Built for companies of 10 to 500 people
Custom AI Development. Your Process. Your Code.
Custom AI built for companies of 10 to 500 people: chatbots, agents, automation and full platforms. Six to ten weeks from scoped quote to production. Fixed price, weekly Friday demos, every line of code yours from day one.
AI Makers is Mark Austen's custom AI software studio (AI Makers LLC). A custom AI app, built around your process, is the main thing I make for companies worldwide.
Fixed price. 24-hour reply. No sales call.
Or see how pricing works · build your own AI system in the interactive builder →
Mark Austen, Founder: I personally reply within 24 hours
Top Rated Plus
on Upwork
What is custom AI development?
Custom AI development is building AI software around one business’s own processes and data, instead of renting a general product. It usually means agents or assistants connected to the systems you already use, with rules for what they may do on their own. You own the code, and it runs where you choose.
What I build
One reasoning core. Specialised agents. Your tools.
A typical custom AI build: the channels your team already uses fan in from the left, a single reasoning core in the middle, and specialised agents fan out to whatever systems run your business. Fixed price, quoted after a short scoping chat with me. See how pricing works.
FigureA generic shape: your build picks the channels, agents, and SOPs that actually matter for your business. Every connection is an MCP tool server. Six to ten weeks end-to-end, fixed-fee, milestone-paid, you own the code. Assemble your own version of this diagram in the builder.
Google AdsConnectedWhat would you like me to do with Google Ads?
Custom AI · built for you
Your custom AI.
Built for your business.
A team of autonomous AI agents, each with its own role, memory, tools and scheduled tasks. Customised for how your business actually works, deployed on your infrastructure. Try the productised version at nora.work before commissioning a custom build.
On your infrastructure. Your data never leaves your cloud.
Email, calendar, CRM, accounting, ad platforms and storage, all wired in.
Per-user permissions, audit log on every action, governance baked in.
Detects its own errors, recovers, escalates only when a human is needed.
WhatsApp, web, Telegram, Slack and email: one brain, many surfaces.
Autonomous loops on schedule. Wakes up overnight, hands you the brief.
60+ Integrations
Connects to everything
you already use.
Pick the tools your business runs on. I wire them into your custom AI.
Deploy anywhere.
Your data stays yours.
Three deployment options. Full data sovereignty on every one.
Managed Cloud
Get started in minutes at nora.work. I host, monitor and update everything. Your AI is up and running the same day.
On-Premise
Install on your office Mac, Linux server, or any machine. Your data never leaves the building. I set it up remotely.
Self-Hosted Cloud
Deployed on your own AWS, Azure, or Google Cloud account. Full control. I build and hand over the keys.
Here's exactly what happens.
From first call to a running AI. No mystery.
Discover
A short scoping chat. I map your tools, workflows and the specific job the AI needs to do. You walk away with a one-page scope, fixed price.
Build
I build your custom AI with the agents you need: sales, support, finance, ops. Each gets its own tools, memory and scheduled tasks. Weekly Friday demos every step of the way.
Deploy
Deployed on your infrastructure, on-premise, or my managed cloud. Connected to your real tools and channels from day one. Real users in week 4 or 5, not week 12.
Evolve
Add more agents, integrations and automations as you go. Your AI self-heals and self-upgrades, so it keeps getting smarter as the business changes.
How pricing works
Fixed price, quoted after a short scoping chat with me.
One written quote for the agreed scope, before anyone writes code. It covers the build end to end: discovery, integrations, deployment and 30 days of post-launch support. It spells out what’s in and what’s out, and you own the code.
| Project | What that buys | Timeline |
|---|---|---|
| AI chatbot or assistant | Support, internal Q&A or lead qualification on your site or WhatsApp. Your knowledge base, 1 to 3 integrations, admin screen, hosting. | 3 to 8 weeks |
| AI automation across systems | AI that takes actions: invoices, reconciliation, documents, orders and returns. 4 to 8 systems, audit trail, human sign-off where a mistake costs money. | 4 to 12 weeks |
| Custom AI platform | A system your team logs into. Multiple agents, your own data layer, roles and permissions, dashboards, an API, deployed to your cloud or on-premise. | 8 to 20 weeks |
| Managed plan after launch | Optional. Hosting, monitoring, model updates and ongoing tweaks once it is live. Cancel any time. You own the code either way. | Ongoing |
What drives the quote: how many systems the AI has to touch, how clean your data is, and what a wrong answer costs you. Multi-country enterprise rollouts with compliance sign-off get scoped separately.
When custom AI isn't the right buy.
I would rather tell you now than sell you a build you don’t need. In these four cases, something else fits better.
Your team mostly drafts and searches
If people need help writing emails and finding answers in documents, ChatGPT Business or Claude Team does that today. Nothing needs connecting, so a build adds cost without adding value.
A mature tool already does the job
Helpdesks, CRMs and booking systems are solved problems. If an off-the-shelf product fits your process, buy it and add AI only where it falls short. I compare both routes in custom AI vs off-the-shelf software.
A single assistant would cover it
If a small team wants one AI assistant across WhatsApp, email and the calendar, Nora is my ready-made product for that. It runs on your own instance and is live the same day.
The process isn't settled yet
AI follows a process well, and it cannot decide one for you. If the steps change every week, write them down and run them by hand for a while, then automate.
Your competitors will ship their custom AI this quarter.
The SMBs that move first will have AI that’s been learning their business for months while everyone else is still on a sixth ChatGPT pilot. 6 to 10 weeks from this call to live in production.
Questions you're probably asking.
Who do you build for?
Mostly companies of 10 to 500 people: boutique professional-services firms, mid-market SaaS, e-commerce brands, real estate teams and local-services businesses. The sweet spot is businesses where AI moves the numbers but who don't want to hire and run an in-house ML team.
What exactly do I get?
A private, custom-built AI system deployed on your server or my managed cloud. You choose what it does: customer support, invoice processing, lead qualification, document analysis, ad management, whatever moves your business. Each AI agent has its own memory, scheduled tasks, and integrations. Your team controls everything through a built-in admin panel.
How much does it cost?
Scope drives it: how many systems the AI touches, how clean your data is, and what a wrong answer costs you. Fixed price for the agreed scope, quoted after a short scoping chat with me. The written quote spells out what's in and what's out, and you own the code.
How long does it take?
Live in 6 to 10 weeks for most projects. Chatbots take 3 to 6 weeks, multi-system automation 6 to 10 and custom platforms 8 to 14. Weekly Friday demos every step of the way, so there is no 12-week silent build.
Can it connect to my existing systems?
If it has an API, I connect it. 60+ integrations out of the box, including Salesforce, HubSpot, Stripe, Xero, AFAS, Exact, Shopify, WhatsApp, Slack, Outlook, NetSuite, SAP and dozens more. No API? I build a custom connector.
Is my data secure?
Your AI runs on your own server, your office machine or my managed cloud: your call. If you host it, documents, customer data and conversations never leave your infrastructure. I don't train any model on your data. SOC 2, GDPR, HIPAA, PDPA controls available based on your needs.
How is this different from ChatGPT or Microsoft Copilot?
ChatGPT and Copilot are great for personal productivity: drafting, summarising, brainstorming. Custom AI is what you need when the AI has to take actions in your real systems, with audit trails, multi-user controls and your specific workflow. I recommend ChatGPT Enterprise to almost every client alongside the custom build.
Do I own the code?
Yes. The code is yours, the data is yours, the IP is yours. No vendor lock-in. I build on OpenClaw, a third-party open-source AI agent framework, so even the foundation is open. You can move providers any time.
What's OpenClaw and why does it matter?
OpenClaw is a third-party open-source AI agent framework. I build on it, and I run Nora, a managed product on top of it, at nora.work. Starting from a battle-tested foundation collapses what used to be a 6-month custom build into a 6 to 10 week project. You get the speed of a productised foundation with the fit of a custom build. See it live at nora.work.
AI Makers built our bank reconciliation system in 3 weeks. It replaced hours of manual data entry every day.
We needed a full booking platform with AI tour descriptions and a customer chatbot. They delivered everything and it just works.
Where I work
Remote-first, with dedicated guides for the markets I ship to most.
Who I build for
What custom AI looks like in your industry or department.
How pricing works: fixed price, quoted after a short scoping chat.
See the week-by-week timeline: what happens in weeks 1 to 10.
Read my 6-stage process, plus the readiness checklist.
Design your system in the builder: assemble agents and integrations.
How custom AI integration works: what each connection costs, and what breaks in month one.
Custom AI
Ready to build your custom AI?
Let's talk.
Tell me what tools your business uses and what you'd like the AI to do. I'll come back within 24 hours and tell you straight if custom is the wrong tool for the job.
Send me a message
I'll get back to you within 24 hours.