Quick answer: AI for estate agents in 2027 comes in three stages. Generative AI writes things when you ask. Insight AI learns from your data and tells you what you missed. Execution AI completes a whole job end to end and hands you the finished draft to approve. Most agencies are still paying for stage one.
Alto has just introduced the next iteration of Alto Intelligence: Six AI skills that take on an everyday agency job from start to finish. Every one of them can be changed by the agency using it, and agencies can build their own. This guide explains what changed, how to tell the three stages apart in a demo, and the seven questions worth asking before you sign anything.
What is AI for estate agents?
AI for estate agents is software that uses machine learning and language models to carry out agency work: matching applicants to properties, writing listings and market appraisals, chasing sales progression and rent arrears, running anti-money laundering (AML) and compliance checks, and answering questions about the agency’s own data in plain English.
The distinction that matters in 2027 is not whether a CRM has AI. Every provider in the sector now says it does. The distinction is how much of the job the AI is allowed to finish.
The three stages of AI in estate agency
This is how Alto CEO Riccardo Iannucci-Dawson splits it, and it is the fastest way to cut through a demo.
| Stage | What it does | Estate agency example | Who does the work |
|---|---|---|---|
| 1. Generative | Produces content when prompted | “Write me a listing description” | You prompt it, you check it, you send it |
| 2. Insight | Learns from your data and surfaces what you didn’t know to look for | “These four sales are at risk of falling through, and here’s why” | Software finds it, you act on it |
| 3. Execution | Completes the whole job and returns a finished draft | Spots the at-risk sale, writes the chase to the solicitor, logs the activity | Software does it, you approve it |
Stage one is the baseline now, and you can get most of it from a free chatbot. Stage two needs depth of data, not just access to a model, which is why fewer tools genuinely operate there. Stage three is where an agency’s day changes, because the software stops being the place you record the work and becomes the place the work happens.
“A feature helps an agent to complete a job. A skill actually does the job.” – Riccardo Iannucci-Dawson, CEO of Alto
Take the arrears run. AI that supports you sorts the list by who’s furthest behind, and you still work out who’s really in trouble and still write every letter, with some AI polish. A skill works that out, writes the chaser for each one, and flags the three that need a phone call. The property manager reads them and presses send.
Most agencies are heading into 2027 still buying stage one
Adoption is no longer the interesting number. Near enough everyone is using AI. What they use it for is the story.
In a February 2026 survey of 225 agents by Realtor Property Resource, part of the US National Association of Realtors, 82% said they had integrated AI into their business. The top uses were writing tools (77.9%), listing descriptions (68.5%), social content (59.5%) and email (53.2%). That is stage one, almost across the board. The same agents named accuracy (63%) and compliance (49%) as their biggest worries.
The UK picture looks much the same. As 2026 closes, agents have AI. Very few have AI that finishes anything.
Here is the honest test. Open your current CRM and ask: in the last week, what did this software do that I didn’t tell it to do? If the answer is “sent the portal feed”, you are still the automation.
What execution AI looks like: Skills
On Wednesday 2nd September 2026, Alto introduced skills, six AI skills that sit inside Alto Intelligence, with more to follow. Each one takes a single job an agent repeats every week and completes it from start to finish.
| Skill | The job it completes |
|---|---|
| Property matcher | Finds the applicants who actually fit a property, not just the ones who match on paper, and shows why each one made the cut. One curated update per contact rather than an email per property. Agencies set what ‘fit’ means for them. |
| Listing writer | Writes the description, the bullets and the room-by-room in your agency’s style, learned from the listings you’ve already done. Better copy, more viewings, nothing to rewrite. |
| Appraisal builder | Writes the whole appraisal report: the properties that actually match this one, the ones you sold on that street last year, sounding like your agency wrote it. Win more instructions without losing your evening. |
| Progression chaser | Tells you which of your sales are in danger and why, in plain English, then writes the chase to the solicitor. You decide what ‘in danger’ looks like. |
| Arrears chaser | Tells you which tenants are heading for real trouble, not just who’s been late longest, writes the chaser for you, and flags the cases that need a call instead of an email. |
| Alto IQ | Ask a question in plain English and get an instant answer, chart, table or number, with an explanation of what’s happening and why. |
Every skill runs on demand from inside Alto. It gathers the data, makes the judgement and writes the words, then hands the finished draft to an agent to approve.
Nothing goes out until an agent says so
Nothing reaches a client and no record changes until a person approves it. The arrears chaser never moves money. The progression chaser doesn’t email the solicitor until you’ve read what it wrote.
That matters commercially, not just ethically. When AI writes a client communication in a regulated industry and something goes wrong, who carries the risk is a live question. Some providers are already on record saying disclosure of AI-generated communications is the agent’s decision, not the platform’s. The Property Ombudsman and the Information Commissioner’s Office (ICO) will form a view in due course. Agents should form theirs first.
Change any skill, or build your own
This is the part that separates skills from an AI feature list, and it’s the part worth asking every provider about.
Every one of the six can be changed by the agency using it: The criteria it judges on, the tone it writes in, and the thresholds that decide what gets flagged. If your definition of a sale ‘in danger’ is different from the next agency’s, you change it.
Agencies aren’t limited to the six either. The same tools Alto’s own team used to build them are available to customers. An agency describes the job it wants done in plain English: which records to look at, what to judge, what to produce, and when to run. There’s no code to write and no technical knowledge needed. The skill then runs on that agency’s own records, in that agency’s own house style, against that agency’s own history.
That reverses how property software has usually worked. Instead of waiting months for someone to build the thing they need, an agency can build it itself in an afternoon.
“There are 6,000 agencies on this platform and they don’t run the same business. The good ones have a way of doing things that’s genuinely theirs. How they qualify an applicant, when they pick up the phone on an arrears case, what their appraisals look like. Software has spent 20 years asking those agencies to work the way it works. This is the other way round.” – Riccardo Iannucci-Dawson, CEO of Alto
Why the data underneath matters more than the model
Generic AI is generic. A model trained on the internet has never seen a chain collapse on a search, a Section 8 notice, or a client account reconciliation.
Alto holds 20 years of workflow data captured across 6,000 agencies, through every market condition and every kind of transaction, now running to 1 billion anonymised data points. When a skill judges which sale is at risk, it is reasoning over what actually separated the deals that closed from the ones that stalled.
“You can build a clever interface in a quarter. You can’t manufacture 20 years of estate agency workflow data. Every CRM can tell an agent what happened. Alto can tell them what may happen next, because we’ve watched it happen across a third of the market. A system starting today can record what’s going on now. It can’t go back and watch the last 20 years and turn that into the decisions a modern agency runs on. We can.” – Riccardo Iannucci-Dawson, CEO of Alto
Alto has confirmed a further set of skills is in development. These will judge against the whole data set rather than a single agency’s records, showing an agency how its own performance compares with what the wider market is actually doing. Alto expects to say more later in the year.
There is a second thing agents rarely get told. Much of what is sold as AI in this sector is a wrapper around someone else’s large language model, licensed and marked up. That isn’t wrong in itself, but it carries token costs that don’t show up on the price card, and the same question can come back with a different answer on Monday and on Tuesday. When you’re writing to vendors, buyers and tenants on behalf of your business, that isn’t theoretical.
AI and compliance: The part agents underestimate
Compliance is where execution AI earns its keep, because compliance failures are almost never failures of storage. They are failures of noticing.
Since 1st May 2026, the Renters’ Rights Act has ended Section 21, moved tenancies to periodic and limited rent increases to once a year. Add AML checks under HMRC supervision, material information on every listing under the Digital Markets, Competition and Consumers (DMCC) Act, deposit deadlines, and gas, electrical and energy certificates to renew on time.
A system of record waits for you to notice the certificate expired. A system of work notices for you. AutoCert cuts certificate renewal admin from around 30 minutes to five, and identity and AML checks run through Alto Verification inside the platform, cutting the time it takes to get a client checked and set up by roughly 85%.
Seven questions to ask before you buy AI for your estate agency
- Which stage is it? Generative, insight or execution. Make them show you, not tell you.
- Can we change it? If the criteria and thresholds are fixed, you’re buying someone else’s way of working.
- Can we build our own? Ask whether you get the same tools their team used, or a request form.
- Is it live today, or still to come? “Coming later this year” is not a feature.
- What is the total cost at my volumes? Licence fee plus token usage, worked out properly, not the headline number.
- Who carries the risk when a generated client communication goes wrong?
- What is it judging against? A platform’s intelligence is only as good as the transaction history behind it.
We’ve gone deeper on this in what to ask before buying AI for your estate agency CRM, and compared the market in AI CRMs for estate agents in 2026: 5 compared.
How to start using AI in your agency without a change programme
- Pick the job that eats your Tuesday. Usually arrears, progression chasing or applicant matching.
- Run it alongside your existing process for two weeks, and compare the draft with what your team would have sent.
- Change the thresholds until the skill flags what you would have flagged.
- Measure one thing: time per task, instructions won, or fall-through rate. Not ‘adoption’.
- Then add the next job, or build a skill for the one nobody else does the way you do.
Alto saves each user 10+ hours a week on average. That comes from removing jobs one at a time, not from a single feature.
The bottom line
Alto Intelligence launched in spring 2026 with five AI tools supporting the roles inside an agency. It now moves from supporting those roles to completing the work they carry out, and from Alto building the AI, to agencies building their own.
The agencies that do well in 2027 won’t be the ones that adopted AI earliest. They’ll be the ones that adopted the right stage of it. Stage one is a free chatbot. Stage three is a different business.
Changing CRM is hard. Staying on one that leaves all the work with you is harder.
“Agents didn’t join this industry to work through an arrears spreadsheet on a Tuesday night. They joined it to win instructions and sell houses. These skills give them more of the job they actually wanted, and they still decide what goes out the door.” – Riccardo Iannucci-Dawson, CEO of Alto
See all six Alto Skills, what each one does and what it replaces, on the Alto Intelligence page
Frequently asked questions
What is AI for estate agents?
AI for estate agents is software that uses machine learning and language models to carry out agency work, from matching applicants to properties to writing listings, chasing progression and arrears, running compliance checks and answering questions about your data. Going into 2027 it ranges from AI that drafts content when asked to AI that completes whole jobs end to end.
What are the skills?
These are skills inside Alto Intelligence that each complete one everyday agency job from start to finish. Alto launched six on Wednesday 2nd September 2026: Property matcher, Listing writer, Appraisal builder, Progression chaser, Arrears chaser and Alto IQ. Each one gathers the data, makes the judgement and writes the output, then waits for an agent to approve it.
Can an agency change how a skill works?
Yes. Every one of the six can be changed by the agency using it, including the criteria it judges on, the tone it writes in and the thresholds that decide what gets flagged. If your definition of a sale ‘in danger’ differs from another agency’s, you change it rather than working to theirs.
Can an estate agency build its own AI skill?
Yes. Alto gives customers the same tools its own team used to build the first six. An agency describes the job in plain English, covering which records to look at, what to judge, what to produce and when to run, with no code and no technical knowledge needed. The skill then runs on that agency’s own records and history.
What is the difference between generative AI and execution AI?
Generative AI produces content when you prompt it, such as a listing description or an email draft. Execution AI works out what needs doing, does it, and returns a finished draft with everything logged. As Alto CEO Riccardo Iannucci-Dawson puts it: “A feature helps an agent to complete a job, while a skill actually does the job“.
Can AI replace estate agents?
No. AI in 2027 removes repeated administrative work such as matching, drafting, chasing and checking, so agents spend more time winning instructions, negotiating and advising. A person still approves every client communication before it is sent.
Is AI in estate agency software safe from a compliance point of view?
It depends on the design. Look for a mandatory human approval step before anything reaches a client, a full audit trail, no automatic movement of client money, and a clear answer on who carries the risk if a generated communication is wrong.
What data do the skills run on?
Our skills judge against 20 years of workflow data captured across 6,000 agencies, now running to 1 billion anonymised data points, alongside the agency’s own records and house style. A further set of skills in development will judge against the whole data set, so an agency can see how its performance compares with the wider market.
