Home
Solent

One AI platform, built to plug into everything you run.

AxonGate connects your company's own documents and know-how to AI — so your team can ask a plain question and get a direct answer, instead of searching through files. It's the same platform other Solent products build their AI features on, and it's available as its own platform for your business too.

  • 76+ file types
  • 100+ languages
  • Answers in under a second
  • Every action logged
AxonGate platform interface showing a plain-language answer generated from company documents
AxonGate answersVerified against 3 documents

Your Q3 return policy allows...

Answered directly from company files · nothing left the building

  • Ask, don’t search

    Type a question about your own documents and get a direct answer, not a list of files to open.

  • Your data stays yours

    Every customer’s documents and answers are kept completely separate — nothing is ever shared between companies.

  • Works with what you already have

    Over 76 file types, from PDFs and spreadsheets to scanned paperwork and recorded audio — no manual conversion needed.

  • Runs where you need it

    Fully in the cloud, or entirely on your own servers, with full control over where your data lives.

Maintenance technician scanning an equipment tag while viewing the Kirpi work-order queue on a rugged tablet

Kirpi AI CMMS

Active maintenance intelligence that speeds up the crew's decisions.

Kirpi manages assets, work orders, and maintenance procedures with AI-supported analysis, suggestions, and controlled automation — turning the full request-to-analysis lifecycle into one real-time platform. AI accelerates the crew's decision; it never replaces it.

  • Asset/equipment management; tag and P&ID instrument tracking on a facility → line → machine → component hierarchy
  • Work-order management with fault log/track — 7 order types, 4 priority levels, sub-orders, automatic work timers
  • Preventive, corrective, and repair maintenance; meter thresholds auto-open work orders (e.g. oil change past 500 operating hours)
  • Shift tracking & handover; maintenance procedures and step-by-step checklists
  • Stock, parts, labor, and cost integration — min/max thresholds, parts auto-deducted from work orders
  • Web + mobile synchronization; real-time performance visibility (mean repair time, SLA and PM compliance)

The maintenance lifecycle

  1. 01

    Request

    A fault or request enters through any of 7 intake channels and becomes trackable in one click.

  2. 02

    Work order

    The request becomes a work order with a type, priority, tasks, checklist, and parts list.

  3. 03

    Execution

    The crew works the order with automatic timers, shift handover, and procedure checklists.

  4. 04

    Completion

    True cost — labor, parts, and external expenses — is computed automatically at close-out.

  5. 05

    Analysis

    Every action lands in an immutable audit trail feeding mean-repair-time and compliance dashboards.

  • OK

    Meaning
    Within the response/resolution window
    Flag
  • Warning

    Meaning
    Most of the window consumed
    Flag
  • Breach

    Meaning
    Limit exceeded
    Flag
    Safety-hazard and downtime tickets specially flagged

Before Kirpi, with Kirpi

  • Work orders lost in paper and Excel

    Digital work orders — tracked from opening to close-out

  • Failures strike unannounced

    Automatic preventive-maintenance planning

  • Stock either short or overflowing

    Smart inventory with min/max thresholds

  • True maintenance cost invisible

    Labor + parts + external expenses computed automatically

  • "Who did what, when?" unclear

    Immutable audit log

AI triage flow showing a request moving from intake to classification to routing

Zeplin Desk — AI-Supported Request & Workflow Automation

The same front door, now classified and routed by AI.

Zeplin Desk makes enterprise support and workflow processes intelligent with AI-supported reception, request classification, and automatic routing. Incoming requests are analyzed with AI, assigned to the right unit, and repetitive work is bound to automation — so the desk answers faster while the team works the exceptions. Full product detail lives on the Call Center & CX page.

  • AI request reception, classification, and prioritization
  • Department-based workflow, approval, and task automation
  • SLA tracking and performance analytics
  • Automatic request intake from email, web, and API channels
Incoming signals converging into a central AI layer and emerging as structured, prioritized queues

Enterprise AI Capability Layer

One shared intelligence, embedded across the portfolio — and adaptable to yours.

The same AI capabilities are embedded in products across the Solent portfolio: object detection and signal fusion in autonomous field systems, smart camera analytics in XPoint, content analysis and generation in SoruLab, and the multilingual AI guide in KBIS. The layer deploys in edge and central architectures — and it can be adapted to custom enterprise projects.

  • Vision: person / vehicle / UAV / smoke-flame detection and classification
  • Natural language: multilingual assistant and content analysis
  • Decision support: rule engines, anomaly detection, and early warning
  • Edge and central AI architectures

See → understand → decide

  • See

    Vision: person / vehicle / UAV / smoke-flame detection and classification.

  • Understand

    Natural language: multilingual assistant and content analysis.

  • Decide

    Decision support: rule engines, anomaly detection, and early warning.

  • Deploy anywhere

    Edge and central AI architectures, adaptable to custom enterprise projects.

The layer, already in the field

Proof in production — SoruLab

SoruLab shows what the layer does under real load. Five AI operations — difficulty assessment, error correction, twin-question generation, answer management, and topic mapping — improve entire question pools in bulk, processed in parallel with automatic retry on failures. Every AI change waits in a workspace with an old/new diff until an expert approves it: nothing publishes without human sign-off, and every change is versioned and reversible. AI proposes; the decision stays human.

  1. 01Difficulty assessment
  2. 02Error correction
  3. 03Twin-question generation
  4. 04Answer management
  5. 05Topic mapping

Who feels the difference

Two roles, two ways the same AI layer accelerates decisions.

Plant maintenance director

Downtime found before it happens.

Pains

  • Unplanned downtime discovered by breakdown
  • Maintenance cost invisible until year-end
  • Knowledge that leaves with the shift

Gains

  • Preventive plans and meter thresholds open work orders before failures
  • True cost per work order computed automatically
  • Shift handover and an immutable history keep knowledge in the system

Digital transformation lead

AI that ships, not AI that demos.

Pains

  • AI pilots that never leave the lab
  • Vendors promising replacement instead of adoption
  • Every product carrying its own siloed "AI"

Gains

  • One capability layer already proven inside shipping products
  • Edge or central deployment to fit existing infrastructure
  • Expert-approval workflows that make adoption auditable

Enterprise AI Software figures

7Work-order types

Work-order types

7Request intake channels

Request intake channels

5SoruLab AI operations

SoruLab AI operations

4Vision classes detected at the edge

Vision classes detected at the edge

Boardroom table with a site analysis app open on a tablet, blueprints, and a city skyline at dusk

Put an AI layer under your experts.

Bring us a process where decisions are slow and evidence is scattered — our engineers will show how vision, language, and decision support accelerate it, with your experts holding the approval key.