What do we mean when we talk about artificial intelligence?
When we talk about ‘AI’ in the context of tools like ChatGPT, Claude, Gemini, or Copilot, we’re referring to computer systems that can perform tasks that usually require human intelligence, such as understanding language, generating ideas, analysing information or solving problems.
These systems learn from large amounts of data, recognise patterns, and use those patterns to produce useful outputs. That might mean drafting text, reviewing documents, summarising regulations, or generating images and diagrams.
Generative AI – the category that includes most of the tools mentioned above – goes a step further by creating new content based on what it has learned. This could be written material, visuals, structured information, or even design‑related suggestions.
Where does a sole practitioner who wants to explore generative AI start and how can it help them?
Keir Regan-Alexander, Co-Founder of OmniChat.uk (which provides an enterprise-grade large language model (LLM) platform alongside digital training and workflow resources to large teams), says: “For better or worse, practically everyone begins with the free version of ChatGPT or a meeting minute bot. In general I would avoid free tools and loads of third-party tools, because they usually retain your data permanently and train on it.”
“My recommendation would be to instead start with an entry level subscription for Claude and seek out some free online training courses by Anthropic (the makers of Claude). These will give some solid foundations, but you will likely struggle to understand how to ‘apply’ this to the domain of architecture and that is where coaches and advisors come in. Sometimes it’s also good to group up with other practitioners at the same stage to help support each other as you learn and test things.”
“Last year we recorded a CPD for RIBA Academy that is available to RIBA Members and our whole purpose is around helping professional teams take these generalised technologies and to learn to apply them to high specialised work.”
What are the most meaningful ways AI can change the workflow of a sole‑practitioner architect?
“I suspect lots of solo practitioner’s time is spent producing appointment contracts, programme planning, fee planning, invoice creation, bid writing, sending out tailored practice info, spec writing and dealing with contractor RFIs,” Keir says. “These are all areas where I am seeing AI used effectively. I also think that a solo practitioner with good hand-drawing, modelling and 3D skills will get outstanding results if they can learn to use Nano Banana Pro (by far the best AI image model we have tested, in my opinion).”
Can AI be used as a digital assistant in a business?
Perhaps a way to justify or frame using AI is to think of the tech as your very own assistant or intern. Use different AI tools for different tasks, give them clear instructions, set boundaries, and let them handle routine tasks.
Keir says that the most simple and straightforward application for AI is relatively wordy and admin-based tasks: things like reviewing new collections of documents when you are bidding for a project, or helping to communicate more effectively by email, sending out proposals, or writing up meeting minutes rapidly.
“For all of these tasks, you need to have to know that the model is working with good data to be able to verify results, be aware of the risks and to be able to apply the baseline theory in order to be able to achieve accurate and repeatable results,” he says. “I would caution against trying to go straight into areas like ‘vibe-coding’ and app development, which may be tempting. Ultimately we should be using AI for tasks that we are competent to perform and where we can discern good (error-free) output from bad.”