How we work
From growth goal to a system that can carry it
You have more demand ahead than your current operation can comfortably carry. This page shows how we turn one growth bottleneck into a working system in your own environment, and how we extend it. You leave knowing the steps, the offers and what you hold at each stage.
- 01
Understand the growth goal
We start from the growth you want: more customers, higher volume or a new market, then trace back to the workflow that limits it today.
We ask you to describe the growth you are aiming for in concrete terms: volumes, markets, products, target response times. The room usually holds a sponsor, an operations or finance lead, and someone who knows the workflow day to day. Together we map how work moves now, which systems it touches and where delays or errors appear. You see your goal and the constraint on a single page.
We then agree what success would look like if that constraint moved. That can be more orders processed per day, faster responses, fewer manual checks or clearer cross-market visibility. We do not start with models or tools. We start with the measurable shift in operating capacity that would support your growth, and we write it down in simple terms you can share with your team.
A written growth objective and the workflow that limits it, described in numbers and plain language.
- 02
Find the bottleneck
We study how the work really happens, quantify the baseline and decide which workflow is worth building first.
Through the Workflow Assessment we take two weeks to observe the target workflow, with read-only access to your ERP, CRM, document stores or exports under NDA. We sit with the people who touch the work and watch how they use the systems. You see your current throughput, error rates and exception handling written down, so the size of the gap between ambition and capacity is clear.
We then test a few candidate workflows against three questions: how strongly they link to your growth goal, how measurable they are and how feasible a first build would be in your environment. In a closing session we share our view: build now or do not build yet. If it is a build, we propose the first workflow and outline an approach. If it is not, we explain why. There is no obligation to proceed.
A two-week assessment that ends in a written recommendation: build this workflow now, or do not build yet.
- 03
Build in your environment
We design and deliver the first system on your existing stack, with approval steps, exception paths and clear controls.
With the First System Delivery we build the first agent system directly on your ERP, CRM, bank exports or document stores, rather than asking you to move. We keep models narrow and let deterministic code calculate amounts, dates and balances. A person still decides consequential things. You see the proposed design in diagrams and test cases before any code is finalised, so you can check that it mirrors your real work.
We wire in approval steps and exception paths that match how your team already exercises judgement. Every action the system takes leaves a trail in language a non-engineer can read. Before anything goes live we agree the real cases that count as acceptance. We write down the target error rate and measure against it. When we hand over, you have a working system in production, with documentation your operations or finance manager can actually use.
The first live system in your own environment, tested against real cases, with approval, exceptions and a readable audit trail.
- 04
Expand with confidence
Once the first system is live, we extend it and apply the method to new workflows, each with its own end date and handover.
Through Embedded Engineering our team stays close to yours for a defined period, usually focused on one workflow at a time. We sit in your planning calls, watch how the first system behaves in production and collect the edge cases it surfaces. You see a steady rhythm of small extensions and new checks, not a long hidden build. Each change is tested against real examples and explained in simple terms to the people who rely on it.
We then apply the same method to the next workflows that support your growth: another finance control, a cross-market reporting link, a high-volume review step. Each workflow has its own scope, time-box and handover. Six months later you have a connected set of systems that share data, route work automatically where it is safe to do so and keep your team in control of the exceptions. The capacity to grow is now built into how you operate every day.
A connected set of workflows that can absorb more demand, with each system owned, documented and understandable to your team.
What we deliver together
Offers that match where you are
Whether you are probing a single process or ready to extend a working system across the business, each offer starts from a clear growth goal and ends in something concrete you can hold, run and measure.
Workflow Assessment
Two weeks to map the growth opportunity against the way work really flows today. We observe the workflow, read your existing systems under NDA and quantify throughput, delays and error rates. The aim is to find the workflow that is both tightly linked to your goal and realistic as a first build in your environment.
Two weeks. Read-only access under NDA. Fixed fee, agreed before we start. Ends in a written build or do-not-build recommendation.
First System Delivery
We design and deliver the first agent system in your own environment. It connects to your ERP, CRM, bank exports or document stores. Deterministic code handles calculations, the model handles pattern recognition, and people approve high-impact decisions. It is validated against agreed real cases before it ever carries live volume.
Scope and fixed fee agreed after the assessment. Ends with one live system, documentation and an agreed handover.
Embedded Engineering
Our engineers stay embedded with your team to extend the first system and apply the same method to the next workflows. Each workflow has its own scope, test cases, acceptance criteria and handover. You get a sequence of working systems that together increase your operating capacity to support growth.
Time-boxed per workflow, never open-ended. Fixed fee, agreed before we start each workflow. Ends with a handed-over system each time.
How we build
Six principles for systems that last
- Build on what you already run
- We connect to the systems you already use: ERP, CRM, document stores, the shared drive. We do not ask you to move core processes to a new stack. The work lives where your team already works, which makes change management lighter and reduces operational risk as you grow.
- Models never invent numbers
- Models classify, match and suggest. Deterministic code calculates every amount, date and balance. A person still decides things with real financial, legal or customer impact. This keeps the line between judgement and automation clear and makes the system easier to trust and to audit.
- Verifiability before automation
- Before we automate a step, we make it easy to verify. People can see what the system did, why it did it and what data it used. Only when they are comfortable checking the work do we raise the level of automation. This keeps control in your hands as volume grows.
- Fewer moving parts, clear trails
- We minimise the number of components in a workflow and choose tools your team or internal IT can understand. Every important decision leaves a trail in language a non-engineer can read. That trail is for your operations and finance leaders, not only for engineers.
- Acceptance against real cases
- We define acceptance using real cases from your own history that match your own edges and exceptions, not generic test data. We write down the error rate we are aiming for before we start. You see how the system behaves on concrete examples that matter to your business.
- Readable for the inheritors
- We structure systems, documentation and dashboards so that an operations manager, a finance manager or an IT generalist can read them. We expect that people who were not in the build will have to run the system six months from now. We write and design for them.
What we will not do
Work that is not for us
Saying no is part of protecting your growth goal. If what you need is on this list, we are not the right partner and we will say so early.
- We do not provide extra hands to operate your existing process without changing the system behind it.
- We do not accept retainers without an end date. Every engagement has a clear scope, time-box and handover point.
- We do not build chatbots just because they look like AI, without a clear workflow and measured outcome behind them.
- We do not take on AI projects that are secretly data migrations or core system replacements wearing a new label.
- We do not automate a process nobody has sat down to observe, map and quantify in its current form.
How we price
Fixed fee, agreed before we start
Every engagement runs on a fixed fee, agreed before we start. There are no usage surprises mid-project. For Workflow Assessments the fee covers two weeks of observation, analysis and recommendations, with read-only access under NDA.
For First System Delivery and Embedded Engineering we agree the scope, time-box and acceptance criteria first, then set a fixed fee around that. You always know what each workflow will cost, what you will receive at handover and when the work will end. There are no open-ended retainers.
If we cannot see a path to a measurable result that justifies the cost, we will recommend that you do not build yet. Our aim is a working, owned system in your environment, not a perpetual project.
Next step
Talk about the growth you want next
If you can see more demand ahead than your current operation can comfortably carry, tell us about it. A senior engineer talks through your growth goal, the workflows behind it and whether a Workflow Assessment makes sense, and replies within two business days.
Fixed fee, agreed before we start. A senior engineer replies within two business days.