Process intelligence for regulated financial services

From process map
to proven AI value.

PinpointProof maps how work really happens in regulated financial firms, scores every step for AI and automation potential, and proves the value before anyone builds anything.

Thirty minutes, straight into the diary. No form, no qualification call first.
Prefer email? hello@pinpointproof.com

The problem

Every firm now has an AI budget, and no method for spending it.

Use cases get picked by anecdote, with no baseline. Nobody can prove benefit afterwards, or show the board how they decided.

01

Chosen by anecdote

The loudest process wins, not the most valuable one. The work with the largest prize is usually invisible because nobody has counted it.

02

No baseline

Without a measured before, there is no after. Benefit becomes a matter of opinion at exactly the moment the board asks for a number.

03

No evidence trail

Six months on, nobody can show how the decision was made, what was rejected, or which controls were considered.

How it works

Four steps from “where do we start?” to proven value.

STEP 1

Map

Facilitated capture with the people who do the work. Activities, systems, handoffs, controls, volumes, exceptions. Validated by a named person, not drawn from a document.

STEP 2

Score

Every step scored against six weighted factors, with two constraint modifiers for regulatory sensitivity and change readiness. Every score traces to the evidence that produced it.

STEP 3

Prove

Each candidate gets the smallest test that could disprove the benefit, with a threshold agreed before it runs. A test that kills a weak idea in two weeks has done its job.

STEP 4

Scale

What survives moves into delivery with its control pattern, human approval points and audit trail already specified. Realised benefit is tracked back to the baseline.

The disposition ladder starts at eliminate and standardise, not automate. Automating a broken process is the most expensive mistake available. See the full scoring method →

Where this goes

HR owns people. IT owns systems.
Nobody owns the work.

Every regulated firm is about to run a hybrid workforce of people and agents. Someone will have to answer for it.

  1. What processes exist, and which of them are regulated?
  2. Who, or what, performs each step?
  3. Under which controls, and with which human approval points?
  4. Producing what evidence, at what cost and performance?
  5. Changed by whom, approved by whom, and when?
  6. And can you show all of that to someone who asks?

Today that answer usually lives in a spreadsheet one person maintains. AI inventory expectations, operational resilience mapping, DORA-style registers and senior manager accountability are all converging on the same artefact. It does not exist yet.

A map built this way is already most of that answer. Our direction is to make it live: the system of record for how work happens, human and machine, and the control tower that proves it.

Where each layer stands

Five layers. We will always tell you which one you are buying.

Most of this is direction, not product. Saying so is part of how we work.

LAYER 1Evidence
Validated maps, AIQ scoring, a ranked opportunity register, and a designed test against every candidate.
Built. This is what the sprint delivers today.
LAYER 2Specification
Every step machine-readable: inputs, outputs, rules, permitted values, controls and KPIs. SOPs ingested as evidence with provenance, including where observed work diverges from the document.
In build. The next increment.
LAYER 3Generation
One specification compiled to any target: an agent scaffold, an automation configuration, or a better SOP and training pack for the work that should stay human.
In design. Not available, not sold, not in any fee.
LAYER 4Assurance
The control tower. Live current versus target operating model, drift detection when behaviour diverges from specification, benefit realisation against recorded baselines.
Direction. Nothing committed to a date.
LAYER 5Network
Reference process libraries by sector, anonymised benchmarks, and certified partners delivering sprints on the platform.
Later, and only if layers 2 to 4 earn it.

Prototype and prove, never operate

Our design team designs and prototypes new features, including functioning mockups tested fully in a sandbox to prove the value. Your in-house technology team then packages and deploys them through your own change control, with our support at handover. We will not run a regulated process on your behalf, and that limit is permanent.

Evidence for your governance, not a compliance opinion

Everything we produce is an input to your governance. We do not calculate a regulatory outcome, produce a filing, or attest to anything.

Why not the alternatives

The right analysis, in the right container.

Alternative
The flaw
PinpointProof
Process miningCelonis, Signavio, UiPath
Needs event-log data from systems of record. Your most expensive processes run on email, spreadsheets and portals, so they are invisible to it.
Captures work as it is actually done, not only as systems record it. Why →
A consultancy assessmentBoutique or Big Four
Six to eight weeks and a large fee for a deck that is out of date within a quarter and cannot be re-run.
A live workspace you keep and re-run after every change.
A workshop and a spreadsheetDone internally
No scoring method, no baseline, no evidence trail. The reasoning lives in one person's memory, so the board sees a conclusion without the evidence that produced it.
A scored register where every recommendation traces to its evidence.
Just start buildingThe most common route
Automates whichever process had the loudest advocate, often one that should have been eliminated instead.
A designed test that kills weak candidates in two weeks, not two quarters.
The regulatory dividend

The map you build for AI is the evidence you owe your regulator.

Because every map carries the controls, owners and obligations, it doubles as evidence for UK operational resilience and DORA obligations. Every score traces back to the process elements that produced it. The result is a living, auditable record of how the firm operates. How one map serves both →

Owners and controls

Every step carries its named owner, its control pattern and its human approval points, captured once and used for both purposes.

Traceable scores

Any score can be opened to show the evidence behind it. Change the evidence and the score changes with it.

It never goes stale

Re-run it after every change. The map is the operating record, not a document about the operating record.

Who it’s for

Built for the people who own the AI mandate.

Regulated financial services firms of 100 to 2,000 staff, UK and Ireland first.

The seat

Chief Operating OfficerHead of OperationsChief Transformation OfficerHead of ChangeHead of Operational ExcellenceHead of Operational Resilience

The sector

Fund administrationAsset & wealth managementInsuranceSpecialist bankingPayments

Who it is not for

Firms under a hundred staff, where there is no change budget and the economics do not work.

Anyone midway through an enterprise process mining implementation. You already have a tool and a team, and a second one will not help.

Anyone who wants a diagram for an audit and nothing else. That is a documentation job, and there are cheaper ways to buy it.

FAQ

Questions, answered.

What is an AI opportunity assessment?

An AI opportunity assessment is a structured review that identifies which parts of a business process could be performed by a machine, ranks them by value and feasibility, and attaches evidence to each recommendation. A credible one produces four things: a validated map of how the work is actually done, a score for every step, a measured baseline to judge benefit against, and a designed test for each candidate before anything is built.

Why does process mining not work for fund administration or insurance operations?

Process mining reconstructs a process from event logs produced by systems of record. It works well where the whole process runs inside one or two transactional systems. In fund administration, insurance operations and specialist banking, the most expensive work runs across email, spreadsheets, shared mailboxes and counterparty portals, which generate no usable event log. The processes with the largest prize are therefore the ones process mining cannot see. Capture has to start from how the work is actually done.

How long does it take, and what does it cost?

Four weeks for one business area, ten to fifteen processes, at a fixed fee. There are no day rates and no variation on the scope agreed at kick-off. A single-process pilot is available if you would rather test the method before committing to the full scope.

How much of my team’s time does it need?

Around twenty-six person-hours in total. Ninety minutes each from the people who actually do the work, a few hours from the operations lead, and ninety minutes from risk and compliance at the challenge session. Late input is the single most common reason engagements of this kind slip, so the commitment is set out in the proposal and agreed before we start.

Do you need access to our systems?

No. No integration, no data extraction, no connection to your environment at any point. Read-only walkthrough access or a screen-share is enough. We do not need investor or client personal data and would prefer not to receive it.

Will you build the automation for us?

We can take you as far as a working prototype. Our design team designs and prototypes new features, including functioning mockups that can be fully tested in a sandbox environment against real cases, so the value is proven before anything is built for production. In larger firms we then work with your in-house technology team to hand over the prototype, its specification and its test evidence, so it can be securely packaged and deployed within your existing technology governance and change control. One limit is permanent: we will not operate a regulated process on your behalf. Your accountable owner deploys, which keeps the accountability line clear.

Is this a compliance product?

No. Everything we produce is an input to your governance. It is evidence you can use, not a verdict we issue. We do not calculate regulatory outcomes, produce filings, or attest to anything.

How many clients do you have?

PinpointProof is early, and we are not going to pretend otherwise. There is no long client list and there are no case studies yet. What there is: a published method, a scoring framework you can inspect before you commit, and twenty years inside fund operations and regulatory technology. Being early is why the fee is what it is, and why you get the founder rather than a delivery team.

Know where to start.
Prove it worked.

Book a thirty-minute walkthrough. Bring one process; we’ll show you what its map, scores and opportunity register look like.

Pick a time that suits you. No form to fill in first.