The operating layer for distributed infrastructure.
GridKeep turns fragmented infrastructure signals into governed, verified action - so operators can run more sites with less ambiguity.
Distributed infrastructure is scaling. The operating context isn't.
The assets are distributed, but the decisions are still stitched together across dashboards, messages, spreadsheets, and field teams.
Signals live in different systems
Telemetry, WhatsApp, SMS, spreadsheets, and field notes describe the same site without sharing one operating picture.
Triage happens after consequence
An asset can drift for hours before anyone knows whether it is a real failure, a sensor fault, or a local power event.
Proof gets lost between teams
A dispatch may happen - but the operator cannot easily show what changed, who confirmed it, or whether it recovered.
The problem is not a lack of data. It is the distance between what the data says and what a team can safely do next.
GRIDKEEP THESISGridKeep closes the gap between signal and action.
One governed control loop adds context, control, and proof to every consequential decision.
See the signal before it becomes a service failure.
Bring telemetry, field notes, messages, and asset context into one operating picture.
One operating layer. Four surfaces.
Command centre, digital twin, decision copilot, and evidence ledger - connected by one accountable control loop.
Operational truth, in one view.
The asset base is growing. The control plane is still missing.
A bottom-up beachhead gives GridKeep a measurable path from one paid proof to recurring operating software - without pretending a large market number is traction.
Zimbabwe beachhead
at beachhead pricing
not a forecast
Assumption-driven, not claimed. Site counts, pricing, and conversion rates are hypotheses to validate in a paid pilot.
More distributed assets create more coordination cost.
Mining, energy, health, and water operators already manage signals across different tools. The missing layer is the accountable handoff between what changed and what someone should do next.
Start where a false positive is expensive.
Our initial wager is mining operations: distributed sites, costly downtime, and complex field handoffs create a sharp need for governed action. The same loop expands into energy, health, and water.
A haul route slows. The operator sees the constraint before production slips.
GridKeep correlates equipment telemetry, power, connectivity, shift notes, and maintenance history before escalating a field response.
Land with a paid proof. Grow by site.
The commercial hypothesis is simple: start with a bounded deployment, then turn the same operating context into recurring software across more sites and asset classes.
We sit between the signal and the consequence.
Existing systems remain useful. GridKeep is the governed workflow and evidence layer above them, so replacing the system of record is not the entry requirement.
Land in mining. Expand through networks.
The first buyer is the mining operator; expansion follows through energy, health, public-program, and capital networks already accountable for the outcome.
Run more sites without adding another layer of confusion.
The first buyer is accountable for production, assets, exceptions, owners, and proof - and needs a shared operating picture before downtime compounds.
The product surface is ready. The proof is next.
GridKeep has enough product surface area to earn a pilot. It does not yet have customers - the next milestone is a signed design partner and the first live deployment.
Demo-ready / not production. Current operating values are synthetic and no live telemetry, messaging, CRM, or authentication integration is claimed.
Cyprian Aarons
Founder & Principal Engineer at Topiax. Builds reliable AI systems, secure data workflows, and production-ready agent architecture.
Turn the thesis into live proof.
We are looking for design partners and a bounded deployment round to connect the loop to real sites, real teams, and real outcomes.
25 design-partner sites
3 failure classes
90 days to learn
Clear by design. Synthetic operating values are labeled. Pricing and milestones are hypotheses. No customer logos, ARR, or production uptime are claimed.