We map the process, define the delivery and automate what actually creates value – with human controls, traceability and measurable results. Two entry points: process analysis on site, or automation of an already described flow.
Process analysis on site suits organisations that know something is inefficient but do not yet have a described process. We interview, observe and document the current state. The result is decision support – even if the decision is not to automate.
Automation of a described flow suits organisations that already have a process description and want it built, tested and operated. Here we start by testing whether the description is sufficient for a defined delivery.
The process, scope, assumptions and versions should be the same through the website, meeting, quote, order, delivery and follow-up. A changed scope must require a new calculation and a new decision – not a silent adjustment afterwards.
We therefore work with versioned records and explicit decision points: purchase, data access, acceptance and production activation are separate decisions.
Not everything should be automated. High-volume deterministic processes can often be solved more simply with an ordinary integration. Processes without an owner, a measurable baseline or clear rules should first be improved manually.
We say so when a simpler solution is better. That is part of the delivery, not a failure.
The organizations we work with typically face at least one of these.
The same information moves between email, spreadsheets, ERP and CRM. Everyone knows the work exists, but nobody has measured what it costs.
Data lives in several systems with different owners, formats and truths. Every handover creates waiting, duplicate work and new errors.
A demo impresses, but lacks acceptance criteria, error handling, permissions, monitoring and a named owner in operations. Then it stops.
Time, quality and cost are not measured before and after. Without a baseline you cannot tell whether automation helped – or merely moved the work.
A selective portfolio where senior expertise makes the biggest difference.
Interviews, observation and a current-state map. You get a described process, exceptions, volumes, systems and decision support. The analysis can end without an implementation purchase.
When the flow is already described, we build a defined automation with a clear start, end, owner, acceptance criteria and error path. No hidden scope.
We compare candidates by volume, variation, system access, risk and value – so you automate what is worth doing first.
We connect the systems the process actually uses. Where an API exists we use it; where it does not, we describe the limitation instead of hiding it.
AI agents can be used where variation and judgement require them. They work with small, schema-bound tools, human approvals and full traceability – not unrestricted production access.
After delivery we track cycle time, error rate, exceptions, cost per case and realised benefit. Deviations and remaining manual work are reported openly.
A clear process from first conversation to delivered result.
We start with the process, not the tool: start, end, owner, volume, exceptions, systems and today's baseline.
We choose a delivery that can be verified, with acceptance criteria, responsibility, dependencies and a clear boundary against customer and third-party commitments.
The automation is built in a defined environment with error handling, permissions, logging and a manual fallback.
We test the normal flow, exceptions, permissions, retries and restore. Only then can the delivery be accepted.
Production activation is a separate decision. After that: monitoring, incident path, maintenance and benefit measurement.
Include the whole workflow, including cases that need manual help. Freed time has a capacity value. Cash savings require a plan for how time is used and which costs actually disappear. All amounts exclude VAT.
Count each effort once. Oversight means work not already included in per-case handling or review. The values above are editable examples.
Enter how much freed time reduces expenditure and the cost that can actually be avoided per hour. Loaded hourly cost is used only for capacity value. More work does not automatically mean an extra payment: assess additional staffing payments separately. Count each cost once. Zero is an explicit assumption; blank means not assessed.
Time and capacity results show full adoption. In the cash flow, both avoided expenditure and additional staffing payments ramp with adoption. Operating costs begin in month one.
Cash flow and payback require an assessed realisation share, avoidable hourly cost, additional staffing payments and operating cost. Enter 0 for costs assessed as zero.
An estimate based on your assumptions, without a quote or guarantee. The model assumes constant monthly volume and costs and excludes interest, tax and residual value. Your inputs stay in your browser.
Start with one sentence. We use this to propose the next step – a process analysis or a defined automation. This is an enquiry; scope and price are agreed before ordering. Keep passwords and other sensitive information out of the form.
Process analysis and automation connect with our consulting in strategy, product development, integrations, data, cloud and security. The Open Source hub collects our technology selection tools. We agree the expertise and scope needed for your assignment.
Pricing is set after scoping. No older price examples are binding.
Interviews, mapping and decision support. Can end without an implementation purchase.
One defined process in a pilot environment with acceptance criteria and a measurement plan.
Production activation, monitoring and maintenance under an agreement.
Learn more about this topic in our articles.
Answers to the questions I hear most often.
Process automation means that a recurring workflow is wholly or partly performed by systems instead of people. It can be deterministic integrations, AI agents or a combination. The choice should follow the process's variation, volume, risk and system access – not whichever tool is popular.
No. The process analysis is a standalone delivery and can end with a decision not to automate, to make simpler improvements, or to procure implementation separately.
In principle anything with a stable API or a documented integration. We first map which systems and permissions actually exist. If an API is missing, we describe the limitation and propose another path instead of promising an integration that cannot be operated.
We set a baseline before delivery: volume, active handling time, lead time, errors and exceptions. After delivery we measure the same metrics for an agreed period. We separate freed time, actually reduced costs and hypothetical value.
They work with bounded tools and permissions, and actions that affect customers or finances require human approval. We log every call and can disable the agent without shutting down the whole process.
Data ownership, storage, access and retention are defined in the agreement. We do not use customer data to train models and we document which external services are part of the flow.
Next step
Do you have an ambitious idea or a technical decision where it pays to think right from the start? Get in touch – no obligation.
Get in touch
