Method
Fast, but verifiable at every step.
Understand the real process, reduce before building, test the error cases, then operate. A VELLORNE project moves forward through concrete results, not presentations.
- Contact person
- One, from the first conversation to operations.
- Commitment
- Scope and budget approved in writing before building starts.
- Duration
- From a few weeks for a targeted flow to several months for a larger system.
How it works
Six steps, each with a result.
- 01
First conversation
You describe the process and what is costing you time. We tell you frankly whether automation makes sense.
Result: a clear opinion and, if relevant, a proposal for an analysis.
- 02
Analysis
We observe the process as it really happens: inputs, people involved, exceptions, volumes, tools, access rights and sensitive data.
Result: process map, risk points, recommendation.
- 03
Design
Minimal architecture, critical rules, human approval points, hosting choice and test plan. AI is only used if it brings something measurable.
Result: scope, architecture and budget approved in writing.
- 04
Build and testing
Integrations, rules, approval interfaces and logging. Error cases are tested as thoroughly as the normal case: system unavailable, duplicate, missing data, insufficient permissions.
Result: a tested system, in an environment separate from production.
- 05
Parallel running
The system runs alongside manual processing, on your real cases. Results are compared and rules adjusted before the switchover.
Result: a go-live decision based on facts.
- 06
Go-live and operations
Gradual roll-out, monitoring, alerts and backups. Documentation handed over and knowledge transferred.
Result: a system in production, documented and monitored.
Go-live conditions
A demonstration is not enough to go into production.
Four conditions must be met before a system is declared ready.
Tested
The normal case and the error cases have been run and checked.
Traceable
Every run leaves a trace: identifier, steps, decisions, result.
Reversible
The flow can be paused and resumed without loss or duplication.
Documented
How it works, operating procedures and contacts are written down and handed over.
Deliverables
What you receive.
A system your team understands, and that another provider could take over.
- The production system and access to it.
- A description of the automated process, the rules and the approval points.
- The technical documentation: architecture, integrations, dependencies.
- The operating procedures: backup, recovery, emergency stop.
- The list of third-party services used and where the data is located.
Your role
What we expect from you.
- A point of contact who knows the process.
- A few real examples, anonymised if necessary.
- Limited technical access that you can revoke.
- Feedback during parallel running.
Method
A short project, with verifiable results at every step.
Start by describing the process. Everything else is built from there.