In partnership with
Hi, I’m media innovation journalist Ulrike Langer and you’re reading News Machines. If this newsletter was forwarded to you becoming a subscriber means you will never miss an issue. If you want to support my work you can spread the word, become a paid subscriber or click on the ad.
This week, a big thank you goes out to all subscribers who filled out my survey about which direction the premium tier of this newsletter should take and to everyone on the free or paid tier who gave me additional input. There is one clear winner for should be next and that is deep dives with operational and actionable details. So that’s what I will be working on first. Stay tuned! The survey is now closed but feedback anytime is always welcome. And now let’s dive into automated civic reporting.
Jonathan Bash wants to bring back civic reporting to news deserts with Locunity
Local government is where decisions are actually made - zoning variances, school budgets, water board appointments. It's also where news coverage disappeared first. A California Bay Area startup called Locunity is trying to fill that gap with AI, and their approach is worth understanding if you're thinking about how far automated reporting can go.
Locunity monitors public meetings - city councils, planning commissions, school boards - and turns them into newsletter-style briefings. The AI scrapes meeting videos, structures what was said, pulls quotes and vote counts, and drafts a report in the style of a Politico newsletter rather than a dry set of minutes. A human editor reviews it before publication, checking names, spellings and numbers.
Sponsored
AI Agents Are Reading Your Docs. Are You Ready?
Last month, 48% of visitors to documentation sites across Mintlify were AI agents—not humans.
Claude Code, Cursor, and other coding agents are becoming the actual customers reading your docs. And they read everything.
This changes what good documentation means. Humans skim and forgive gaps. Agents methodically check every endpoint, read every guide, and compare you against alternatives with zero fatigue.
Your docs aren't just helping users anymore—they're your product's first interview with the machines deciding whether to recommend you.
That means:
→ Clear schema markup so agents can parse your content
→ Real benchmarks, not marketing fluff
→ Open endpoints agents can actually test
→ Honest comparisons that emphasize strengths without hype
In the agentic world, documentation becomes 10x more important. Companies that make their products machine-understandable will win distribution through AI.
CEO Jonathan Bash co-founded the company with Dev Iyer, formerly a tech product manager at Twitch. Bash's own background is in public communications - he's a three-time winner of a Public Relations Society of America award - and he came to this through personal frustration. He lives in a news desert. His local paper in Martinez died in 2021. The East Bay Times, which covers all of Contra Costa County with its more than one million residents across 20 cities, has one civic reporter for 450 elected officials. "They generally only have the time to focus on bribery and other big salacious headlines," Bash told me.
Locunity Reports: A browsable archive of all public meeting reports and videos (all backend screenshots courtesy of Locunity)
The process starts before the meeting. Locunity feeds the system context in advance - speaker rosters, background on the agency - which catches most common errors before the AI ever touches the audio. The system then scrapes the meeting video, structures what was said, and builds out an agenda summary with policy background, quotes, vote counts, and stakeholder positions.
From that it drafts a newsletter with headlines and bullets, structured around what matters, what each side said, and what happens next in the policy process - closer to Politico than to an Otter AI transcript dump. The most common error is name misspellings - Christine with a K versus a C. Quote misattribution happens roughly one in ten times, which is why a human editor checks names, quotes, and numbers before anything goes out.
Locunity currently covers 100 jurisdictions - this week adding Solano County and the remaining cities and special districts in Alameda County. The target market is around 90,000 local government bodies, including city councils, school boards and special districts across the US.
Advocacy groups first, newsrooms second
The paying customers currently are mostly advocacy organizations, not newsrooms. Bash is straightforward about why: The company's immediate goal is survival. Advocacy groups have both the need and the budget. Most small local newsrooms are financially struggling. However, they are next on the priority list because Locunity wants to serve “suburbs and small cities, the ones that are most likely to be news deserts,” Bash told me. Publishers in Locunity’s service area can access Locunity’s platform immediately for free. If they are outside, they will be added to a waitlist.
Coastside News puts the fact that its local meeting coverage is AI-generated front and center and embraces the AI style writing as a differentiator from its human-produced content
One newsroom partnership is already running. Coastside News is a community-owned publisher (a public benefit corporation with local shareholders) of the Half Moon Bay Review and Pacifica Tribune in San Mateo County. It publishes Locunity content under a branded section called "Coastie," clearly labeled as AI-generated. The disclosure goes further: Two Coastside News directors have personally invested in Locunity.
One of them is Rich Klein, the CEO, who responded to my inquiry by email. He describes the editorial process as mostly checking proper names for correct spelling, supplemented by corrections from readers who attended meetings. Of several dozen reports published so far, one had a substantive error - a math problem and a speaker misidentification. Klein considers writing style a question for later, once accuracy is consistent. For now, he says the gap between Coastie's voice and the paper's regular coverage is deliberate: "The big style difference between the Locunity reports and our human-based stories is something of a feature. It helps highlight the fact that these are AI-based stories."
That's a defensible position. Transparency through awkwardness is still transparency. But it also means the primary editorial intervention between AI output and publication is spellchecking - which is a thinner layer than the word "fact-checking" usually implies.
Locunity Policy Radar: Publishers can create custom daily roundups on up to 249 issues and subtopics, as well custom, real-time keywords
Bash wants Locunity to become infrastructure - the layer that every local newsroom plugs into rather than building their own meeting coverage tools from scratch. CT Mirror's Legitalk is doing similar work. When I raised it, Bash agreed that this kind of parallel development doesn't need to stay parallel. I spoke with Angela Eichhorst at CT Mirror a few weeks ago about exactly that work. The question of whether civic AI coverage converges toward shared infrastructure or stays fragmented across individual newsrooms is one of the more consequential open questions in local news right now. Locunity is making a bet on convergence.
The accountability questions haven't been fully resolved. Locunity serves both labor unions and the companies they're negotiating with. It serves advocacy groups on opposing sides of local policy fights. Bash's answer is that structural neutrality - serving everyone - makes bias economically irrational. He acknowledged the limits: Even decisions about what to summarize and how much space to give a topic carry editorial weight. "It won't be perfect," he said.
Contra Costa County Board of Supervisors: A snapshot of a typical "journalistic style" meeting report transcript, agenda and original reporting (left) and a report version with video (right)
What publishers can take from this case study
Investor-publisher overlap is becoming a feature of local news survival, not an exception. The people willing to fund struggling community papers are often the same people funding the tools those papers use. That's worth tracking as a pattern.
Before partnering with an AI vendor, map the financial relationships. Who owns what, who has invested in what, and how does that affect editorial independence. Then disclose it proactively.
If you use AI-generated content, build the disclosure into the product, not the fine print. Coastside's "Coastie" section makes the sourcing visible by design - readers know what they're reading before they read it.
Vendors like Locunity can only become infrastructure if they're both neutral and flexible enough to serve newsrooms with very different editorial standards, audiences, and workflows. That's a harder product problem than covering meetings.