Read-only first
No automatic control changes and no self-created work orders.
Ireland · Pharmaceutical manufacturing · Customer discovery
Predumi is testing whether Irish pharmaceutical plants need a vendor-neutral path from condition evidence to approved maintenance work and post-work verification.
A pilot, not a promise
The category already has sensors, dashboards, APM suites, remote analysts, and AI claims. The starting point is one unresolved workflow—not another horizontal platform.
No automatic control changes and no self-created work orders.
First probe: 10–20 intermittent AHU fan/motor trains—not a claim to monitor the whole plant.
Test the existing route, vendor, BMS, historian, CMMS, and installed sensors before adding another system.
A condition-monitoring specialist reviews diagnostic recommendations.
Source, context, rationale, reviewer, change, and outcome remain traceable.
The final deliverable can conclude “do not deploy” if the economics or evidence fail.
Conditional 8–12-week design-partner sprint
A proposed fixed scope only after a pharma site identifies an unmet criterion in the incumbent/current process. Price and acceptance criteria remain discovery questions.
Agree the current process, unmet criterion, asset criticality, failure modes, intended use, data, responsibilities, and head-to-head success test.
Assess data quality, operating context, current condition, existing maintenance workflow, and a value baseline.
Monitor read-only, log evidence, and have recommendations reviewed before the plant decides any action.
Where maintenance occurs, compare post-work condition and record what changed, what did not, and what was learned.
Deliver an honest expand, revise, or stop recommendation—with limitations and unresolved risks stated.
Trust starts with boundaries
Predumi is at customer-discovery stage. Accuracy theatre and premature compliance language would make the company less credible, not more.
For Irish plant practitioners
No confidential data. No product theatre. Just the current process, the last real failure, and what would make a small pilot useful—or pointless.
We are looking for problem interviews and one paid design partner for a narrow, read-only pilot after the evidence supports it.
Request a discovery conversation