00Approach
Tested step by step, and yours to keep.
Every engagement follows the same path: audit, pilot, deploy and, if you want it, operate. Each phase is scoped and priced on its own, so you commit to one step at a time. What we build for you is yours.
How we keep your risk low.
One baseline from start to finish
The hours baseline from the audit stays the yardstick. The pilot, the launch and every monthly report are measured against it, with the same definitions.
Test on your own data
Your real cases go into the tests, including the scanned, handwritten and incomplete ones. The pilot results, measured against the baseline, decide what goes live.
A person approves what matters
Each system has a human in the loop (a person who approves a step before it runs) wherever a mistake would be costly. A new system starts in draft mode and runs steps on its own only when the results justify it.
The simplest tool that works
We start with the software you already pay for and add tools such as n8n or Make where they fit. Custom code comes in only where they fall short. A system with fewer moving parts has fewer things to break.
Documented, so you can leave
Every system ships with a runbook (how to run and repair it), its documentation and its test cases. Your own staff or another supplier could run it from those alone. We would rather keep your work by doing it well.
A decision after every phase
Each phase ends with measured results and a choice that is yours: continue, change course or stop. If the numbers do not support the next step, we say so before you pay for it.
From audit to daily use.
-
Audit
Find where the hours go, decide what is worth automating and define what a first pilot has to prove.
What we do
- We agree the scope on a short call: one or two related processes, the people who run them and the systems involved.
- Everyone who does the work spends about an hour with us, talking it through and showing it on screen.
- Real cases are timed step by step, and volumes come from samples or read-only access to your systems.
- Every exception, workaround and handover is listed, with how often it happens.
- We rank each opportunity by hours returned, effort and risk, and note any personal data it touches.
What you get
- Process map, checked by the people who do the work
- Hours baseline for every step in scope
- Ranked list of opportunities, with a data and security review
- Pilot proposal with a fixed scope, price and timeline
What we need from you
- A process owner who can make decisions on scope and next steps
- Space in your team's week for the interviews and screen shares
- Read-only access or anonymised samples, whichever you prefer
-
Pilot
Test one workflow on your own cases and compare it with the baseline before anyone depends on it.
What we do
- We build the workflow from the pilot proposal in a test environment, in accounts opened in your company's name.
- Past cases become test cases, each with the answer your team considers correct.
- The workflow runs in draft mode alongside your team, and a person checks every output before it counts.
- With the people who will handle exceptions, we tune the rules and thresholds that send a case to a person.
- Time per case, errors and the share of cases sent to a person are measured against the baseline.
What you get
- One workflow, tried by your team on real cases
- Results against the baseline, with the raw figures
- A recommendation to deploy, change or stop, with reasons
- A fixed-scope proposal for deployment, if results support it
What we need from you
- A batch of real past cases, such as last month's invoices or emails
- A reviewer who knows what a correct result looks like, for an hour or two a week
- Access to the systems involved, limited to what the pilot needs
-
Deploy
Make the pilot safe for daily use, train the people around it and switch it on without stopping the work.
What we do
- We set up production access with the narrowest permissions each connection can work with.
- Monitoring, alerts and logs are added, so every run can be traced and every failure reaches a named person.
- You set the approval thresholds with us: what runs on its own and what waits for a person.
- We write the runbook and documentation and train everyone who reviews or approves the system's work.
- The switch happens in stages, and the old way stays available as a fallback until the results match.
What you get
- The system running in production, in your accounts
- Runbook, documentation and test cases
- Monitoring and alerts, each with a named owner
- A trained team that knows when to override the system
What we need from you
- Sign-off on thresholds, roles and who receives alerts
- Training time for everyone who will use or review the system
- An admin or IT contact to set up live accounts and sign-in
-
Operate
Keep the system accurate as your business, data and models change, and extend it when you are ready. You can also take it in-house.
What we do
- We review alerts and failed runs, find the cause and fix it.
- We check samples of the system's work against the test cases to catch any slow decline in quality.
- The tests run again after any change to a model, prompt or connected system.
- Each month you get a written report on volumes, hours returned, exceptions and changes.
- When you are ready, we scope the next workflow from the ranked list.
What you get
- Monthly report on hours returned, exceptions and changes
- Fixes and improvements, each logged and documented
- New workflows, each scoped and agreed before we build
What we need from you
- A named contact who reads the monthly report and sets priorities
- Early notice of changes on your side, such as a new system, form or supplier
Three models, each agreed in writing.
Fixed scope
A defined piece of work for a price agreed in writing before it starts. The proposal states what the system will do, how we will measure it and what we need from you.
- If the build runs over our estimate, the agreed price still holds.
- A change of scope is quoted and agreed in writing before we start on it.
- Tools and model usage are paid from your own accounts, so you see running costs from day one.
Best for Pilots, deployments and any build whose result can be described in advance.
Operate & improve
After launch, we keep the system working and improve it: monitoring, fixes, retests and agreed changes, with a written report each month.
- The monthly fee is agreed upfront and names the systems we look after and what is included.
- Larger additions, such as a new workflow, are scoped and priced as fixed-scope work first.
- We review the arrangement with you against the monthly reports, so the fee stays tied to visible work.
Best for Workflows people depend on daily, where an unnoticed failure would cost hours.
Performance-linked
A share of our fee depends on a result both sides can measure, such as hours returned or meetings booked. It is available for selected projects.
- We set the measure, the baseline and the measurement period in writing at the start.
- Both sides read the same figures, taken from your CRM, accounting or other systems of record.
- We offer it only where our work drives the result and outside factors, such as seasonality, play a small part.
Best for Processes with a clean baseline and a result that can be counted without debate.
What we need from you.
- One process owner who knows the work, can make decisions and stays with the project from audit to launch.
- Time with the people who do the work, mostly during the audit and the pilot.
- Real examples to build and test against, such as last month's invoices, orders or emails, anonymised if you prefer.
- Access to the systems involved, granted from accounts you control and removed when it is no longer needed.
- Clear rules, or someone who can set them. A system is only as precise as the rules behind it.
- Technical staff are optional. If you have an IT or data protection lead, we involve them in access and data decisions.
What stays with you.
- The code we write for you, kept in a repository registered to your company.
- Every workflow, prompt and test case, with the documentation that explains how they fit together.
- The accounts the systems run in, such as automation tools, model providers, databases and hosting, all in your company's name.
- Credentials and API keys, held in accounts and vaults you control. We work only with access you grant, and you can withdraw it at any time.
- Logs, records and reports, stored in your accounts where you can read and export them.
- The audit report and pilot proposal, whether or not you continue with us.
What happens after launch.
Support is part of Operate & improve. We monitor each system, fix what breaks and keep its tests current as your business changes. You can raise a problem by email or in your team chat. Anything that stops your team working comes before improvements.
- Monitoring: failed runs and unusual results alert us and the person you name on your side.
- Fixes: we trace each failure in the logs, correct the cause and record what changed.
- Changes: new rules, fields or reports are run against the test cases before they go live.
- Monthly report: volumes handled, hours returned against the baseline, cases passed to people and changes made.
- Handover: if you take a system in-house, you get the runbook, the test cases and a walk-through.
Start with the audit, then decide.
The audit takes 1–2 weeks and has a fixed fee, agreed in writing before it starts. The report and the pilot proposal stay with you either way.