There’s no shortage of potential local content. But the time it takes to pull news sources, review city council agendas, and look at zoning applications and permit filings stops most people before they get started.
Houston Realtor Katie Day built a system to make that process easier, using AI to uncover and summarize the latest news, then deciding what stories are worth covering for her audience.
Day leads the #MOVEMETOTX real estate team at REAL and built Houston Unlocked into a 46,500-follower account by covering local news, real estate trends, and new developments across Houston.
In a recent conversation with The Broke Agent, Day broke down how she uses AI to do more research without handing over the most important part of the job: deciding what matters to the people of Houston.
Here’s how the system works and why the human element is still the most valuable part of it.
Let AI Handle Research, Not Story Selection
Day uses ChatGPT and Claude for research and scripting. Both tools can search the web and pull up news stories on their own. The choice of what to cover stays with her. She keeps that part of the process manual for a few reasons:
- She doesn’t post about every story she comes across, so a stack of AI-suggested articles doesn’t save her time if most of them get tossed
- She wants to read the original source herself to understand what’s happening in Houston, not skim a summary
- AI search results aren’t always current. Even when she asks a tool to skip anything older than five or seven days, older stories still turn up
In her words:
“I don’t do that because I don’t post about everything, one. And two, I actually want to read the stuff and read the articles and check them out because I want to know what’s going on. And then three, sometimes I find that when you use AI to search, it doesn’t necessarily pull all of the most recent stuff.”
Anyone who’s used AI to assist with content creation and topic research is probably nodding their head right now. Day is aware of AI’s limitations and has learned to play to its strengths.
She hands off scripting and summarizing once she already knows what she wants to cover. She’ll give a tool bullet points from a news article or a city council update and ask for a script draft, then edit it into her own voice.
BAMx members get access to a system built around this same idea, made up of three parts:
- A weekly content playbook, with customizable templates, that lays out what to post and when
- Live content audits with real-time feedback, examples, and Q&A with The Broke Agent
- Around-the-clock access to the Hyperlocal AI Content tool, which pulls hyperlocal news stories and turns them into ready-to-use scripts
Together, these three pieces give agents a research and reporting system similar to Day’s, already built and ready to use. Sign up here for instant access.
Train AI to Monitor City Council Agendas & Permit Filings
Local governments produce a steady pile of paperwork: planning agendas, plat filings, zoning notices. Only a small share of it turns into a story. Sorting through that pile by hand, every week, isn’t a good use of an agent’s time.
This is where AI, and agentic AI in particular, can shine.
Day has Claude scanning multiple city sources at once for this reason:
“I have it searching the Houston city council websites for planning agendas and plat filings and different stuff like that and giving me kind of a summary of the things that are happening because there’s so many different things that are being filed and talked about and most of them aren’t newsworthy, but I want it to kind of comb through them and also remind me when the meetings are so that I can go look at the physical agendas themselves just cuz, again, AI might miss something that’s actually a big deal to the people of Houston.”
This is a process shaped by experience with Claude, etc. Day treats the AI summary as a starting point and checks the primary documents herself too.
She has the AI remind her when city council meetings are scheduled, so she can go look at the actual agendas on her own to find topics of interest to her audience.
Because unlike Day, AI doesn’t scan those filings with the perspective of a Houston resident.
For agents outside Houston, the same setup could point at a city council or a county permit office instead.
Here are the steps to set up a Claude agent that scans city council websites for planning agendas, etc.
- Identify your source pages: council meeting agendas, planning/zoning commission agendas, and plat or subdivision filing reports.
- Check whether your city uses Legistar, Granicus, or CivicPlus. Many do, and these often have search or RSS built in, which saves the agent from having to parse raw PDFs.
- Write one clear prompt for the scheduled task that says what to check, what counts as new since the last run, and what you actually care about (specific districts, zoning types, project names, dollar thresholds).
- Set the schedule to match how often those sources actually post, rather than checking daily out of habit.
- Pick where the summary goes: email, a message, or a doc you’ll actually open.
- Run it for a couple of cycles and compare the output against the real agendas before you trust it fully.
And even if the AI agent’s output matches the real agendas, the summary will still need your human eye and perspective to judge which news items are worth including in your content.
AI Can Find the Story. You Make it Better.
Day’s system works because she doesn’t let AI take over the human side of content creation. It helps her monitor sources, process information and move faster, but it doesn’t decide what Houston residents need to know.
Your knowledge of the neighborhood, conversations with clients and understanding of what matters locally are what turn local news and government filings into useful content. So while AI can surface a zoning proposal, you can explain how that proposal could affect traffic, development, home values or daily life.
Use AI to help you become faster, more informed and more consistent. But don’t give up the human judgment that make your content worth following.
