AI Voice for EV Charger Installers: Enquiry Triage

An EV charger installer's phone rings with the same opening line all day: "how much to fit a charger?" It sounds like a booking question. It almost never is. Before anyone can quote a price or send a van, the caller's property, parking, electricity supply and grant status all have to be established, and any one of them can change the job from a routine two-hour fit to a survey, a consumer-unit upgrade or a distribution network application.

The demand behind those calls is now structural, not a fad. Since 15 June 2022, Approved Document S has required new residential buildings in England with associated parking to have an electric vehicle charge point, so every new-build handover is a potential enquiry. On 1 April 2026 the government's EV chargepoint grant rose from £350 to £500 per socket, which changes the sums for the renters, flat owners and landlords who qualify. More EVs, more grant complexity, more calls, and each one needs qualifying before it is worth an installer's time.

That qualifying work is repetitive, rules-bound and easy to get wrong at 7pm when the office is closed. It is exactly the work an AI voice agent should carry, and exactly the work it must be scoped never to overreach.

This guide is shipped by the team behind Dilr Voice, enterprise voice AI built for regulated deployments. Or see DATS, our five-stage AI consulting system.

What can an AI voice agent actually do on an EV charger enquiry call?

An AI voice agent like Dilr Voice answers every EV charger enquiry, captures the caller's address, property type and parking, works out which grant route they fall into, screens the electrical supply for obvious blockers, and books a site survey when anything is uncertain. It logs the call to the installer's job system and escalates any decision that needs a qualified electrician. It handles volume and out-of-hours calls without quoting a price blind.

None of that replaces the electrician. It replaces the receptionist's triage sheet: the twenty questions that decide whether a call becomes a survey, a quick fit or a polite decline. Because the agent asks the same questions in the same order every time, the survey team arrives with a complete picture instead of a name and a postcode. For a busy installer, the voice AI agents sit in front of the diary, not inside the van.

Why is every EV charger enquiry really an eligibility triage, not a booking?

Because the price, the grant and the feasibility are all unknown until the caller is qualified. A charger fit for a renter in a flat with a shared car park is a different job, on a different grant, from the same charger at a detached house with a full consumer unit. The agent's first job is to place the caller in the right lane, and there are several lanes, not two.

The government funds EV charge points through five separate schemes, and which one a caller qualifies for depends entirely on their tenure and property. This is the single biggest reason an enquiry cannot be quoted from the phone alone. An owner-occupier of a single-family house with a private driveway, the caller most people assume is the easy win, is in fact eligible for no household grant at all.

The distinctions in that table, drawn from the government's own grant guidance, are precisely what a caller cannot self-diagnose and a rushed receptionist gets wrong. Getting the grant route right on the first call protects the installer from quoting a price that later has to be corrected. It is the same enquiry discipline we apply to availability quotes for self-storage operators, and it runs through the other AI voice guides for UK trades and services.

What does the agent need to know before it can quote or book a survey?

Enough to know whether the job is standard or not. The agent gathers the property type, whether parking is off-street or on-street, the location of the consumer unit relative to the parking, whether the supply is single or three phase, the caller's EV and connector, and their grant route. If every answer points to a routine install it can offer survey slots. If any answer is uncertain, it books a survey rather than guessing.

That last rule is the whole discipline. An EV charger draws a sustained load, and whether a property can take one depends on the existing main fuse, the spare capacity in the board and the earthing arrangement, none of which a homeowner can report reliably down the phone. So the agent is built to escalate to a survey the moment it hits an unknown, not to talk itself into a quote. The path from ringing phone to firm price looks like this.

Structured this way, the survey is booked with a purpose and the quote is anchored to something real. It is the same qualification logic we build into an AI operating model consulting engagement, and it is why the trade verticals we work in, from heating and boiler firms to glazing installers, all end up with an enquiry step in front of the diary rather than a booking form.

Do EV charger installs need DNO approval or notification?

Yes, and the agent needs to know which side of the line a job sits on. A standard single-phase domestic charger is usually a "connect and notify" job: the installer tells the local Distribution Network Operator after the work, under the Energy Networks Association's Engineering Recommendation G98. A larger, three-phase or vehicle-to-grid install is an "apply to connect" job under G99, where the operator must approve the connection before any work starts.

The agent should never promise a DNO outcome, but it can and should ask the questions that flag which regime applies, so the survey and quote account for it. The MCS guidance on notifying DNOs sets out the G98 and G99 framework that governs this. Where a caller mentions three-phase supply, a high-powered unit or vehicle-to-grid, the agent routes straight to a human, because those jobs carry a lead time the caller needs to hear about early. Building that routing rule is the kind of thing our AI execution office owns during a rollout.

Which rules and standards apply to a UK EV charger installation?

Several, and they do not all cover the same nations, which is why an agent scripted for "the UK" has to be careful. In England, Approved Document S governs charge points in new and materially changed buildings. In England and Wales, Part P of the Building Regulations covers electrical safety in dwellings. Across Great Britain, BS 7671, the IET Wiring Regulations, is the wiring standard, and the hardware itself must meet the Electric Vehicles (Smart Charge Points) Regulations 2021.

Those smart-charging rules are worth understanding, because they shape what an installer can legally sell for home and workplace use. The regulations require relevant charge points to be smart, and they draw the line at the cable: a non-smart cable is defined in regulation 3 as "an electrical cable which is a charge point but which is not able to send and receive information." An installer whose hardware pre-dates those rules, or a caller asking to reuse an old unit, is a case the agent should flag rather than book. The grant schemes add one more requirement on top: to claim any of the chargepoint grants, the work must be done by an OZEV-authorised installer, and Part P compliance in England and Wales is the safety obligation an installer demonstrates, through a competent-person scheme or building-control notification, not an optional badge. Naming who each rule binds, and where, is the difference between a compliant script and a liability, and it is the first thing we check in a design and compliance review.

Set against that regulatory picture, the scale of the transition is easy to underestimate. The public network alone reached 121,171 charging devices across the UK by 1 July 2026, of which 28,887 were rated rapid or above at 50kW and over, according to the Department for Transport. Home and workplace installs, the pipeline an installer actually works, are counted separately and driven by different rules, but the direction of travel is the same.

What must an EV charger AI voice agent never decide on its own?

It must never confirm a fixed installation price, guarantee grant eligibility, sign off electrical feasibility, or promise a DNO connection, because every one of those depends on a survey or an operator decision the agent cannot make. Dilr Voice is scoped to qualify, book and escalate, not to quote. The moment a call reaches a judgement about load, safety or a firm number, it goes to a qualified electrician.

This is not a limitation to apologise for; it is the design. A voice agent that quotes a price it later cannot honour costs more goodwill than a missed call ever did. The same boundary governs the trades where a safety line is legally fixed, such as the Gas Safe boundary for heating firms. We spend as much time defining what an agent must not do as what it should, and that discipline is central to how Dilr Voice is built and to the wider DATS methodology behind it.

How does AI voice fit an installer's existing tools and calls?

It sits on the phone line and writes into the systems already in use. A production setup connects a telephony provider such as Twilio to the installer's CRM or job-management software, so a qualified enquiry lands as a survey booking with the caller's details attached. It records calls under UK data protection law, and it keeps follow-up reminder calls on the right side of the line between a service message and marketing under PECR.

That plumbing is where the build-it-yourself platforms and the managed approach diverge. Tools such as Vapi, Retell AI, Synthflow and PolyAI give a capable installer the parts to assemble an agent themselves. A managed deployment instead hands over the integrations, the call recording compliance and the escalation rules as a finished system. The same qualification logic underpins the deployments our AI execution office owns, from first enquiry through to a live rollout. Which path fits depends on how much of the wiring an installer wants to own, and that is genuinely a business decision, not a technical one, which is where our approach to placing AI starts.

What is the best AI voice setup for a UK EV charger installer in 2026?

There is no single best; there is a best for your size and appetite. A solo installer who enjoys tinkering and takes low call volume can self-build on Vapi or Synthflow for very little and keep full control. A multi-crew regional installer fielding survey enquiries across grant schemes, with a CRM to integrate and call-recording duties to meet, is usually better served by a managed platform such as Dilr Voice, where the compliance and routing are owned end to end.

The honest concession is that the self-build tools win on upfront cost and on how much you can fiddle with them, and for a one-person operation that can be the right answer. Where the managed approach earns its keep is the moment a missed or mishandled enquiry costs a real survey slot, and the qualification logic has to be right every time rather than most of the time. That crossover point, not a feature list, is what an AI operating model consulting engagement is designed to find. Across enterprise adoption more broadly, McKinsey's State of AI research found in November 2025 that most organisations now use AI, yet only about a third have moved it into production, and the gap is almost always execution rather than tooling.

Does an AI agent replace the electrician's site survey?

No, it protects it. The agent's job is to make sure a survey is booked only when it is worth a visit, and that the surveyor arrives with the property, parking, supply and grant details already gathered. That turns a cold visit into a prepared one. The electrical judgement, the load assessment and the sign-off stay entirely with the qualified installer, exactly where the regulations put them.

Will callers know they are talking to an AI?

Yes, and they should. Good practice, and the direction of travel in AI regulation, is to tell callers plainly that they are speaking with an automated agent and to route them to a human whenever they ask. Dilr Voice is configured to disclose that it is an AI at the start of the call, which also tends to make callers more direct about what they actually need, shortening the qualification.

Want to see this in production? Try Dilr Voice live, book an AI placement diagnostic, see our DATS methodology, or read about our approach to placing AI inside field-service businesses.

Written by the Dilr.ai engineering team, practitioners who ship enterprise AI in production. Follow us on LinkedIn for shipping notes, or subscribe via the RSS feed.