AI & DataPillar 02 of 04 · AI & data systems

AI that does the work, on data you can act on.

Useful AI and commercial-data infrastructure, wired into your business. Document automation, on-brand content, public-data pipelines and dashboards, with a human keeping the final say.

Henry Goff leads every build. A data & AI engineer from the bench joins for pipelines and model work as the scope demands.

See the run-sheet

Free 30-minute callFixed price before the buildYou own the code from day one

Admin hours saved by one CRM
1,000sCyntex Group · in the director's words
To list a bike, down from an evening
MinutesLozMoto · in the owner's words
Vehicle-data pipelines live
2Sell Your Cab · LozMoto
listing · draft outLozMoto admin after generation: a full listing in the owner's voice plus a social caption ready to shareThe listing generator on a phone, on the forecourt
LozMoto · the generated listing, and on the forecourtreal admin · Claude
Drawing
Pillar 02 · AI & Data
Sheet
1: home · 2: packages
Drawn by
Henry Goff
Stack
Claude · GPT · n8n · Postgres · DVLA & DVSA APIs
Status
2 pipelines live
Rev
2026-09-08

§01 The workRun-sheets · screens as shippedOne regulated, one on the forecourtLozMoto screens are the live admin. Cyntex Group's screens are withheld at the client's request and drawn as a redacted document. Bars are illustrative.

A regulated CRM and a dealer's listing engine, and what each one replaced.

Two builds where AI does a specific job on specific data, with a person keeping the final say. Neither is a chatbot.

run 01Cyntex Group1,000s of hours saved

FCA-aware CRM & document automation

Bespoke CRM for a regulated claims business: client intake, document generation, change management and records in one compliant system. The letters, packs and correspondence that ran on copy-paste are now generated from case data.

As an FCA regulated business, we needed someone who understood our field and could deliver a fast, simple and compliant custom CRM with document automation to match. Henry delivered on every count. Thousands of hours have been saved.
Muhammed Khan · Director, Cyntex Group
Survived compliance

FCA-regulated environment: intake, generated documents, change management and records held in one auditable system, with the data handling the regulator expects.

intake · documents · change log · recordsdrawn

Where the work livessystems per client record

Before4 places
After1 system

Intake, documents, change management, records: four functions consolidated into one compliant system. Bar lengths illustrative; screens withheld at the client's request.

run 02LozMotoMinutes to list a bike, down from an evening

AI-assisted listings engine

A registration lookup fills make, model, year, engine, colour and MOT history. Loz adds one line of highlights — “recent service, new tyres, two keys” — and Claude drafts the full listing and a social caption in his own voice. He reviews, tweaks, publishes.

Honestly, it's how we run the shop now, not just a website. I punch in a reg and the bike's details and MOT history fill themselves in. Listing a bike used to eat a whole evening. Now it's done in minutes.
Loz Ridd · Owner, LozMoto
Survived compliance

Vehicle details and MOT history pulled from source by registration; every draft lands in an editable field and is reviewed by a person before it ships.

Time to list one bikeper listing

Beforean evening
Afterminutes

Bar lengths illustrate “an evening” against “minutes”, in the owner's words; not timed.

run 01

FCA-aware CRM with document automation at its core.

run 02

AI-drafted listings from a registration lookup.

pipeline

Vehicles valued from DVLA & DVSA lookups the moment a reg is typed.

Commercial intelligence, wired into your business.

Public data is a working asset, if it arrives clean, current, and connected to the systems you run. HG.Studio builds the pipelines that turn public records and market signals into feeds your business actually acts on.

data · vehicle lookupVehicle registration lookup pipeline: DVLA VES, MOT history, verify, valuation
Sell Your Cab · DVLA VES, MOT history, verify, value — by registrationlive

Already running in the wild: Sell Your Cab values vehicles from DVLA & DVSA lookups the moment a driver types a reg, and LozMoto's admin fills in a bike's details and MOT history from a registration alone.

The engineering, brieflyfour parts
  • Scheduled pipelines & jobs
  • API integrations & enrichment
  • Structured storage with history
  • Dashboards, alerts & logging
Inside the rulesread them

Public data comes with licences, rate limits and data-protection duties. Knowing those rules is exactly why you want a specialist: every pipeline is designed inside them, so the intelligence you act on is intelligence you're allowed to use.

  • Part 01

    A live feed of commercial intelligence

    Public data pipelines, lead and source monitoring, market and location datasets: the signals your market gives off, collected on schedule and delivered where you work.

  • Part 02

    Public records, turned into something usable

    Companies House, GOV.UK, planning and business-listing feeds, plus FOI request workflows, transformed from scattered records into structured, queryable data.

  • Part 03

    Data that's watched and acted on

    Dashboards, alerts, enrichment and logging, so a change in the data becomes an action in your business, not a report nobody opens.

  • Part 04

    AI doing real work on top

    The AI systems above run best on good data: classification, document automation and enrichment wired straight to the feeds, with a human keeping the final say.

Try a piece of it: the AI Prompt Builder, free

Six parts, from documents to dashboards.

Practical AI runs best on good data, so the parts come as a set: the pipelines feed the models, and the dashboards show what both are doing.

  • Part 01

    Document automation

    Generate, assemble and populate the documents your business produces by hand — letters, packs, reports — straight from your data.

    • Templated & generated documents
    • Data-to-document pipelines
    • Compliance-ready output
  • Part 02

    Content in your voice

    Descriptions, captions and copy drafted on-brand and on-message, tuned to sound like you, not like a chatbot.

    • Brand-voice tuning
    • Drafts you edit, not auto-posts
    • Listings, captions & email copy
  • Part 03see also

    AI inside the workflow

    Claude and GPT wired into real workflows — n8n, your CRM, your inbox — so AI handles the steps a person used to grind through.

    • LLM steps inside n8n
    • Triggered, queued, retried
    • Human-in-the-loop checkpoints
    Explore Automation
  • Part 04

    Classification & routing

    Triage at machine speed: read, tag and route enquiries, tickets and documents so the right thing lands with the right person.

    • Intent & topic classification
    • Smart routing & tagging
    • Priority & sentiment scoring
  • Part 05

    Commercial-data pipelines

    Public records and market signals collected lawfully, structured, and delivered where you work, with dashboards and alerts on top.

    • Scheduled pipelines & jobs
    • API integrations & enrichment
    • Structured storage with history
  • Part 06see also

    Dashboards & alerts

    Surface what the AI is doing and what the data is saying: summaries, trends and ROI in a dashboard your team will actually open.

    • AI-generated summaries
    • Live performance & ROI
    • Alerts before problems land
    Explore Apps & Products

Find where AI fits, design the guardrails, build, then watch it.

The prompts, the brand voice and the checkpoints where a human signs off are designed and agreed before anything ships.

How a build runsfour steps
  1. 01

    Find where AI fits, free

    30 min

    We find the jobs where AI genuinely earns its place — and, just as honestly, the ones where it doesn't.

  2. 02

    Design prompts & guardrails

    agreed before build

    The prompts, the voice, the data sources, the licences, and the checkpoints where a person signs off.

  3. 03

    Build & integrate

    tested on real data

    The AI steps and pipelines built and wired into your real tools, with safe fallbacks for when a model has an off day.

  4. 04

    Monitor & improve

    every generation logged

    Every generation and every pipeline run is logged and reviewed. Prompts stay sharp and output stays on-brand as the business moves.

Studio leadaccountable

Henry GoffFounder & developer, London

Benchspecialist · contract
  • Product designeras scope demands
  • iOS engineernative apps
  • Data & AI engineerpipelines · models
  • SEO specialistsites that must rank

Every figure names where it comes from.

FigureWhat it measuresSource
1,000sAdmin hours saved by the Cyntex Group CRMrun 01 · in the director's words
MinutesTo list a bike on LozMoto, down from an eveningrun 02 · in the owner's words
2Vehicle-data pipelines live in productiondata systems
1Free AI tool built and given awayAI Prompt Builder

The questions people ask before booking.

Q1Is the anchor the price I'll pay?

No. from £4,000 +VAT is where this kind of build starts; an intake-and-documents system typically lands at £6,000–£12,000. Every project is scoped on a free 30-minute call and priced as a fixed scope, plus VAT, before anything is built. The number you sign doesn't move unless the scope does.

Q2Is AI genuinely useful for a business my size?

Sometimes, and the discovery call is where we're honest about it. We look for jobs AI demonstrably does well — drafting documents, writing listings, sorting and routing — and put a number on the hours they cost you today. Where it doesn't earn its place, we'll say so.

Q3Isn't a ChatGPT subscription enough?

For a person, often yes, and we'll tell you so. A £20-a-month seat is the right answer when one person wants help drafting. A built system earns its keep when the work is repeated, when it has to pull from your own data, and when the output has to be consistent, logged and reviewed — a claims pack generated from case data, or a listing drafted from a registration lookup.

Q4Won't AI-written content sound generic?

Not if it's tuned properly. LozMoto's listings are drafted in the owner's own straight-talking voice, and every draft is reviewed by a human before it ships. The AI does the typing; the business keeps the final say.

Q5Is collecting public data even legal?

Done properly, yes, and the rules are exactly why you want a specialist. Every pipeline is designed within ICO guidance on AI and data protection, the Open Government Licence, and each source's own terms, from Companies House to Google Places. Public doesn't mean free-for-all; that is the point.

Q6What happens if Henry is unavailable?

You own the repository from day one, so nothing is held hostage. Every build is documented to a written standard, a named backup developer is written into the contract, care plans carry stated response times, and the contract includes a repo-ownership clause. The studio is small on purpose; the continuity policy is what makes that safe to buy.

Q7Who owns the code and the accounts?

You do. The repository, the hosting account, the domain and every third-party service are set up in your name or transferred to it at handover. We hold access only for as long as a care plan says we should.

Put AI to work on the boring bits.

A free 30-minute call: we find where AI genuinely saves your team hours — drafting, sorting, summarising — and exactly what it would take to build it. It ends with a fixed price.

from £4,000 +VATintake & documents typically £6–12k · fixed scopeSee all packages