What an AI agent is
An AI agent is software that does a job, not software that answers questions about one. A general AI tool can draft a reply if you paste in the message. An agent watches for the message, writes the reply within your rules, sends it through your systems, and records what it did.
The difference that matters to an operator is who moves first. A chat tool waits for you to ask. An agent is already working when the request arrives, and it only comes to you when something falls outside the rules you set.
The work agents handle in property management
The best candidates are the tasks that come up every day, have a clear finish, and follow rules your team could write down. In property management that covers a lot of the week.
- Collects move-in paperwork, follows up on missing items, and prepares key release for your approval.
- Reads a signed management agreement, prepares owner and property records, and follows up on missing paperwork.
- Compares vendor bills with work orders and agreed rates, flags differences, and prepares bills for review.
- Collects maintenance requests, follows your response rules, contacts the right people, and records the next step.
- Organizes inspection notes and photos, prepares repair requests, and drafts deposit paperwork for your team to review.
- Tracks lease end dates, prepares renewal offers using your rules, and follows up on approved offers.
- Answers listing questions, collects inquiry details, and helps arrange tours. Questions outside its approved answers go to your team.
What an agent needs from you
Three things: access, rules, and a definition of done. Access means the same accounts you would give a new hire on day one, limited to that role. Rules mean your response times, spending limits, approved vendors, and the decisions that stay with people. Done means the record your team expects to see when the work is finished.
An agent without those is a demo. With them, it is a member of the team you can audit.
What should stay with people
Every Openhour agent has actions it never takes, however long it has been running. They are the decisions where a wrong call harms a resident, an owner, or your company.
- Issue a door code on an uncertain file.
- Set up or change bank accounts or payouts.
- Pay anything.
- Replace your emergency response service or decide who owes repair costs.
- Send or serve a notice.
- Decide not to renew, negotiate exceptions, or penalize accommodation requests.
- Tell anyone they do not qualify.
Built into your software, or built around your work?
The property management platforms now ship AI features of their own, and for simple work inside one system they can be enough. A custom agent makes sense when the job crosses systems, follows rules the built-in feature cannot express, or needs a team to keep it running after your software changes.
We have written up how Openhour compares with the AI in the major platforms, using each vendor's own published descriptions.
How an agent goes live without a leap
An agent should never touch real records on its first day. It is tested on made-up records first, then it watches your real work without changing anything, and your team reviews what it would have done. Only after you approve does it take real actions, on properties you choose, with limits that are written down.
Questions to ask any vendor
Whether you are looking at Openhour, a feature built into your property management software, or a standalone product, the same questions tell you whether the agent will actually work for your company.
- What does it connect to in my software today, and what would my team still do by hand?
- What does it do on its own, and what does it ask a person about first?
- Can I see it run on my process, with my rules, before it touches real records?
- What happens when it cannot finish? Who is told, and how?
- Who hosts it, who fixes it when my software changes, and what is the bill?
Common questions
What is an AI implementation team?
We combine experience running service businesses with technical implementation. We investigate where work gets stuck, agree on a priority, implement and test the setup, and train your team. We stay involved through your agreed support and evaluate relevant developments as your needs and AI capabilities change.
What is an AI system?
An AI system uses the information it receives to produce outputs such as answers, recommendations, predictions, or decisions. For a business, we bring together the models, information, applications, and connections that make those outputs useful.
What is an AI agent?
An AI agent uses a model to choose steps and tools as it works toward a task. It can be one component of a larger AI system. We set the permissions and rules for what it can do and when a person must approve an action.
What starts at $500 a month?
The AI Implementation Team starts at $500 a month for agreed help with existing tools, setup, and training, one priority at a time. We confirm the included scope before work begins. Software subscriptions are separate, and custom builds are priced separately.
How are custom systems priced?
Custom systems have a separate build fee and ongoing retainer based on the work involved. Your proposal explains the deliverables, support, hosting, software and AI usage costs, and any limits before you agree.
Do we have to be in Charleston?
No. Our founders are based in Charleston, and we work with businesses elsewhere through video calls, screen sharing, and agreed implementation work.
Who owns a custom system?
Ownership of custom code, infrastructure, reusable components, and handover arrangements is agreed in each project’s contract. Third-party software remains subject to its provider’s terms.
Can you connect the software we already use?
Yes. We connect existing software and build custom integrations, using supported APIs and other appropriate methods. We agree on the connections and permissions for your tools before building.
Can a custom system use different AI providers?
Custom infrastructure can be designed to work with different AI providers and suitable open-source models. We choose models around your needs and evaluate and test changes before switching. Hosting, model operation, and migration work are scoped for your project.