One recurring process, its owner, and examples of real cases.
Automation readiness assessment
Before building a bot, find the process worth automating.
We turn one real business workflow into a clear, pilot-ready decision by documenting how the work happens today, which data and systems it uses, where exceptions and risks lie, and how success will be measured.
Five practical documents to help you decide whether and how to run a focused pilot.
Run a pilot now, fix the prerequisites first, or wait before automating.
When an assessment helps
You have an AI idea, but you do not yet know whether it is the right process to automate.
An assessment helps when the team knows where time is lost but scope, data, risk, or expected benefit is not yet clear enough to build.
Too many manual handoffs
People re-enter data across email, spreadsheets, CRM, ERP, documents, and other internal systems.
Long queues and delays
Requests stall because information, decisions, and ownership are not available in the same step.
Unclear exceptions
The standard case looks simple, but non-standard cases, bad inputs, and approvals carry most of the risk.
Too many possible tools
It is unclear whether the process needs rules, traditional automation, an AI assistant, an agent, or a combination.
What we examine
Process, data, and control before technology.
A good candidate needs more than an impressive demo. It also needs a sensible scope, reliable inputs, clear ownership, and a safe way to test results.
- Case volume, cycle time, manual steps, delays, rework, and the current effort required.
- Data sources, record quality, access rights, and available APIs, webhooks, connectors, or MCP tools.
- Business rules, decisions, exceptions, and the moments that require expert judgement.
- Actions a bot may propose, perform, or must send to a person for approval.
- Quality, safety, and operational measures that can be compared with the starting point.
Concrete deliverables
Five practical documents for making a decision.
You do not get a generic slide deck. You get a compact set of materials that defines the real process and the conditions for a fair pilot test.
Process map
The main path, exceptions, roles, systems, inputs, outputs, and the points where work waits or moves backward.
Starting measurements
Available starting measures such as cycle time, number of manual steps, rework rate, response time, and case volume.
Risk and approval matrix
Which actions may be automated, which need human review, and how to limit access, log actions, and stop the workflow safely.
Pilot plan
A one-page definition of the problem, scope, users, integrations, boundaries, owner, measures, and criteria to continue or stop.
Test cases
A representative set of common, edge-case, incomplete, and invalid inputs with expected outcomes for repeatable bot or agent testing.
How it works
From real cases to a clear pilot plan.
Select
Narrow the list to one process with an owner, sufficient frequency, and a clear business problem.
Map
Walk through the main path, exceptions, data, systems, decisions, and current bottlenecks.
Measure
Record the current results and define test cases, risks, and approval points.
Decide
Recommend a focused pilot, prepare the data or process first, or clearly explain why the project should not proceed.
An assessment does not promise return on investment or a predetermined result. Benefit and cost estimates depend on available evidence, real volume, integration constraints, and the outcome of a controlled pilot.
Related topics
Move toward a solution after the assessment.
Questions
Before starting the assessment.
Do we need to know which model or platform we want?
No. The assessment starts with the process, data, and required controls. A model, agent framework, or platform should be selected only after the real requirements are understood.
What do we need to prepare?
A process owner, a short description of the current workflow, and a few real, anonymised examples are enough to begin. If starting measurements do not exist, the assessment will identify what to track first.
Does every assessment lead to an AI pilot?
No. The right conclusion may be traditional automation without AI, data preparation, a process change, or a decision that the current case is not valuable or safe enough to pilot.
First step
Send us a process that takes too much of your team’s time.
Include who performs it, how often, which systems it uses, and where it usually stalls. We will reply with a proposed assessment scope.