Stop Asking AI for Answers. Start Giving It Jobs.
Most people treat AI like a search engine. Type a question, get an answer, move on. It's not wrong, exactly — it just leaves a lot on the table.
The problem with that approach is that it puts AI in charge of figuring out what you actually need. When you ask "what should I write in this email?" the AI has to invent the context, the tone, the purpose, and the recipient from almost nothing. It'll produce something — that's what it does — but it'll be generic, because generic is all it can be without direction.
The frustration that follows — this is useless, AI isn't ready for real work — is a predictable result of the wrong mental model. Not evidence that the tool doesn't work.
Stop Asking. Start Assigning.
Here's the shift: stop asking AI for answers and start giving it jobs.
A job is specific. It has a defined output. It gives AI the context it needs to produce something actually useful.
Compare these two:
Question: "What should I write in this email?"
Job:
Write a follow-up email to buyers I showed three homes to yesterday who
haven't responded. They seemed most excited about the second house but
mentioned price was a concern. Keep it under 100 words. Warm but not pushy.
One clear ask: let me know if they want to revisit any of the properties.
The second one defines what's being produced, what the raw material is, what constraints apply, and what the goal is. The output will be usable. The first will produce something you delete.
That one shift — from asking to assigning — changes most of what you get back.
See the Gap Between Raw and Usable
Once you stop treating AI as an answer machine, you start noticing a different kind of opportunity in your workday.
A typical day is full of raw material: voice notes from showings, scattered bullet points from a buyer consult, a dense email thread about a deal going sideways, a set of agent remarks you need to turn into a showing summary. That information exists. It just isn't organized in a way that makes it easy to act on.
AI closes that gap almost instantly. Showing notes become a structured buyer feedback summary. A messy back-and-forth email becomes a clear written decision with next steps. A rough outline becomes a first draft.
When you start seeing your work through that lens — where is the raw information, and what does it need to become? — you stop wondering what to use AI for. The opportunities become obvious.
Think of It as a Talented Colleague, Not a Button
The most useful mental model for AI isn't a search engine or an autocomplete function. It's a collaborative partner — a capable one that needs clear direction to do its best work.
Imagine a colleague who has read everything: every contract form, every negotiation tactic, every market report, every piece of real estate marketing ever written. They can draft, summarize, organize, and research faster than anyone you've worked with. But they don't know your clients, your market, or the unspoken context behind your request. And they will occasionally state something confidently that turns out to be wrong.
Your role isn't to trust them blindly. It's to direct them clearly, review their work, and apply your own judgment to what they produce.
The division of labor: you provide the direction, the context, the expertise, and the final call. AI provides the speed, the drafting, and the structure. Neither works well without the other.
Skepticism Is Part of the Job
Shifting how you think about AI doesn't mean trusting it more. It means understanding precisely where it needs checking.
AI can be wrong — and it won't signal uncertainty the way a careful person would. It will state an incorrect closing date, a wrong commission calculation, or a made-up zoning rule with the same confident tone it uses for accurate information. This isn't a bug that'll get fixed. It's a known characteristic of how the technology works, and it has a name: hallucination.
The right response isn't to avoid AI. It's to approach every output the way you'd approach any draft — as a starting point that requires review before it goes to a client or into a document.
The agents most at risk aren't the skeptics. They're the ones who trust AI outputs because they sound authoritative. Skepticism isn't a barrier to using AI effectively. It's a core part of using it responsibly.
One Thing to Do Differently Starting Now
This mindset shift doesn't happen automatically. It requires a deliberate decision about what kind of tool AI is and what role you're going to play when you use it.
Here's your assignment:
- Pick one task you do repeatedly — something you do often enough that a reliable approach would actually matter. Writing showing feedback emails, drafting offer summaries, updating clients on transaction milestones — pick one.
- Write a job, not a question. Include the context, the output format, the constraints, and the raw material.
- Run it. Evaluate the output as a draft, not a final product.
- Refine the job based on what's missing. Run it again.
That's it. Not a new platform to learn or a subscription to evaluate — a way of thinking about a tool you may already have. Every practical workflow, every time-saving result follows from getting this one thing right first.
- Jason