The work crosses departments
Sales starts it, operations continues it, finance closes it — in three systems.
Solution · Engineering
Custom AI Systems
For processes that cross departments and systems, where the work does not fit any standard product — built as software, phased, and scoped before anything is committed.
Scoped and delivered in phasesEach phase approved separately
Who it's for
- A process that crosses departments and systems
- Rules specific to your business, built up over years
- Internal or legacy systems with an interface
- An owner who can decide across those departments
The problem
Some processes do not fit a product.
01
02
The rules are specific to you
Built up over years, held by experienced staff, written down nowhere complete.
03
Off-the-shelf tools cover part of it
So the gaps are filled by people, spreadsheets and re-keying.
Starting scope
What a first project actually looks like.
What the live trial covers
- The narrowest part of the wider process that can run on its own
- The narrowest useful slice of the wider process
- A named set of systems for that phase
- Approval rules defined per action
- A written record of the rules as implemented
- A decision point before the next phase is committed
What the trial should show
- The phase delivers a usable outcome on its own
- Handovers between steps carry the state the next step needs
- Financial and irreversible actions stop for a named approver
- Configured actions are recorded and attributable
When this is not the right fit
- The systems involved cannot expose what the process needs
- No single person can decide across the departments involved
- The process is being redesigned at the same time
- A single-workflow live trial would answer the question more cheaply
€0 upfrontRun it in your business for 3 days before you pay. If you keep it, a Live AI Workflow starts from €1,500 for one workflow.Start a free live trial
Full working functionality for the agreed workflow — not an artificially limited demo. The workflow itself is agreed with you before the trial starts.
What the system does
Several agents, one process, defined boundaries.
Where a workflow spans several departments, multiple agents can coordinate, each with its own scope and permissions, and a defined handover between them.
Coordinates several agentsEach responsible for one part of the process, with an explicit handover.
Integrates systems without a standard connectorIncluding internal and legacy systems, where an interface exists.
Encodes your specific rulesWritten down during the build, which is often valuable on its own.
Builds internal AI applicationsWhere staff need an interface rather than an automation running unseen.
Applies governance per stepPermissions, approvals and logging defined per action rather than globally.
Delivers in phasesEach phase scoped, approved and useful on its own.
Before and after
What changes for a cross-department process.
01todaySales starts it, operations continues it, finance closes it — in three systems.with the systemOne process carries the request, with a defined handover at each boundary.
02todayThe gaps between tools are filled by people and spreadsheets.with the systemThe gaps are closed by integration, with exceptions surfaced.
03todayThe rules exist only in experienced people's heads.with the systemThe rules are written down during the build, which is valuable on its own.
04todayNobody can say where a request currently is.with the systemState and ownership are visible at every step.
Example workflow
A cross-department request, carried end to end.
01Request enters at any pointSales, support, a portal or an internal system event.RECEIVED
02Routed to the responsible agentBased on the type of work rather than who happens to be free.DONE
03Data gathered across systemsEach read under its own permissions.DONE
04Handover between agentsWith the context and state the next step needs.DONE
05Commitment or cost involvedThe workflow stops for approval by the accountable person.NEEDS APPROVAL
06Actions executed and recordedAcross each system involved, with events logged for review.DONE
Illustrative workflow · configuration varies by project
Integrations
It connects to what has an interface.
Feasibility depends on access rather than ambition. The first question in scoping is always what each system can actually expose, and what it cannot.
Systems with usable APIsLegacy systemsCRMERPDatabasesDocumentsPrivate modelsInternal tools
Integration feasibility is confirmed during scoping.
Human control
Governance is designed per action.
In a multi-agent system the question is not whether humans stay in control, but exactly where. That map is produced during design and agreed before build.
Financial commitmentsalways approved by a person
Cross-department handoverslogged and attributable
Actions visible to customersapproval rule set per action
Unrecognised casesescalated, not attempted
Example implementations
Where this fits.
Multi-site operations
The same process run differently in each location
One system with local variations, instead of several incompatible habits.
Regulated environments
Approval and evidence requirements
Approval flows and logging designed into the workflow from the start.
Established businesses
Long-standing internal systems
Automation built around systems that cannot realistically be replaced.
Tell us the workflow, we build and connect it, you run it live for 3 days, then keep it or switch it off. Larger systems start from €7,500.See pricing
Questions
What technical buyers ask.
We do not quote the whole programme up front. For complex multi-agent programmes, deeper discovery is agreed as part of scoping to establish system access, dependencies and sequencing. Delivery is phased so each phase is approved on its own merits. Where possible, the first phase is deliberately kept small to validate the approach before expansion.
Project ownership, infrastructure accounts and handover arrangements are defined transparently before implementation begins. Our preferred approach is to avoid unnecessary vendor lock-in and to make infrastructure ownership clear from the start.
Deployment options include your own cloud, an isolated environment or on-premise where technically appropriate. The choice is driven by data sensitivity and your infrastructure policy, and is agreed during design.
Try this in your real business for 3 days.
Where a narrow slice of the wider process can run on its own, it can be trialled live for 3 days first. Larger programmes are scoped with you directly.