AI for government intelligence

Ask any question
about a place

The answer was already in your data. We built the way to reach it.

$1M
in missed revenue from 21 unpermitted short-term rentals, identified from public records in one town
97
housing units identified in 15 minutes — sites, feasibility scores, revenue projections
<2 min
from new question to verified answer with mathematical models running underneath

Governments collect mountains of data.
Nobody has ever unified it.

Assessor cards. Permits. Zoning codes. Water meter readings. Meeting minutes. Utility filings. Regulatory dockets across five states. Inspection records going back decades. It all exists. It sits in filing cabinets, siloed databases, and scanned PDFs that no one reads systematically.

The information was always there. The intelligence was not.

ArtifexAI ingests it all — links records to parcels, links parcels to geography, links geography to policy. Then it runs mathematical models across the whole and lets anyone ask questions in plain English. Not guesses — verified computation on your own data.

Municipal intelligence
Eight analytical engines.
One question at a time.
For municipalities, each engine turns a different class of public record into a different kind of intelligence. All queryable in plain English.

Who owns your town?

Every parcel classified by resident, absentee, LLC, trust, or seasonal owner.

What percentage of North Chatham is absentee-owned?
92.3% of North Chatham parcels are absentee-owned. Median home 2,694 sf on 23,920 sf lots. The "neighborhood character" argument is being made by people who don't live here.

Where is energy vulnerability?

Heating fuel, building materials, and assessed value combined into an energy risk score for every parcel.

I have $100K for a heat pump pilot. Which homes?
40 resident-owned homes under $600K. Two tiers: electric baseboard and oil-heated. None are STRs. Mapped in geographic clusters for installer efficiency.

Are assessments fair?

Hedonic pricing models predict what each property should be worth. Divergences flagged by inspector, neighborhood, type.

Are there systematic biases in the assessor's work?
Inspector NF shows a 140% hedonic residual, assessing at 2.4x model predictions. 51 properties carry full interior grades with no evidence of inspection.

Where can housing actually be added?

ADU feasibility scored by physical capacity, financial viability, infrastructure, and ownership.

How many ADUs could realistically be built?
1,573 eligible parcels. At 10% adoption: 157 units generating $124K annually. Top corridor: Deer Meadow — 25 properties with 1,500+ sf basements.

Which assessor cards have problems?

Integrity engine audits every card for math errors, phantom renovations, ghost verifications.

Were all inspections actually performed?
3,131 parcels flagged as ghost verifications: full interior grades with no evidence of a site visit. COD 38.26, well above IAAO standard of 20.

What does zoning actually allow?

Digitized bylaws computationally applied to every parcel.

Which parcels gained ADU eligibility under the new law?
847 parcels previously ineligible now qualify under MA Affordable Homes Act. 62% in R20 districts with adequate lots but prior restrictions.

Where is affordable stock disappearing?

Teardown tracker identifies completed and pipeline demolitions.

How many affordable homes have been torn down?
159 completed teardowns at 5.7x value multiplier. Top 20 destroyed $47.2M in affordable value. 835 more in the pipeline.

Who is operating without a permit?

Cross-reference bedroom counts, septic permits, ownership, and listing activity.

Find unpermitted STRs owned by LLCs
21 parcels identified. Bedroom/septic mismatch on 14. Missed revenue: $1M+ annually.
Beyond parcels
The same architecture applied
to every kind of government data.
Community intelligence

595 speakers. 202 topic clusters. Every voice linked to a property.

Meeting transcripts become structured intelligence: who spoke, what they said, how sentiment breaks down, and what property they own. Ask "what are residents saying about affordable housing?" and get 62 mentions across 4 topic clusters with named speakers, their property stakes, and the trajectory of the debate. Look up any speaker and get their address, assessment data, meeting history, and how their claims hold up against the numbers.

Document intelligence

13,000+ municipal files across 40+ committees, structured and queryable

Planning board transcripts, zoning board decisions, select board minutes, committee reports — ingested, indexed, and cross-referenced with property data. Not just searchable. Structured into topic clusters with temporal patterns, sentiment analysis, and speaker identification.

Regulatory intelligence

Cross-jurisdiction policy analysis at scale

For New York State, we analyzed 19,000+ utility commission docket filings across five states to benchmark LMI electrification policy. Stakeholder sentiment, funding mechanisms, cost-shifting evidence — extracted from regulatory proceedings no human team could read in full.

Water & utility compliance

Consumption patterns reveal what permits don't

Cross-reference real-time meter data with property records, septic permits, and irrigation registrations. Bayesian models score the probability of non-compliance. We identified hundreds of unregistered irrigation systems from water usage patterns alone.

Policy simulation

What happens if you change the rules?

Model residential exemption scenarios and see the tax impact on every parcel. Simulate ADU adoption rates and project revenue. Run a geopolitical energy scenario and identify which neighborhoods lose heat first. The digital twin runs the counterfactual before you make the decision.

Case studies
One conversation. A complete strategy.
Real sessions from a single municipality. The system connected parcels, meeting transcripts, speaker records, and property data to produce what no consultant could deliver in six months.
01

"What are residents saying about affordable housing?"

The system searched 595 indexed speakers across 202 topic clusters. It found 62 mentions across 4 major debate threads, identified the key contested projects, mapped sentiment patterns, and surfaced the core finding: near-universal rhetorical support paired with fierce implementation opposition on every specific site.

02

"Show me the neighborhood where the opposition is concentrated."

The system identified the most vocal housing critic — 40 recorded topics across 22 issues — found her property, mapped all 52 parcels on her street, located the Town-owned development site she opposes, and profiled the neighborhood: modest ranches and capes, median $626K, 91% absentee-owned.

03

"Does the data support her arguments?"

Cross-referencing her meeting statements against actual property data, the system found that every home on her street — including hers — violates the density requirements she invokes. She defends "neighborhood character" on a road where 91% of homes sit empty most of the year. The system flagged where she is right (scale, pedestrian safety) and where the data contradicts her claims.

04

"What housing solutions would actually work here?"

The system analyzed every lot size, basement square footage, building footprint, and ownership status. It found 19 homes with convertible basements over 600 sf, 20 lots large enough for detached cottages, and 29 absentee-owned properties sitting empty most of the year.

Result: 39 potential housing units hiding in the existing neighborhood — more than the 20-unit project the town spent 2.5 years fighting over.
05

"I have a $300K grant. How do I optimize it?"

The system ranked every property by conversion feasibility, modeled cost per unit ($45K for basements vs. $200K+ for new construction), identified the 5 best candidates, designed a demonstration cottage, and allocated $40K for replicable program infrastructure.

6 affordable units at $50K each — with a template that scales to 2,400+ eligible properties town-wide.
06

First question to complete strategy: one conversation.

No consultant. No RFP. No six-month study. Community sentiment connected to property data connected to policy design — in real time, from the town's own records.

01

"What are residents saying about energy costs and infrastructure?"

The system found 37 mentions on water and sewer infrastructure (the dominant issue), 12 on budget and debt, 7 on grinder pump equity, and 9 on energy and solar. It identified the most active voices, surfaced the core tension: a sewer project ballooning from $300M to $650M with only 17% of homes connected, and mapped it against actual energy vulnerability by fuel type and neighborhood.

02

"Show me homes getting hit from both sides — oil heat AND no sewer."

The system identified 1,056 oil-heated single-family homes facing dual infrastructure costs. It ranked neighborhoods by squeeze intensity: the town center had 431 oil homes (41% of the total) with the lowest median values. It then isolated 321 modest homes under $800K with vulnerability scores above 60 — the households where costs genuinely hurt.

249 of the most vulnerable homes concentrated in one neighborhood. Average crisis-year energy cost: $1,846 — on top of upcoming mandatory sewer hookups of $15K–$30K.
03

"Show me one street. What's the energy picture for every home?"

The system mapped all 57 parcels on a single road and produced a house-by-house energy portrait: 8 oil-heated homes (avg. built 1941, vulnerability score 78) alongside 28 gas-heated homes (avg. built 1968, vulnerability 50). It identified the starkest contrast: a 1924 home with Fair condition throughout, oil heat, zero permits in 45 years of ownership — three doors from a 2014 gas-heated home worth twice as much.

04

"I have $500K for a heat pump pilot on this street. How do I spend it?"

The system designed a four-tier program: $195K for full oil-to-heat-pump conversions on 4 long-term resident homes (prioritized by vulnerability and tenure), $100K for weatherization on 3 gas homes with ancient building envelopes, $60K for 2 subsidized STR conversions tied to permit compliance, and $145K for monitoring, electrical panel upgrades, and a replication toolkit.

9 homes converted at $55K each — spanning five construction eras from 1800 to 1977. Annual energy savings of $10,600 at crisis pricing. Every conversion documented as a replicable case study for the 1,056 oil homes town-wide.
05

The system sequenced by equity, not efficiency.

The longest-tenured resident — 46 years, zero renovations, Fair condition throughout, oil boiler likely original — goes first. Not the LLC. Not the STR operator. The person who has been there longest and has the least. Energy cost drops from $2,855 to $1,400. A street becomes the story. When town leadership debates scaling town-wide, they drive down one road and see the future.

01

"What's the full picture on the sewer and wastewater situation?"

The system found 83 mentions across 6 active topic clusters spanning 25+ meetings. It surfaced the headline: a sewer project originally estimated at $300M has ballooned to $650M+ with only 900 of 5,300 planned connections complete — 17% of the 2030 target. It identified grinder pump equity fights, nitrogen loading regulation battles, regional capacity diversions, and a PFAS-contaminated drinking water crisis running in parallel.

02

"How many homes are stuck in the same trap — over-assessed, non-conforming lots, still waiting for sewer?"

The system filtered 7,733 parcels down to 888 homes on interior R40 lots that are over-assessed, non-conforming, and still on septic. Total assessed value: $1.52 billion. Median over-assessment: 176% above model-predicted value. More than half have never had a verified interior inspection.

888 homes paying premium taxes for infrastructure they haven't received, on assessments that can't withstand scrutiny, facing five- and six-figure connection costs when the pipes finally arrive.
03

"If we prioritize the next 100 connections for maximum equity, where do we dig first?"

The system ranked all 888 homes by combined pain — tax distortion, construction age, vulnerability score, ownership tenure — and identified the town center as ground zero: 54 resident-owned homes in 1960s subdivisions, median assessed $558K, over-assessed by 136%, on 10,000 sf lots with 60-year-old septic systems. Not one resident from these streets has ever spoken at a board meeting.

04

"I have $2M. Phase the first 30 connections on the four worst streets."

The system designed a three-phase construction plan: 12 connections on two parallel streets sharing a trunk line corridor ($780K), 8 connections completing an adjacent street ($520K), and 10 connections on the longest street's resident core ($700K). It sequenced by tenure — the first shovel goes at a home owned by the same person since 1962. Sixty-three years waiting.

30 connections at $65K each. 83% serving year-round residents. Engineering follows natural gravity flow. Each phase extends the previous trunk line — no orphan mains, no dead-end infrastructure.
05

The system found the silent majority.

The most vocal infrastructure critic has made 209 public appearances. The homeowners on these four streets — the most over-taxed, most under-served, most modestly-housed year-round residents in town — have made zero. This plan speaks for them with data. Not one person from these streets has ever attended a board meeting. The system gave them a voice anyway.

01

"Find every short-term rental in town and show me which ones are compliant."

The system scraped active listings from Airbnb, VRBO, and other platforms, then used computer vision and address matching to link each listing to its actual parcel in the assessor database. It matched listing photos to property card images, cross-referenced bedroom counts against septic permits, and flagged ownership type — resident, LLC, trust, absentee. Out of hundreds of listings, it identified 43 operating with no valid permit or with expired, rejected, or stopped registrations.

02

"Show me the worst one."

The system surfaced a waterfront property whose permit application had been rejected by the Health Agent months earlier — record status: Stopped. The owner signed an affidavit acknowledging $200/day fines for violations. The Airbnb listing was still active. Guest reviews proved continued operation for 10 months after rejection, with names, dates, and trip details documenting paying guests in an uninspected property.

One property. 313 days in violation. $62,600 in accumulated fines. Without the system, nobody in town government knew she was still operating.
03

"Are there other patterns?"

The system found a second category: properties operating under pre-existing bed-and-breakfast licenses that predate the STR regulation. One had a 2017 B&B license, an active food establishment permit, 32 reviews, and zero STR registration. Operating through a trust. The owner knew they were running a commercial hospitality business — they just never registered under the new rules. Two different violation types, both invisible without cross-referencing permits, listings, and property records.

04

"How much revenue is the town losing?"

The system calculated exposure across all 43 confirmed violations: permit fees, room excise tax, inspection charges, and statutory fines. It cross-referenced bedroom counts against septic capacity — 14 of 21 LLC-owned properties had bedroom/septic mismatches, meaning the listed occupancy exceeded what the property was designed to handle. That is both a revenue problem and a public health problem.

Estimated missed annual revenue: $1M+ across permit fees, room tax, and inspection charges. Conservative settlement value on accumulated fines: $86K–$129K from this one batch alone.
05

The system closes the gap between rejection and enforcement.

A health agent rejects a permit. The applicant keeps operating. Without automated monitoring, nobody follows up — for months, sometimes years. The system watches continuously: matching new listings to parcels, checking permit status, flagging guest reviews that prove occupancy. One click on the dashboard gives the enforcement officer the rejected application, the active listing, the reviews, the assessor card, the AI photo match, and the fine calculation. The entire case, ready to send.

"This totally changes the paradigm around what's possible through government intervention."
State Energy OfficialNY Dept. of Public Service
"It stopped me in my tracks. I'm not technical and I'm not easily impressed."
Affordable Housing Board MemberMashpee, MA
"You've done this in a much more useful way than we've been able to do it so far."
Policy AnalystNYSERDA
Solutions
Same platform. Different questions.
Municipalities

Your town's missing intelligence layer

Housing analysis, assessment integrity, STR compliance, energy planning, and policy simulation — from the data you already collect.

Annual subscription
State Agencies

Parcel-level precision for statewide programs

Target LMI investments at individual homes instead of census tracts. Measure reform outcomes with real data at the resolution where policy happens.

Engagement-based pricing
Federal & Defense

Digital twin for installation communities

Military housing requirements, BAH market modeling, infrastructure condition assessment across privatized housing portfolios.

Active DoD engagements
Questions government teams ask us
What kinds of questions can we ask it?

Anything that touches your data. "Where can we add housing?" "Are our assessments fair?" "Who spoke against this project and what property do they own?" "If we adopt a 25% residential exemption, what happens to every parcel?" "Which homes should get heat pumps first?" "Where should we dig sewer next?" The system cross-references parcels, meeting transcripts, permits, zoning, ownership, and utility data — so the question can span any combination of those.

Can it help with policy decisions — not just data lookups?

That is the core use case. The system runs policy simulations: model exemption scenarios and see the tax impact on every parcel, project ADU adoption rates and revenue, sequence infrastructure investments by equity impact, and design grant programs optimized to a specific budget. It connects community sentiment to property data to policy design in one conversation.

Does the AI actually do the math, or just guess?

Every answer runs through mathematical models underneath: hedonic pricing, assessment ratio analysis, feasibility scoring, displacement tracking, clustering algorithms. The conversational interface makes it easy to ask questions in plain English. But the answers come from verified computation on your actual data — not generated text.

Can our staff use this without technical training?

Yes. You ask questions in plain English. "Show me all properties on Elm Street" works. "Which neighborhoods have the worst assessment quality" works. The system handles the data joins, the spatial queries, and the model runs. If your staff can type a question, they can use it.

What data do you need from us?

We start with publicly available records: assessor databases, GIS parcels, zoning bylaws, and permit records. Most municipalities already publish this. For deeper analysis we can integrate CAMA exports, water meter data, meeting minutes, and inspection records — all data you already collect. No new data gathering needed.

How is this different from hiring a consultant?

A consultant takes months and delivers a static report. This is a living system you can query any time. The housing question you ask today becomes the STR compliance question tomorrow and the energy targeting question next month — same data, same platform, no new engagement. And when your data updates, the answers update with it.

Does it work with our existing systems?

We ingest data from standard municipal formats — CAMA systems like Vision, ProVal, and AssessPro, GIS exports, OpenGov records, utility billing systems, and PDF meeting minutes. We do not replace your existing systems. We unify what they already contain.

How long does setup take?

Two to three weeks for a town under 10,000 parcels. The first week is data ingestion and parcel linking. The second is model calibration and quality assurance. By week three, you are asking questions.

Can we see it working before we commit?

Yes. We run a live demonstration on real municipal data — real parcels, real assessments, real meeting transcripts. Bring your hardest question. Book a 30-minute demo.

Bring your hardest question

15 minutes. Live data. No slides.

See it live

Privacy Policy

Effective September 2024 · Last updated January 2025

1. Introduction

ArtifexAI ("we", "us", or "our") is committed to protecting the privacy of our customers' data. This Privacy Policy describes how we collect, use, disclose, and safeguard information when you use our platform and services (collectively, the "Service"). This policy applies to all users of ArtifexAI, including government agencies, organizations, and individual users.

2. Information We Collect

Account Information: Name and contact information, organization name, account credentials, and billing information.

User Content and Data: Documents and files you upload, queries and interactions within your instance, data you designate for analysis, reports and analyses you create, and user preferences.

Usage Information: Log data (IP addresses, browser type, device information), access times and dates, and usage metrics for performance and billing purposes.

Cookies: We use cookies and similar technologies to maintain user sessions, remember user preferences, and provide security features.

3. How We Use Your Information

Service Delivery: Provide and maintain the Service, process your queries, authenticate users, generate reports, and provide customer support.

Service Operations: Maintain service performance and reliability, identify and fix technical issues, and generate usage reports for your organization.

Security and Compliance: Detect and prevent security incidents, monitor for unauthorized activity, comply with legal obligations, and enforce our Terms of Service.

4. Data Protection Commitments

Customer Data Isolation: Each customer's data is logically segregated within our systems. Your proprietary data is never accessible to other ArtifexAI customers.

No Training on Customer Data: We do not use your proprietary data, queries, search patterns, or usage behaviors to train AI models or improve services. Your documents are not used for machine learning training. Your queries are not used to improve models for other customers. Your usage data is not aggregated with other customers' data. Your data is used exclusively for providing services to your organization.

Public Data vs. Proprietary Data: ArtifexAI aggregates publicly available government data, which may be used to improve our data collection and indexing systems. Any non-public data you upload or create remains your exclusive property, is protected by access controls, is never shared with other customers, is not used to train models, and can be deleted upon your request.

5. How We Share Your Information

We do not sell your data. ArtifexAI does not sell, rent, or trade your personal information or proprietary data.

Service Providers: We may share information with third-party service providers who assist in operating our Service, including cloud infrastructure providers (Amazon Web Services) and payment processors. Service providers are contractually obligated to protect data.

Legal Requirements: We may disclose information when required by law, including in response to valid subpoenas or court orders. Where legally permissible, we will notify you of such requests.

6. Data Security

We implement security measures including encryption of data in transit and at rest, access controls and authentication mechanisms, security monitoring and logging, and regular security assessments.

7. Data Retention

We retain your information for as long as your account is active. Upon account deletion, we delete your proprietary data within 30 days. Backup copies are purged within 90 days. Some information may be retained if required by law.

8. Your Privacy Rights

You have the right to access your personal information, request a copy of your data, update your account information, request correction of inaccurate information, and request deletion of your account and data. To exercise these rights, contact us at russ@artifexai.io. We will respond within 30 days.

9. Government-Specific Considerations

We understand that government agencies may be subject to public records laws. Upon receiving a public records request affecting your data, we will notify you promptly, cooperate with your legal review process, and follow your instructions regarding disclosure. For government customers, we can provide additional privacy documentation upon request, including data processing agreements and privacy impact assessment support.

10. Contact

For privacy-related questions or requests: russ@artifexai.io