Ontology-driven · AI-powered · Private by design
I build ontology-driven, AI-powered apps for small and medium-sized businesses. Describe what you need in plain English, ask questions of your own data in plain English, and get an app that always matches the way your business really works.
Which jobs are running behind this week?
Your app: Here are the late jobs, grouped by engineer, with the customer for each. Shall I draft an update to each customer?
Yes, and add a rule: no job can be booked without an engineer.
Your app: Drafted. The new rule is ready to review in Test before it goes live.
ILLUSTRATION ONLY
The story of ontology-based apps
An ontology sounds grand, but the idea is simple: a clear, written description of how your business works. Everything else is generated from it.
The nouns of your business: customers, jobs, products, invoices, engineers, suppliers.
How those things connect: a customer places jobs, a job is done by an engineer, a job is billed on an invoice.
What must always be true: every job needs an engineer, an invoice needs an approved job.
Things, relationships and rules, written down once.
Tables and links generated, not hand-drawn.
Enforced consistently, everywhere in the app.
Screens and workflows that use your own words.
It understands your business because it reads the same ontology.
Because the database, the rules and the app all come from the same description, they can never drift apart. Change the ontology and the app follows, so the software always matches the business.
Architecture
Each customer gets their own Test and Live environments. Everything the AI needs runs alongside the app, inside your boundary.
The app, database, local AI and MCP gateway all sit inside your environment. Nothing is sent to public AI providers.
Every change lands on your Test server first, so you can see it working before it touches the business.
The build you approved in Test is the very same build promoted to Live. No surprises.
Hosting options
Both options keep your data away from the public internet and public AI providers. The difference is who shares the hardware.
Your own large language model, running on your own dedicated servers. It is yours alone, and nothing leaves.
A jointly-hosted LLM that lives inside our private network. It is multi-tenanted to keep costs down, and still never sends your data to the internet or to public AI providers.
How changes happen
Your business will change, and your app should keep up. Here is the journey of a change, from a sentence in plain English to your live system.
Describe the change in plain English.
The description of your business changes, and the app is regenerated from it.
Tests run to catch problems before you ever see them.
The new version appears on your Test server.
Try it, then give the go-ahead.
The same build you approved goes live.
If something is not right, go back to the previous version. Changes are packaged as versions, so stepping back is quick and calm.
People, and the AI itself, can only see and do what their role allows. You decide who can ask for changes, who can approve them and who can go live.
Why HA Consulting
Big-business software ideas, delivered at a sensible scale for small and medium-sized companies across the UK.
Built around how smaller businesses really work, with no enterprise bloat and no jargon.
Outgrown your spreadsheets? I turn them into proper apps with real rules, real structure and an assistant you can ask.
Your information is organised once, in your own words, and used consistently across the whole app.
Choose a fully sovereign set-up or a shared private one. Either way, your data stays off public AI.
Because the app is generated from the ontology, changes arrive quickly and stay consistent.
You work with Hugh directly, remotely, wherever you are in the UK. No account managers in between.
Pricing
Every business is different, so let’s have a conversation about what you need. Drop me a line and we’ll talk it through.