About MoCo Monitor

An independent, AI-processed digest of civic information for Montgomery County, MD — sourced, summarized, and linked back to the public record.

Disclosure: this site is processed by AI

100% of the content on MoCo Monitor is produced by AI reading public county emails. Two models do the work:

Anthropic Claude — reads each email and extracts the structured fields (title, plain-language summary, category, dates, source links).
Google Gemini Embedding — converts each item to a numeric fingerprint used only for grouping near-duplicates and surfacing related items. It does not write content.

No generative AI produces opinion, analysis, or editorial framing. Every item links back to the primary public source so you can verify against the original record.

How the pipeline works, at a high level

Collect. We subscribe to the county's public email bulletins — Council releases, committee agendas, DEP / DHHS / MCDOT alerts, Planning Board notices. Same feed any resident can sign up for.
Extract. Claude reads each email and returns one or more structured items, each with a title, snippet, objective summary, category, key dates, and source links.
Fingerprint. Gemini embeds each item as a vector. Items with near-identical fingerprints across emails are merged so you see one canonical record, not repeats.
Publish. The result is rendered to this static site — no ads, no trackers, everything linkable. The footer shows when this page was last built.

Found a mistake? Please tell us.

AI extraction is useful but imperfect. If a title is misleading, a date is wrong, a source link is broken, or an item is miscategorized:

Email corrections@mocomonitor.example with the item URL and what's wrong.
We review every correction. Fixes typically land within 24 hours.
If the error came from the underlying county email itself (rare but possible), we'll add a note pointing at the primary source rather than silently rewriting.

Roadmap

Things coming soon or being considered — in rough priority order. Have an opinion? Reactions on item pages feed back into what we prioritize.

Spanish (EN / ES). Every item rendered bilingually, toggled from the header.
Personalized preferences + topical digests. Set your district and topics of interest, then get a periodic email or on-site feed scoped to them. Deferred until we know the right shape — inline drawer vs. standalone page, anonymous vs. account-backed.
Public roadmap with +1 voting. This list, but interactive — readers upvote what matters; feeds directly into what we build next.
Server-side feedback aggregation. Today "Useful / Not useful" reactions live in your browser. Move to a small endpoint so signals aggregate across readers.
District tagging on content. Extraction-time tagging so preferences and digests can filter by council district.
Support / Oppose widget (policy items). A second reaction set for policy items — Support / Oppose / Unsure — layered on top of the Useful signal. Off by default; turned on per-item when ready for public review.
Semantic search re-rank. Search is fast today (BM25 client-side). Next, use the existing Gemini embeddings to re-rank top hits so synonyms like "housing" and "real estate" find each other.
WordPress publishing. Mirror each item as a WP post for SEO and longer-form commentary.
Open-data export + public correction log. Transparent downloads and an audit trail of edits.

Principles

Every item cites its public source. If we can't link back, we don't publish.
AI extracts, it does not editorialize.
No ads, no trackers, no selling data. Ever.
Corrections over cover-ups — if we get something wrong, we fix it visibly.
Free forever for residents.