Pilot evaluation

Prove the workflow before broad deployment

An eight-week, fixed-scope evaluation on your own transformers: approved evidence in, agent-drafted analyses reviewed by your engineers, and an evidence pack your organization can take to a decision.

Request a pilot See the sample report

Defined pilot scope
Evidence-readiness checklist
Review workflow and sample report
Conservative value framework

Pilot timeline

Eight weeks from scope to decision

Three duration bands, each closing on a deliverable you keep — no open-ended discovery phase.

Weeks 1–2

Evidence intake

Asset list, approved historical records, and source files land in the workbench. Quality gates flag gaps and duplicates while the evidence model is configured for your fleet.
Deliverable

Evidence intake report

What passed the quality gate, what is missing, and exactly what the pilot will analyze.
Quality-gated DGA trend
- H2
- CH4
- C2H4
- C2H2
H2 watch level Gassing event
Weeks 3–6

Agent drafts + engineer review

The agent drafts per-unit analyses with cited evidence. Your engineers approve, edit, reject, or escalate in weekly review cycles — every decision lands in the audit log.

Deliverable

Reviewed work packages

Engineer-signed work packages for the units that need action, formatted for your CMMS.
Weeks 7–8

Evidence pack + decision

Findings, rationale, and the full review history are assembled into the final evidence pack, with a value memo scoped to your success criteria.

Deliverable

Evidence pack + value memo

A procurement-ready pack: condition evidence, AI rationale, the sign-off trail, and a conservative value summary.

Health index

82 / 100
Good

Poor 0–40 Fair 40–70 Good 70–100

Deliverable previews show synthetic specimen data, not customer results.

Scope

What the pilot covers — and what comes later

A pilot proves the workflow on a bounded population before any enterprise commitment.

Pilot scope compared with full deployment, by dimension.
Dimension	In the pilot	Full deployment
Transformer population	A focused population where historical evidence exists	The full fleet, onboarded in criticality order
Integrations	Approved historical records and controlled datasets — no live integrations	Live CMMS, EAM, and historian connections, evaluated after the workflow proves useful
Decision workflow	One decision workflow with named review owners	Additional workflows and teams, added one at a time
Outputs	Evidence pack, reviewed work packages, and a value memo	Continuous review cycles and fleet-wide reporting
Deployment	An engineer workstation inside your perimeter — no OT connection	Deployment shaped by your IT and security review

Success metrics

Measure decision quality, not hype

Evaluation focuses on preparation time, evidence completeness, review clarity, traceability, and usefulness for lifecycle and sustainability planning.

Preparation time: Time to assemble a review-ready diagnostic file, measured before and during the pilot.
Evidence completeness: Share of scoped units with a complete, quality-gated evidence record.
Health-index clarity: Whether engineers can explain each unit's score drivers from the workbench alone.
Traceability: Every recommendation resolves to its source records in one step.
Engineer confidence: Reviewer assessment of the draft-and-review workflow after each cycle.
Lifecycle usefulness: How useful the outputs prove for lifecycle and sustainability planning.

Cross-industry context from the U.S. DOE FEMP O&M Best Practices Guide — what functional predictive-maintenance programs have averaged in independent surveys:

25–30% ¹

reduction in maintenance costs

35–45% ¹

reduction in downtime

10x ¹

return on investment

Engagement model

What each side provides

A pilot is a joint effort with a deliberately small footprint on your team.

Your team provides

Named engineer-reviewers for the weekly review cycles
Approved historical records and asset identifiers
The decision workflow and success criteria to evaluate
A workstation inside your perimeter

GridAPM provides

Workbench deployment and evidence-model configuration
Agent-drafted analyses for every scoped unit
Facilitation of the weekly engineer review sessions
The evidence pack, work packages, and value memo

Get started

Two ways to start

Request a pilot

Tell us about your fleet, your evidence, and your timeline — we reply with a scoping call.

Go to the pilot request form

Build your pilot brief first

Answer a short readiness questionnaire and paste the structured brief into your request.

Open the pilot brief builder

Pilot FAQ

Questions before the first evaluation

How many transformers should a pilot include?

A focused first pilot starts with a small population where historical evidence exists and the team has a clear review workflow to evaluate.

Do we need live integrations for a pilot?

No. A pilot can begin with approved historical records and controlled datasets. Live integrations can be evaluated after the workflow proves useful.

What does GridAPM claim during a pilot?

Decision support with engineers holding final authority. The pilot measures preparation time, traceability, and review quality on your own evidence — it makes no claims of guaranteed failure prevention, autonomous control, or quantified performance until your evidence supports them.

How long does a pilot take?

Eight weeks on a fixed scope: evidence intake in weeks 1–2, agent drafts and engineer review cycles in weeks 3–6, and the evidence pack and decision in weeks 7–8.