Platform

Connect site signals, decisions and field response.

Gen318 brings cross-brand monitoring, AI-assisted investigation, maintenance and reporting into one operating record for mini-grid and C&I teams. Follow an issue from the first signal through the response and the evidence it leaves behind.

Explore real console screens →
Gen318 · fleet illustrationSample · not customer data
Connected
SITES ONLINE
Review
COVERAGE
Open
OPEN INCIDENTS
Per site
DIESEL BASIS
Fleet load · 24h Actual Forecast
Incidents · investigation
ADAMAWA-07CRITICAL

Inverter drift — inspect the supporting evidence

09:41 WAT
KADUNA-12WATCH

Battery cell divergence above baseline

08:17 WAT
NIGER-03PLAN

Soiling indicator — review cleaning conditions

06:55 WAT
Connected
Signals and field response
Local
Edge buffering
Reviewed
Operational decisions
Traceable
Reporting evidence
Control room

One place to pick up the next decision.

An operator investigates an issue. An O&M lead coordinates the work. A performance reviewer follows the resulting evidence. Each role sees its authorised workflow; investor access is through assigned portfolios, not every operational screen.

Sample console

GEN
318
Gen318 / Fleet

Fleet Dashboard

Start with your highest-risk sites, then grow into a single operating view across solar, battery, and diesel — every site, every signal, one cadence.

Operational domains

View
Active···
Real-time Monitoring
Active···
Ask & Dispatch Planning
Active···
Predictive Maintenance
Active···
Performance Reporting
01 · Architecture

From panel to decision, one conductor.

Supported cloud adapters and commissioned edge nodes feed a shared record. Local buffering depends on power, storage and configuration. Remote views remain stale while disconnected; physical control requires separate enablement.

SITE HARDWARE
Victron · SparkMeter · Huawei · Growatt · Modbus
EDGE NODE
Local inference · queued synchronisation
FLEET BRAIN
Specialist analysis · evidence · model evaluation
YOUR TEAM
Console · technician PWA · permissioned reports
02 · Module 1 · Monitor & predict

Related signals in. A clearer response out.

Group related alerts, apply suppression and route the issues that need attention. Inspect the evidence behind impact estimates and suspected causes. Delivery and trust coverage answer different questions; both matter.

Detect
Signals and alerts

SoC fluctuations, voltage dips that self-resolve, comms blips from 2G — the noise that makes operators stop reading dashboards.

Group
Related signals together

A battery fault overworks the inverter, starts the genset, spikes the meter — that's one incident, not four alerts.

Review
Prioritised operator attention
ACT NOW
THIS WEEK
THIS WEEK
PLANNED
PLANNED

Review ranked incidents with the impact basis and missing evidence visible. This diagram illustrates the workflow, not a measured reduction.

Build toward earlier failure warnings.

Demand forecasting can use market days and seasonal context; battery and inverter models surface degradation for review. The target is enough lead time to turn emergency downtime into planned work. Live-site precision and lead time still have to be measured.

  • Forecast bands, not point guesses — confidence shown honestly
  • Remaining-useful-life estimates with evidence limitations
  • Inverter fault indicators for investigation
  • Recorded maintenance findings for model evaluation
Demand forecast · example
BANDNOT MEASURED ACCURACYACTUAL / FORECAST / UNCERTAINTY
Battery health · estimate
84%SoH
Remaining useful life412 days
Cell divergence+1.8% / wk
Estimation methodMODEL BASIS

Estimates, not observed customer results. Review model confidence and coverage before deciding on maintenance.

14:22OFFLINE · QUEUED
Work order
WO-0142KADUNA-12
AI diagnostic context

Battery string B2 — cell divergence, thermal stress pattern.

REVIEW EVIDENCEPARTS: CHECK AVAILABILITY
Step 4/7 · Work — was the AI prediction correct?
Record the observed condition
Record work and parts used
Attach the supporting photos
Location evidenceSYNCS ON SIGNAL
03 · Module 2 · Field service

The work order arrives with its context.

Give the technician the issue, available diagnostic context and assigned work. Cached jobs support queued field findings and photos offline. Keep the technician record distinct from later evidence assessment and review.

  • Assignment and technician coordination
  • Parts demand and depot sourcing where stock evidence exists
  • Field findings, parts used and supporting photos
  • Cached work and queued updates through interrupted connectivity
  • Recorded causes can be matched to predictions for evaluation

Next, follow the completed job into evidence assessment and post-repair review.

Parts and depot sourcing

Check the part before planning the visit.

Assess demand, inspect site and assigned-depot stock, and decide whether the next step is a transfer or a purchase. Missing stock records remain unknown, so absence of a record is never treated as an empty shelf.

  1. Demand

    Review the forecast and the consumption or equipment history supporting it. Insufficient evidence stays visible.

  2. Availability

    Check recorded site stock and the assigned depot. Depot availability is separate from allocation to a site.

  3. Transfer

    Track dispatch, in-transit stock and receipt, keeping the sourcing decision connected to actual stock movement.

Maintenance record · completion to reviewWorkflow · not a product screenshot or customer result
  1. 01Reported

    Field completion

    Technician

    Keep the work and its evidence together.

    Record the cause, action, parts and photos against the job. Offline updates queue on the device and sync once the server accepts them.

    Record integrity

    A matching record hash proves the sealed content has not been altered since the technician filed it.

  2. 02Gaps visible

    Evidence assessment

    Evidence processing

    See what supports the repair record.

    Inspect the evidence assessment and missing items. Later evidence can receive a new assessment without rewriting the sealed record.

    Evidence quality

    A zero score means an assessment ran and found no qualifying support — which is not the same as a record nothing has assessed yet.

  3. 03Window matters

    Post-repair findings

    Configured evaluation

    Look at what happened afterwards.

    Check whether the monitored signs of the fault cleared and whether it returned. See how each finding was assessed, over what period, and how much expected data arrived and passed validity checks.

    Observed outcome

    A completed evaluation is not the same as a successful repair, and the result tells you which one you are looking at.

  4. 04Review or dispute

    Supervisor review

    Authorised owner or admin

    Keep the reviewer and their basis on record.

    See who reviewed the repair record, what they considered and whether they accepted or disputed it.

    Human review

    The review retains its reason and the version inspected. Unreviewed records remain identifiable; the underlying assessment stays separate.

When the follow-up cannot establish a result
Field work: completed
The technician reported the action. This closes the administrative task.
Follow-up: not evaluable
The available evidence cannot support an outcome. Show the reason and coverage; do not record a successful repair.
Review: unreviewed or disputed
No review is not approval. A reviewer can dispute the record and retain their reason without rewriting it.

Record integrity, repair findings and supervisor review are separate checks—not independent certification. Follow-up evaluation and supervisor review require enablement, suitable evidence and role permissions; confirm availability during discovery.

04 · Module 3 · Compliance & finance

The report behind the performance review.

Prepare client uptime reports, generator environmental records, assigned-portfolio reports and programme exports from the relevant operating records. Review period, completeness and calculation basis alongside each output. Odyssey remains the programme partner.

  • State-aware rule context with recorded provenance and federal fallback
  • Client uptime, energy and reporting-period completeness
  • Generator log book and point-source inventory, each with its fuel-evidence grade stated
  • Portfolio revenue proxies clearly distinguished from collections
  • Reviewable fleet briefs linked to operating evidence
Programme evidence · illustrationOPERATOR REVIEW

Review records against the applicable programme requirements. This is not claim approval; automated milestone logic requires reconciliation before relying on current DARES terms.

CONNECTION EVIDENCEAttached
SOURCE RECORDS
DATA COMPLETENESSReview
PERIOD BASIS
Connection records attachedRECORDED
Consumption period recordedRECORDED
Site requirements need reviewRECORDED
Missing signal coverage needs reviewREVIEW
Site-specific programme requirements REVIEW
Review before export · finalisation preserves content

Rule packs record jurisdiction, cadence and provenance, including federal fallback. They support reporting review, not a legal determination. Investor reports have separate portfolio permissions.

Operational and commercial reporting, loss evidence and PCC settlement each prepare evidence for a person to check and sign. Review the inputs, outputs and human decisions required →

Specialist analysis across the workflow:Battery HealthInverter DiagnosticsDemand ForecastDispatch PlanningComplianceWeatherExplore the intelligence →
05 · Value statement

What your sites delivered. How sure each figure is.

A weekly and monthly statement of what each site delivered, in the units its sources measure. Every figure carries one grade: measured, modelled, estimated, declared, unpriced or insufficient. Figures are grouped by site type, never blended across them, and grid power displaced and diesel saved are never added together.

  • Naira only from a recorded contract or site tariff, or from diesel days above the confidence floor. Otherwise the figure is unpriced and shows its units. There is no platform default price.
  • A missing source is graded insufficient, with the reason it was withheld. It is never shown as 0.
  • A period's figure is only as strong as its weakest day.
  • A day stays provisional until its seven-day late-data window closes.
  • Availability is shown raw beside excused; delivered telemetry is never shown without trusted beside it.
Weekly value statement · illustrationPROVISIONAL
Energy delivered
kWh
MEASURED
Availability · raw beside excused
%
MEASURED
Diesel saved
Litres
MODELLED
Diesel saved
No ₦ · below confidence floor
UNPRICED
Grid power displaced
No ₦ · no tariff recorded
UNPRICED
Connections
No source recorded · never shown as 0
INSUFFICIENT
Provisional
Sent with the weekly summary. Hashed.
Final
Frozen, chained to the previous final, revisions listed.

The provisional statement is written once the period ends and is never rewritten. The final follows at least seven days later, once late data has arrived. It is frozen, hash-chained to the previous final, and lists what changed from the provisional. A final with no provisional before it says it was not compared.

Whose result is it? Each line says.

Every line sits on one rung of a claim ladder. The rungs are reported apart and never merged into one total.

  1. 01

    Site outcome

    What the site achieved over the period. It is the site's result and is never credited to Gen318.

  2. 02

    Recommendation adopted

    A recommendation someone approved. It is a count, never a saving: an approval is not proof that anything happened.

  3. 03

    Platform enabled

    An action whose execution was confirmed, by a verified equipment readback or by the generator running in the recommended state. It is valued over the action's own hours only.

The reader decides what a statement shows. Team members read internal and external lines; an investor reads external lines only, with no approval counts, expected values, internal notes or share-link opens.

Weekly value summary

The week's figures, for people who will not open a dashboard.

Every Monday morning, recipients with the weekly digest turned on get a short summary of last week's provisional statement. It leads with the figures that matter for their role, with the same grades and pairings as the full statement, says the day the week becomes final, and links to the full statement. Owners, admins and operators also get one suggested next step when something needs doing, and none when nothing does.

Email

Every role. Owners, admins, operators and viewers get the account's week, with goal progress first when goals are set.

SMS

Owners. Fitted to two standard text segments, with the statement link kept intact.

Web push

Technicians, in the field app. Their own week, over the sites of their own work orders.

Investors

One email per portfolio they belong to, carrying external lines only.

06 · The Guide

Help inside the console. It explains and suggests; it never changes anything.

The Guide is built from Gen318's own help cards. A card ships only once it has been reviewed and approved, and a card hidden for your role, your site types or a switched-off feature is never shown to you.

For you

What needs you, and why

A short ranked list of suggestions for your role, each with the reason it is there. During setup it becomes the onboarding checklist. Nothing pops up; a new risk item shows as a dot.

Prepared forms

The form, already started

“Do it” opens the right form with the values Gen318 already holds, marked “Filled by Guide”. It never supplies a value only you can declare, such as a diesel price. It can ask you for one and marks your answer “You entered this”. You always press Save.

Tours and previews

See it before you go

Tours spotlight each step on the page and never click for you; Follow along keeps a checklist of the steps. A preview shows a page in your terms before you open it, and never advertises a page your role cannot open.

Ask about Gen318

Questions about the product

Separate from Ask's fleet investigations, this chat answers how Gen318 works from the reviewed help cards, citing each one. A claim gate checks every paragraph for pages you cannot open, features named without their card, wrong settings, credit taken for a result and claims to have done something for you. An answer that fails twice is withheld, and you get the page's help card, or the closest one, instead.

Field app

Help for technicians

A Help sheet, tips on each job step and field-app tours, saved on the phone the first time they load with signal so they work without it. A job also shows the site's goals, as context and never as a personal target.

Goals

Targets on the value ledger

Owners set targets on figures Gen318 already measures, such as diesel use, contracted availability, the client SLA, DARES milestones or fault resolution time. Status is worked out every night from final statements and never set by hand, and no goal has a status before four usable final weeks. An admin's goal is a proposal the owner approves.

The Guide, its chat, goals, the value ledger and the weekly summary are each off by default and switched on per deployment.

07 · Fleet benchmarking

Compare operating records. Keep the basis visible.

Review fleet comparisons with their coverage, population and methodology. The illustrations below show comparison formats, not a Gen318 customer cohort or evidence of vendor superiority.

1NIGER-0394.2↑+2
2ADAMAWA-0191.8↑+1
3KADUNA-0589.4↓−2
4SOKOTO-1186.1·
5ADAMAWA-0781.7↓−1

Site comparison

Comparison rows, not customer scores. Actual comparisons need an explicit population, methodology and coverage; investor permissions are separate.

97.1%A
B
C
D
E

Hardware comparisons

Equipment groups A–E, not measured vendor performance. Compare like-for-like conditions and inspect the underlying evidence.

Comparison over a declared period

Trend and comparison line. This is not a published cohort or a measured Gen318 improvement. Coverage and calculation basis belong with the result.

08 · Decision and execution

Keep decisions distinct from outcomes.

A recommendation, an approved proposal, a command and an observed outcome are different records. Ask actions require enablement; dispatch actuation is separately gated and off by default.

01

Recommendation

Analysis proposes a response with evidence and limitations.

02

Approval

An authorised operator reviews the proposal. Approval records a decision, not proof of physical execution.

03

Execution

Equipment control follows a separate enabled path with current permission and safety checks.

04

Observation

Telemetry and field findings show what happened afterward. Completion alone does not establish success.

Follow the proposal, approval, command and observed response as distinct stages.

09 · Discuss your scope

Start with the task. Confirm the deployment fit.

Comply + Monitor

Discuss monitoring and reporting scope.

  • Supported equipment connections
  • Reviewable programme exports
  • Related alerts in context

Predict + Plan

Discuss asset risk and diesel evidence.

  • Dispatch recommendations
  • Asset-health investigations
  • Work orders and forecasts
Evidence review

Assure

Discuss the evidence your reviewers need.

  • Reporting basis and limitations
  • Field evidence and review
  • Finalised portfolio reports

Build + Assure

Discuss the commissioning handoff.

  • Equipment and signal mapping
  • Maintenance record continuity
  • Confirm scope and availability

Bring your equipment into one operating view.

Connect supported systems and organise their readings into a consistent fleet record. SparkMeter remains the metering and billing partner; Odyssey remains the programme platform.

Victron Energy
Inverters, BMV, MPPT via VRM API + Modbus
SparkMeter
Telemetry from SparkMeter / Koios
Huawei FusionSolar
String inverters & monitoring
Growatt ShineServer
Hybrid inverters
Odyssey
DARES compliance report export
Generic Modbus
Site-specific genset and BMS mappings

See the workflow relevant to your fleet.

Explore your team's workflow, the inputs it needs and the output you would use.

Request a demo