Updated 20 September 2026: first published on Elite Agent in November 2025. The article has been refreshed as an AI basics guide to identifying and testing a useful new workflow.
In November 2025, I wrote about spending much of the year speaking, learning and quietly building Ailsa, our AI journalist at Elite Agent.
At events, more hands were going up when I asked who used ChatGPT. People were drafting letterbox drops, email sequences and social captions. Those are useful places to practise. Building Ailsa pushed me to ask another question: what could we do now that had previously been difficult to offer?
For years, agents had asked us to tell the stories behind their sales. The challenge was finding the time for interviews, writing, editing and checking. AI gave us a way to reconsider that process while retaining human editorial work.
What could I build with AI in my real estate business?
Begin with a problem you recognise, not an app you have just discovered.
Perhaps useful inspection feedback never reaches the person preparing the next campaign. Perhaps colleagues repeatedly answer the same process question. Perhaps the story behind a sale disappears when everyone moves on to the next listing.
Describe who needs help, what they are trying to achieve and where the current process gets stuck. Ask whether better instructions or a simpler process would solve it before adding another tool.
That is the beginner version of AI-first thinking: reconsider the task with the available tools, while keeping the outcome and the people involved in view.
What building Ailsa taught me
I learned automation workflows with AI’s help and explored conversational AI through videos and NotebookLM. The learning took time. A useful prototype did not eliminate the need to check facts, organise handovers or decide who was responsible.
Ailsa’s process starts with an interview about an agent’s actual work. The draft goes through human editing and the agent’s approval before publication on Elite Agent. That last part matters as much as producing the draft.
I had also seen Elite Agent articles appear when researching agents with AI for keynote examples. Published stories can provide evidence of someone’s work. An appearance in one answer does not guarantee future citations, rankings or recommendations.
Try a small design exercise
Help me explore one possible AI workflow for a real estate business.
People who need help: [ROLES]
Outcome they need: [OUTCOME]
Current steps and bottleneck: [DETAILS]
Approved information and tools available: [DETAILS]
Human decisions and approvals that must remain: [DETAILS]
Ask questions where the process is unclear. Suggest three options: improve the process without AI, use AI to assist one step, or test a new workflow.
For each option, explain the information needed, human checks, handover, likely failure points and a small test. Do not invent integrations, capabilities, costs or time savings.
Recommend the smallest experiment that could tell us whether the idea is useful. Define what we should observe and what would make us stop or revise it. Do not treat a prototype as ready for clients.Build something you can check
During a podcast conversation with Ben White, we explored making a small calculator around solar panels and property value. That was a prototype. Its assumptions and calculations needed checking before anyone could rely on the result.
The same applies to a first agency experiment. If you draft a question-answering assistant, use a small source document and questions with known answers. If you organise campaign notes, check the output against those notes. If you generate a calculation, compare it with an independently checked example.
Do the first trial internally with fictional or approved information. Write down what happened rather than assuming that a convincing demonstration represents normal performance.
Decide who takes over next
A useful workflow has an owner. Someone maintains the information, checks exceptions and responds when the system cannot finish the task.
For Ailsa, publication follows an editorial and approval process. For your experiment, the next step might be a colleague checking a summary before it reaches a client. Make that step visible in the design.
A small trial can tell you to continue, simplify or stop. Each is more useful than building a complicated system around an assumption nobody tested.
Put the idea to work
If you have a sale worth telling, explore Ailsa. If you want to work through opportunities in your own agency, contact me about a practical team AI session.
The question is worth keeping: what could we make possible now, and what is the smallest useful way to test it?
The original version appeared on Samantha McLean’s Substack. Read the refreshed source article on AI Powered Real Estate by Samantha McLean on Substack.