​

VH3 AI Field Service Intelligence

​

Start with the working day

At 07:30, a field service team needs a clear view of the day. Which engineers have the right experience for each job? Which sites need attention before the first visit? Which customer has waited too long for a fix? Which account needs a call before a small issue becomes an escalation? VH3 AI brings those answers together. It pulls in the records your operation already creates, connects each job to the right customer, site, engineer, and outcome, then keeps that picture current as work changes. The morning brief arrives with the relevant history. A coordinator finds similar faults without guessing the words an engineer used. An account manager enters a review with evidence instead of spending two weeks rebuilding a report. The platform removes the manual grind around operational knowledge. Your team spends less time joining spreadsheets, searching old notes, and chasing updates. It spends more time deciding what to do next. Every new job adds to the past job history in your account, so tomorrow’s brief starts with more evidence than today’s.

​

The layer behind that relief

VH3 AI is the intelligence layer alongside your field service management system. It connects the systems your operation already runs, keeps records current, and gives the same operational picture to people, automations, agents, and applications. One field service model. The layer maps every incoming record to the concepts that drive field service work. Jobs connect to people, sites, customers, equipment, outcomes, and history. The model keeps those meanings consistent when different source systems use different fields or names. Hybrid retrieval. Hybrid retrieval combines relationship traversal, vector search by meaning, structured job records, and current monitoring results. A request can follow the graph from a customer to its sites, search fault descriptions by meaning, read the enriched outcome fields, and include active risk signals before an agent writes a response. The system can find a similar fault even when two engineers used different words. Prepared context. The enrichment pipeline classifies each job once, resolves its entities, and stores the result for every downstream consumer. Reports, briefings, automations, and agents read the prepared record instead of extracting the same facts again. In these technical guides, the substrate names the decoupled, persistent operational knowledge graph and vector index in the customer’s account. VH3 AI stores enriched records, relationships, search representations, cases, and operating rules there. An LLM call performs ephemeral inference in a context window and returns a response. That temporary inference reads the substrate but does not replace it.

​

Work that moves while you are doing other things

VH3 AI runs deterministic background sentinels against the operational record. Each sentinel evaluates a defined condition on a schedule or after relevant data changes. It uses graph data and stored fields at detection, so the evaluation does not require an LLM call or a token budget. When a threshold is met, the platform records the result, links the supporting jobs, opens a case when configured, asks Connie for a draft brief when needed, and routes the work to the team that owns the customer or region. That sequence gives an operator a queue with evidence already attached. Sentinels find change. Cases hold the work. Teams own the response. Connie supplies a cited draft when a person needs a narrative. People steer and verify the outcome.

​

Why this architecture matters

Raw-record chat can answer a narrow question. Operational work asks an agent to cross-reference history, compare hundreds of thousands of jobs, assess a pattern, and produce an answer that someone can act on. Those tasks need a maintained operating context with resolved entities, current relationships, structured outcomes, and clear scope. VH3 AI had production AI agents running against real operations by late 2024. Processing hundreds of thousands of jobs across multiple large clients exposed the limits of search over raw records. The platform now prepares the operating context at ingest and makes it available to every downstream consumer. The intelligence layer combines relationship mapping, meaning-based search, structured records, and continuous monitoring into one synthesis path. An agent reads connected context. It does not rebuild the operation from fragments on every request. Read how the intelligence layer works →

​

One foundation for every consumer

A team might use a shared dashboard today, a custom internal tool next quarter, and an agent-powered interface after that. Each consumer can read the same enriched, connected record. Reports, briefings, automations, custom apps, and AI agents work from one current operating picture. VH3 AI includes authentication and user management, 28+ workflow automation templates, and guides for connecting coding tools and agents to the intelligence layer. Builders can start from working infrastructure and add the workflow their operation needs.

​

AI costs should work like a utility

VH3 AI separates platform work from model inference. The platform fee covers ingestion, enrichment, entity resolution, relationship mapping, monitoring, rules, third-party context, and the AI in the platform’s own tools. Pricing follows job volume and capability tiers. It does not follow headcount. BYOK means that your organisation supplies the model provider key for Connie and other model-backed agents. The selected provider bills each inference call to that key. You can see the call cost, choose the model, and change the model without a VH3 AI token margin. Deterministic discovery, graph traversal, aggregation, vector search, and sentinel evaluation reuse prepared data and do not invoke an LLM at detection. The enrichment pipeline performs its interpretive work once at ingest. Repeated reads share that work, which lowers the model cost per query as more people use the same foundation.

​

Freedom to build and leave

The intelligence belongs to your organisation. Query it through the API, consume it through automations, connect it to your own models, or build applications on top of it. Dedicated, isolated infrastructure is available on Enterprise plans. The platform does not charge for unused seats or gate exports behind support tickets. If you leave, your enriched operational data comes with you. We keep customers because the platform earns its place in the operation.

​

What the API gives you

The VH3 AI API gives one access point to the outputs of the intelligence layer.

​

How it works

​

The operating knowledge that stays with the business

Field service businesses lose institutional knowledge when experienced people leave. An engineer takes fault patterns with them. A contracts manager takes the context behind difficult client relationships. A new joiner spends months piecing together the operation from conversations, inboxes, and tacit knowledge. VH3 AI keeps that knowledge in the maintained operational picture. Every job, outcome, pattern, and relationship adds evidence to the account. A new engineer receives a pre-visit briefing with faults seen at the site, the fixes that worked, and the people who attended. An operations manager asks about the past five years and receives cited answers. An account manager retrieves customer history, communication patterns, and outstanding risk in minutes. The platform supports onboarding, offboarding, and upskilling. People move on. The operational knowledge remains available to the team.

​

Supported FMS platforms

VH3 AI ingests and normalises data from the FMS systems that connect to it. BigChange provides the live production integration today, including ingestion, polling, discovery, enrichment, and entity resolution. Joblogic, Simpro, Jobber, Uptick, ServiceNow, Commusoft, and other FMS connections remain on the roadmap. Organisations that run more than one system can bring those records into one intelligence layer as each connection becomes available.

​

Next steps