get the person who does the work — no bench, no account team, no juniors learning on your budget.

03

Plain answers

A clear recommendation in language your board and your engineers can both act on — including, when it applies, "don't do this yet, and here's what to fix first."

04

Fixed scope

Defined deliverables and a price agreed before we start. No open-ended day-rate drift, and no engagement that quietly becomes a permanent fixture.

Sectors

Built for firms that are regulated, legacy-bound, or both

Deep in three adjacent markets rather than shallow across twenty.

01 — WEALTH

Wealth management

Adviser productivity, client data that spans four systems and three acquisitions, suitability and reporting under scrutiny, and AI that has to be explainable before it can be useful.

02 — ASSET

Asset management

Investment and research workflow, data cost and vendor sprawl, distribution analytics, and the operational reporting burden that grows faster than the AUM behind it.

03 — FINTECH

Fintech

Getting from a working product to a defensible data and AI capability — one that stands up to enterprise procurement, institutional due diligence and an investor asking hard questions.

How it works

What an engagement actually looks like

Three worked examples of the shapes this work usually takes — a second opinion, a review leading to a proposal, and an MVP build with incremental delivery. These are illustrative, not client accounts, so you can see what you'd be buying before you get in touch.

Example — Wealth management

Second opinion

Typically 2 — 5 days

You're weeks from signing a platform contract and want an independent read

Typical trigger: A vendor proposal and an internal business case, both built substantially on the vendor's own benefit projections. Nobody in the room is neutral, and the number is large enough that being wrong is expensive.

What I'd do: Read the proposal against your actual data estate rather than the idealised one. Test whether each claimed use case is supported by data you hold, in the state you hold it. Pressure-test the benefit assumptions and the implementation timeline.

What you'd end up with: A short written view: which use cases hold up, which need upstream work first, what the proposal doesn't mention, and the questions to put to the vendor before signing.

Where it usually goes next: Either you sign with better terms and a rescoped phase one, or you defer and fix the data foundations first. Sometimes nothing — which is a valid outcome.

Example — Asset management

Review, then proposal

Typically 5 — 8 weeks

You've run pilots for eighteen months and nothing has reached production

Typical trigger: A portfolio of proofs of concept across research, distribution or operations. Individually promising, collectively going nowhere, and the board has started asking what the spend has bought.

What I'd do: Assess the portfolio against value, feasibility and the governance bar each would have to clear. Establish why things stall here specifically — usually data, ownership or risk appetite rather than technology.

What you'd end up with: A prioritised shortlist with the rest explicitly stopped, a commercial case per initiative, and a roadmap with sequencing, ownership and governance defined — written to survive an investment committee and a risk review.

Where it usually goes next: Your team delivers it, or I stay on in a retained capacity to lead the first release through to production.

Example — Fintech

MVP and
incremental delivery

8 — 16 weeks,
then by increment

The use case is agreed and now someone has to actually build it

Typical trigger: There's a decision to proceed and a rough design, but hiring a team for something still unproven is the wrong first move — and the internal engineers who could build it are committed to the product roadmap.

What I'd do: Build the narrowest version that settles the question in production conditions: the data pipeline it depends on, the model or service itself, and the governance evidence it needs to stand up to a due diligence question.

What you'd end up with: Something running with real data and real users, the evidence to decide whether to scale it, and documentation and handover so your own team can own it rather than inheriting a black box.

Where it usually goes next: You scale it in-house, or we extend it in increments — each scoped and priced on its own, so you can stop after any of them. Or you stop now, with a small bill and a clear answer instead of a large one and a maybe.

Practice: DACUITY LTD
Company number: 17431332
Led by: Aleš Hejmalíček — former Head of Data & AI
Focus: Data strategy, AI strategy and governance, technology architecture, delivery
Based: Edinburgh — working with firms across the UK and Europe

Who you're actually hiring

I've spent my career as Head of Data and AI — owning strategy, architecture, development and governance, and being accountable for whether the thing worked, not just whether it was recommended.

That's the perspective I bring: I've sat on your side of the table, made these calls with real budget and real regulatory exposure behind them, and lived with the consequences afterwards. I know which vendor claims survive contact with a legacy data estate and which don't.

The most useful thing I bring is knowing where the line falls between a problem technology can solve and one it can't. A lot of failed data and AI work is a technical answer aimed at something that was really about ownership, incentives, process or headcount — no platform fixes that, and buying one usually makes it harder to see. Telling those two apart early is where most of the value is, and it's the part a vendor is structurally unable to do for you.

The name is data and acuity. Acuity is the sharpness of vision — the ability to resolve detail that's genuinely there but that a blunter instrument would miss. That's the job: seeing your data and AI position clearly enough to know what's real, what's noise, and what's worth acting on.

Get in touch

Get in touch for a 30-minute conversation

No pitch. Bring the decision you're weighing or the problem you're stuck on, and you'll get a straight view on whether it's worth doing, and whether I'm the right person to help. If it isn't and I'm not, I'll say so.

Prefer email? hello@dacuity.io
Or connect on LinkedIn.

Take the maturity assessment