The alpha, twice a week
That was 2 min.
Get the best new moments in your inbox every Wednesday and Saturday. It's free.
Check your inbox
We sent a link to your address. Click it to finish signing up. Not there in a minute? Look in spam.
or
MTS · 2 min · Oct 22, 2026
Local data gaps make AI answers about officials unreliable
Sunil Rajaraman
Founder & CEO, Hamlet
Theo JaffeeHost · Co-founder & host, MTS
One of the first members of a16z's new media team; earlier hosted the Emergent Ventures-backed Theo Jaffee Podcast.
Sophia DewHost · Host, MTS
About this segment
Sunil says AI tools often get local officials wrong because government data is scattered across inconsistent websites. Hamlet has to build pipelines to scrape, identify speakers and assess sentiment.
From “Why Local Sentiment Kills Data Center Projects | Hamlet CEO”, Oct 22, 2026. Watch the full episode
Thanks. We'll look at and fix what's off.
“you have 19,000 American cities. There's 19,000 different websites. You have 3,000 plus counties, 80,000 special districts.”
Takeaways
- AI data about local officials is still quite inaccurate.
- There are 19,000 city websites, all structured differently.
- Hamlet's pipeline includes scraping, speaker identification and sentiment analysis.
In this episode
Full episode- 14:17Local data centers face different community objections2:38
- 9:56Understanding local power can solve the GovTech business problem2:23
7:36Local data gaps make AI answers about officials unreliable2:12
- 12:14Local governments still rely on outdated, fragile record systems2:08Up next
Transcript
Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.
- 7:39
so what do you think about this upcoming election do you think that ai is going to have a huge influence in the regular voter turnout and how they are informed about making decisions it already it's already happening um i've seen articles and i uh i have to pull the exact source but there has even been articles about um people prompting to see who their local officials are Mm-hmm. and use it as a resource for voting on, you know, local officials at their elections.
- 8:04
The thing I would caution you there is the data about local officials is still quite inaccurate. Why? You know, I can't speculate. I mean, I can only speculate as to what the reasons are, but it's just not an important enough problem probably for somebody at one of the foundation model companies to think about, you know, when it comes to federal or state. It's probably very important. but the data quality gets considerably worse once you get to local.
- 8:31
Interesting. I figured there'd be something that I could easily scrape from. Like, isn't there some county website that gets updated with the latest names and who's in charge and what the latest stuff is? Yeah, you would think this is the original sin of kind of local government is the CMS providers, there's a few of them. They were just not very good pieces of software. And so you have 19,000 American cities.
- 8:53
There's 19,000 different websites. You have 3,000 plus counties, 80,000 special districts. They're all structured in a different way. So, you know, the way we think about how to solve this problem is, you know, we have to, we first had to build the data pipelines across thousands of cities. Then you have to identify speakers. Then you have to assess sentiment. And so, you know, I want to thank our partners.
- 9:16
Like Databricks makes our job a lot easier. We use LakeBase. And, you know, for that, data branching has helped a ton. So I highly recommend that. Assembly, who we do some transcription work with. I want to shout them out, but they've been helpful there. BrowserBase helps us with scraping. Vercel helps us with preview deployments. And so this is not like an easily tractable problem with one model or solution, but it's a hairy problem.