01 · Authority
Make permission explicit.
Define who may review, approve, deny, override, or stop consequential actions—and under which conditions.
Fieldborne Tech / EU AI Governance
Make AI-assisted decisions reconstructable.
Human authority. Traceable evidence. Reviewable decisions.
The Fieldborne EU AI Act Governance Pilot applies explicit authorization, evidence lineage, bounded testing, and policy controls to one consequential AI workflow so the resulting decision path can be reviewed afterward.
Pilot boundary. The pilot supports governance implementation and conformity-readiness evidence for selected applicable EU AI Act requirements. It is not legal advice, certification, CE marking, regulatory approval, or a guarantee of compliance.
CP1Bounded engineering scope complete
155Tests completed
0Recorded failures at Engineering Gate
L-01Closed within original bounded review scope
Inspect the bounded record →
The governance problem
Evidence can exist without the decision being reconstructable.
Model logs, policies, tickets, approvals, screenshots, and spreadsheets can all exist independently. The harder question comes later: who authorized this action, based on which evidence, under which defined controls?
Fieldborne is built around making that decision path reviewable instead of requiring a new forensic reconstruction every time it is challenged.
Governance approach
Authority, evidence, and reconstruction stay connected.
The pilot is intentionally bounded to one workflow so the authority model, evidence obligations, test conditions, and acceptance criteria can be defined before consequential execution.
02 · Evidence
Bind decisions to their basis.
Preserve relevant system state, evidence, actor identity, model/provider context, timestamps, exceptions, and outcomes.
03 · Reconstruction
Make the path reviewable later.
Connect requests, authority, evidence, action, and result so a reviewer can inspect the chain without rebuilding it manually.
01Define the boundary
Freeze intended purpose, actors, permitted sources, system boundary, and evidence scope.
02Map selected controls
Translate applicable governance requirements into reviewable controls and evidence obligations.
03Require authorization
Route defined consequential decisions and actions through accountable human approval gates.
04Execute bounded tests
Run predeclared reliability, policy, and TEVV checks against fixed acceptance criteria.
05Preserve traceability
Capture decisions, actors, model/provider identity, timestamps, exceptions, and audit evidence.
06Review and remediate
Preserve failures and limitations, assign remediation, and close with a bounded evidence record.
Engineering evidence
Bounded proof, not generalized claims.
Fieldborne separates what was actually reviewed from what has not yet been established.
Recorded at the later bounded Engineering Gate exact-head review.
Recorded in that same bounded Engineering Gate run.
Closed within the defined bounded Engineering Gate scope.
Windows execution/UI evidence closed within its original bounded review scope.
Claim boundary. These engineering records do not establish production readiness, security certification, regulatory compliance, EU AI Act conformity, institutional endorsement, market validation, QCE scientific validation, or whole-product validation.
Founding pilot
One consequential workflow. One bounded evidence program.
The engagement is structured to produce a reviewable governance record around a defined workflow—not an unrestricted deployment claim.
Closeout objective
At closeout, the bounded record should show who authorized what, based on which evidence, under which defined controls—along with recorded exceptions, limitations, testing outcomes, and remediation.
1bounded AI workflow
6–8 weekstarget duration
€15,000fixed founding-pilot fee
50%due at kickoff after approval
Deliverables
What the customer receives
Scoped workflow/system boundary, control matrix, risk register, authorization records, audit/traceability package, TEVV results, remediation record, and final pilot evidence report.
Evidence domains
What the pilot examines
Human oversight, logging/audit integrity, risk/control mapping, robustness/cybersecurity, technical documentation, QMS/change control, post-market monitoring design, and supplier/model governance.
Best fit
Use Fieldborne when the decision path matters.
- AI governance, risk, assurance, or compliance teams
- Consequential AI-assisted workflows with accountable reviewers
- Organizations that need reconstructable authorization and evidence
- Teams willing to define one workflow and fixed acceptance criteria
Not this engagement
Keep the boundary explicit.
- Unrestricted production deployment
- Certification, CE marking, or regulatory approval procurement
- Legal advice or guaranteed compliance
- Generic AI consulting without a defined workflow boundary
Commercial path
Qualification first. Commitment only after scope approval.
The public website starts qualification. It does not create a contract, authorize a charge, or permit access to production systems.
No public checkout. An intake submission does not authorize payment. Approved engagements use invoice-based billing under the executed agreement.
Trust & engagement
Know the boundary before you submit anything.
Public materials describe the workflow, bounded engineering record, pilot methodology, and separate QCE research framework. Restricted or sensitive material should not be submitted unless a specific engagement explicitly authorizes it.