Workflow Assessment
Two weeks with your process owners and system records. You get a measured baseline and a clear decision on what to build first, or not to build.
Two weeks. Read-only access under NDA. Fixed fee.
First System Delivery
We build in your environment on your existing systems and validate against real cases before the system carries live work.
Scope and fixed fee agreed after the assessment.
Embedded Engineering
Our engineers stay alongside your team to extend the system or build the next one, each with its own handover.
Time-boxed, with a fixed fee per workflow.
How we work
How we build systems your team can own.
Our engineers assess the work, build in your environment and test against real cases. Every stage has an outcome you can inspect.
- 01
Assess the work
We observe the process, systems and exceptions before proposing a build.
OutcomeA measured baseline and a clear statement of the workflow to improve.
- 02
Design the first system
We decide what AI reads, what rules decide and what a person approves, for one bounded first workflow.
OutcomeA written build or do-not-build recommendation after a two-week assessment.
- 03
Build in your environment
We connect your existing systems and test on your own cases before anything goes live.
OutcomeA live system in your environment, with approvals, exception paths and a readable audit trail.
- 04
Hand over and improve
We document the system, train your team and extend it only where the next workflow justifies the work.
OutcomeA system your team understands and owns, with a defined scope for any extension.
What Your Organisation Owns at Handover
Every engagement ends with the same handover pack, held in your environment: source code, rules, test cases, the decision log and the operating manual.
- Code stored on your side
- Rules your team can read
- Models you can swap
- Hosting you choose
- src/The agents, the rules and the connectors to your systems
- rules/approval-rules.yamlWho approves what, and above which amount
- evals/real-cases.csvYour own cases the system is tested on
- audit/decision-log.sqlEvery decision, with its evidence and approver
- operating-manual.mdHow to run, monitor and change it
- hosting.mdWhere it runs and which models it uses
- handover-signoff.pdfSigned by your team and ours
What we deliver together
Work scoped to a clear outcome.
Each step is scoped and priced on its own. The next one starts only if the results justify it.
How we build
How we keep the system dependable
- Build on what you already run
- We connect to your ERP, CRM and document systems. Core processes stay where your team works.
- Models never invent numbers
- Models read and suggest. Code calculates every amount, date and balance. People decide what carries financial, legal or customer impact.
- Acceptance against real cases
- Acceptance is defined on your own historical cases, with the target error rate agreed before we start.
- Readable for the people who run it
- Systems, documentation and logs are written for the team that operates them, not only for engineers.
- Capacity, not headcount
- We measure success by the demand your team can now serve, never by the people removed. The system is built to support their judgement, not replace it.
What we will not do
Work we decline
We decline engagements where the result cannot be measured.
- Staff augmentation to operate an existing process without improving the system behind it.
- Open-ended retainers without a defined scope, time frame and handover.
- Chatbots without a defined workflow and a measured outcome.
- Data migrations or core system replacements presented as AI projects.
- Automation of a process that has not first been observed, mapped and measured.