Each agent is scoped to a single defined work package, trained on your documented way of working, connected to your systems, and bounded by rules about what it may decide alone and what must come to a person. That scoping discipline is why our deployments hold up under audit.
Invoice matching, reconciliation prep, onboarding checks — defined inputs, defined outputs, measurable success criteria.
Your SOPs, templates and business rules become its operating manual, with a named owner and version control.
Autonomy stops where you decide. Anything outside tolerance escalates to a person, and every action is logged.
The same capability, three commercial models. Most clients start in the middle and move outward as confidence builds.
We configure and operate an AI-enabled capability for one of your defined processes, under agreed service levels and controls. You keep the process; we run the agent.
You transfer responsibility for the whole process to Oban. We determine the right blend of technology and people, and we are accountable for the outcome.
The agent performs the repeatable work, an Oban professional supervises quality and exceptions, and you retain final approval only where risk or regulation requires it.
Nothing goes live until it has run in parallel with your existing process and met acceptance criteria that you set.
A short form — 10 to 15 minutes — to confirm fit, the target process, executive support and any hard constraints.
60–90 minutes per process to validate the real workflow, controls, exceptions and the outcome you want.
We review SOPs, representative inputs and outputs, exception cases and actual performance data.
We score automation potential, implementation effort, control requirements and expected value.
Your findings, recommended operating model, pilot scope, roadmap, ROI model and commercial proposal.
The agent runs alongside the existing process on live work until your agreed acceptance criteria are met.
Production under service levels, governance, reporting and change control — with agreed exit arrangements from day one.
The same principle that governs our middle-office teams applies to our agents: they execute, they never hold authority.
Each level sets who reviews the work, who approves it, and how much of the outcome Oban is accountable for. You can sit at one level for a first process and another for a second.
The agent drafts and prepares; your team reviews every output. The fastest way to build internal confidence.
Oban reviews every output before release; you approve exceptions only. The usual first production level.
Oban owns the process to an agreed SLA with risk-based sampling; you approve only high-risk or regulated items.
Full process ownership, a dedicated operations pod, continuous optimisation and executive reporting.
About two minutes. The questions adapt to your answers, so you only see what’s relevant to you. You’ll get an instant AI Readiness Score across the same seven dimensions we use in our full assessment, plus where to start.
If you would rather talk it through than fill in a form, we are happy to start there.
Start the Conversation →