New startup ideas · AI and software · Agent infrastructure

startup idea

Civitrace

Casework agents for US agencies that turn every verified fact into reusable public evidence.

Civitrace runs eligibility and permit casework as a managed agent service for US state, county and city agencies: it reads submitted documents, checks them against source systems, drafts determinations, and hands a human caseworker a decision packet with citations.

AI agent as a service
Government and public sector
$10-100B market
Creates a new category
US first

3/5

venture judge

similar startups, last 2 years (3 all-time)

98%

of 4 nearest real companies still alive

yes

8 matching federal grants and programs

Direction supported by government programs and grants

Test it before you build it

$500 · 6 weeks · 20 prospects

Prove two county human services directors will sign dated, paid pilots for agent-drafted renewal determinations, for $500 in 6 weeks, before any integration is built.

Riskiest assumption · A county human services director will commit to a dated, paid pilot in which an agent drafts eligibility determinations, rather than letting the fear of one wrongful denial, the risk analyst's moratorium scenario, freeze every conversation at the unpaid-demo stage.

1Focus group: who and where

The human services director or SNAP/Medicaid eligibility program manager at a county agency in a county-administered state, whose renewal timeliness is flagged in federal reports and who is heading into renewal season with the same headcount as last year.

where to find 20 · The Digital Benefits Network at Georgetown's Beeck Center, the practitioner community where state and county benefits staff already convene; USDA FNS SNAP timeliness and corrective-action reports crossed with county agency directories in the county-administered states (CA, NY, OH, NC, WI, MN, CO, NJ, VA) to build the 20-county target list; APHSA's ISM conference, the annual health and human services IT event, plus warm intros from Code for America's safety-net practitioner network.

2Sell first, build later

A 90-day pilot in which Civitrace handles document intake and verification for 500 renewal packets and delivers citation-backed determination drafts; the county caseworker signs every determination; work starts within 60 days of the signed data agreement.

the ask · $8 per cleared renewal packet, about $4,000 for the 500-packet pilot, against roughly $30 of loaded caseworker cost per renewal.

a real yes · A signed LOI with a dated start and the security review initiated is a yes that completes when the pilot purchase order issues; compliments, unpaid demos, and 'send us more information' are not.

3Small experiments

The first one attacks the riskiest assumption; each ends with a number that says whether to run the next.

1. Fifteen director pilot-ask calls
$150 · 21 days
Build the 20-county list from FNS timeliness data, reach directors through Digital Benefits Network intros and direct email, and pitch the per-packet price on a 20-minute call. End every call with one ask: a scoping session with their eligibility supervisor in the room and a pilot start date on the table.
keep going if · 5 of 15 directors book a scoping session with their eligibility supervisor within 2 weeks of the call
2. Hand-built decision packet demo
$100 · 7 days
Take the target state's public renewal forms, fill them with synthetic applicant data, and hand-produce 5 citation-backed determination packets using the founders plus an LLM, exactly as the agent would output them. Send the packets to each director 48 hours before their scoping session.
keep going if · 4 of 5 scoping sessions end with the director asking for LOI terms
3. LOI with a dated start
$250 · 14 days
Present a one-page LOI reviewed by counsel: a 90-day pilot covering document intake and verification on 500 renewal packets at $8 per cleared packet, caseworker keeps every signature, start within 60 days, contingent on a signed data-sharing agreement and the county's security review. Count signatures, not enthusiasm.
keep going if · 2 directors sign the LOI with a start date filled in

4Collect a deposit up front

Tesla took $1,000 refundable reservations for the Model 3 and $100 for the Cybertruck before building either: the deposit is the measurement, not the revenue.

no cash yet: take a signed commitment

how · A signed LOI with a dated start, a calendar-blocked kickoff with named caseworkers, and the county's data-sharing and security review formally initiated; a county agency cannot legally or credibly prepay for casework touching applicant records before that review, so the dated commitments are what gets collected instead of money.

what it reserves · A slot in the first three-county cohort and the $8-per-packet founding price locked for year one.

refund · The LOI binds no payment, and either side may withdraw in writing any time before the data-sharing agreement is signed.

target · 2 signed LOIs with start dates from 15 director conversations within 6 weeks

before taking money · Benefits eligibility is federally regulated casework: touch no real applicant PII and draft no live determination before a signed data-sharing agreement and the county's security review, with the caseworker keeping the signature throughout.

Go: build it if

2 signed LOIs with dated starts, security reviews initiated in both counties, and 5 of 15 directors having taken a scoping session.

Kill: stop if

0 LOIs and fewer than 3 scoping sessions after 15 director conversations, or every interested director deferring indefinitely to legal with no date.

5 Scripts to run itoutreach message, landing copy, deposit terms · click to open

outreach message

Your eligibility team is heading into renewal season with last year's headcount, and the FNS timeliness reports already flag your queue. I run a service that reads renewal documents, checks them against source systems, and hands your caseworker a determination packet with citations; your staff keeps the signature on every case. You pay $8 per cleared packet, about a quarter of loaded caseworker cost, with nothing to install. Can I get 20 minutes this week to walk you through a sample packet built on your state's own forms?

landing page

Renewal packets cleared for $8, your caseworker keeps the signature $8 per cleared renewal packet; the 90-day, 500-packet pilot runs about $4,000, under your small purchase threshold. Book a 20-minute scoping call

deposit terms

This letter of intent reserves your county a slot in our first three-county pilot cohort at $8 per cleared renewal packet, locked for year one, with a start date within 60 days. It is contingent on a signed data-sharing agreement and your security review, and no money moves until both are complete. Either party may withdraw in writing before the data agreement is signed.

Would you run this test?

One tap. The yes-share feeds the Demand pillar of this idea's score; nobody sees who answered.

Budgets are out-of-pocket estimates for a team of one to three, US market. Size the deposit to the deal, and check the terms before taking money in a regulated line.

Scorecard

One score that balances how trendy the idea is, the demand for it and its potential for 100x, with competition measured relative to every other idea in the catalog. Recent startup trends first, government priorities second.

Idea Score, 0-100 · raw 45.5 x 1.61

Open

competition: more crowded than 6% of ideas · headwind x0.97

+2.3

government priorities, secondary (8 matching grants)

Trend

Is the wave forming now? 2025-26 entrants vs 2023-24, rounds since 2025, the sector's live-batch direction, the 2026 trend analyst.

Entrants 2025-26 vs 2023-24 (similar companies)3
Rounds announced 2025+ in the sector4
Sector direction (live batch)100
2026 trend analyst25

Demand

Does anyone want it? YC's current RFS, companies already paid for something similar, the operator judge, founders' yes-rate in decks, readers who would run the test.

YC asks for it (current RFS: idea / sector)30
Someone already pays (similar companies, recent / all-time)25
Operator judge: real pain75

100x potential

Can it return a fund? The venture judge (double weight), market-size and moat axes, neighbours still alive, the technologist judge.

Venture judge50
Market size axis67
Moat axis100
Neighbours still alive81
Technologist judge25

Score = 100 x cbrt(Trend x Demand x 100x) x (1 - 0.5 x crowding) + government bonus (max 5), calibrated so the 95th-percentile idea scores 90 (order never changes). A geometric mean: a weak pillar cannot be papered over. Percentiles are among the 272 ideas in the catalog; the terms matched were casework, agencies, verified, fact, reusable, eligibility, permit, managed.

The idea in full

What: Civitrace runs eligibility and permit casework as a managed agent service for US state, county and city agencies: it reads submitted documents, checks them against source systems, drafts determinations, and hands a human caseworker a decision packet with citations. Every verified fact it establishes (income, residency, licence status, inspection result) is written to a consented attestation record the applicant controls and any other participating agency can reuse. Agencies pay per completed determination, not per seat, and Civitrace's field team does the records and integration work at each launch.
Why now: NSF awarded roughly $1.5M to a Safe-OSE center for security-assuring AI agent communication and interoperability standards and another ~$200k to Collaborative Research: SODA: Self-Organizing Decentralized Agents, so the standards work for cross-organisation agent trust is being funded right now; meanwhile the 2026 cohort in this cluster (107 companies) is aimed almost entirely at enterprise IT, with SpaceFlow Technologies, Inc., Decawork and OneCLI all selling agent runtimes to corporate buyers and nobody taking the public-sector determination workload.
Wedge: first customer and entry point: One county human services department drowning in annual benefit renewals: take document intake and verification only, priced per renewal packet cleared, with the caseworker keeping the signature.
Path to 100x: US state and local administrative and IT services spend is in the tens of billions, and determinations are a countable unit priced at a fraction of the loaded caseworker cost. The scaling mechanism is the attestation network: the second agency to accept a Civitrace-verified fact skips the verification entirely, so cost per determination falls as agencies join and a late competitor has to re-verify everything from scratch, which is how the category of a shared public verification fabric gets created.
Ceiling: If agencies refuse to accept each other's attestations for legal reasons, the network never forms and this becomes a per-contract implementation business whose revenue tracks the field team.
Closest real companies, as the generator saw them: SpaceFlow Technologies, Inc. sells a managed runtime for enterprise agents, Decawork an agent control plane for IT teams, OneCLI a sandboxed assistant per employee, Alkera AI reliable data agents. All sell tooling to technical buyers; Civitrace sells the finished determination to a programme director and owns the cross-agency attestation record, which none of them touch.
Main risk: One wrongful denial traced to the agent triggers a state moratorium and every pipeline deal freezes for a year.

Five judges

Each judge scores every idea in the catalog with a named rubric; the venture judge decides whether a card is shown at all (4-5 is venture-grade).

Venture investor
3/5
The cross-agency attestation network is genuinely compounding, but a field team at every launch ties revenue to headcount until it forms.
Bootstrapper
2/5
Per-determination pricing is sound but a field team at every county launch makes revenue track headcount, not margin.
Operator
4/5
Benefit renewal backlogs are a programme director's own problem, priced per determination against loaded caseworker cost with the human keeping the signature.
Technologist
2/5
Document reading against source systems is standard agent work, and the attestation network depends on agency legal acceptance rather than engineering.
Risk
2/5
One wrongful denial triggers a state moratorium, and the attestation network needs legal acceptance that no agency has yet granted.
trends
2/5
Eligibility casework automation was sellable in 2024; the why-now leans on a $1.5M NSF standards grant, which is worth the secondary point at most.

Similar startups in the directory

Companies whose pitch matches most of the idea's terms (casework, agencies, verified, fact, reusable, eligibility, permit, managed): 3 all-time, 0 from the last two years. Same matching as Idea Check.

Sohar Healthyc S23 · 2023 · Healthcare and bioacquired
AI-powered insurance verification to boost patient conversion.
Heymirza500global 500G GA 29 · 2021 · Fintechalive
Developer of a fintech platform designed to deliver tax-efficient employer stipends to a network of caregivers. The company's tools integrate family financial planning and childcare funding, and also provide assistance in corporate policies for parental leave, fertility options, and modern parenting advice for employees, helping working professionals to work towards achieving their financial goals.
Factagoraplugandplay · Agent infrastructurealive
The ultimate fact-checking tool for AI—powered by a reusable knowledge graph.

Run this as an Idea Check →

The generator's reference companies

Real companies the model named as closest when it wrote the card, with their fate. A check mark is a company the radar could verify in its directory.

Public money in this direction

US federal grants, SBIR/STTR awards and open opportunities from the radar's public-money feed, matched to the idea's terms; the sector totals give the context.

grants and programs matching the idea

startup-relevant grants in Agent infrastructure

$16M

awarded in the sector, tracked

I-Corps: Translation Potential of an Intelligent Mobility Platform for Travel Planningawardhigh relevance
National Science Foundation · I-Corps · $50K · posted 2026-08-10
ENT-based Approach to Cervical Phrenic Nerve Stimulation for the Treatment of Central Sleep Apneaawardhigh relevance
NIH / NHLBI · SBIR phase II · $982K · posted 2025-09-01
FIRE-WUI: Agentic Artificial Intelligence for Wildfire Evacuation under Mixed Autonomyawardmedium relevance
National Science Foundation · TIP-CHIPS KTA-5 Disaster mgmt · $2M · posted 2026-08-20
Collaborative Research: Elements: AI-Agentic Digital Twin for Collaborative Building Auditing and Retrofit Simulationawardmedium relevance
National Science Foundation · Special Initiatives, Software Institutes · $300K · posted 2026-07-01
Collaborative Research: Elements: AI-Agentic Digital Twin for Collaborative Building Auditing and Retrofit Simulationawardmedium relevance
National Science Foundation · Software Institutes, Special Initiatives · $300K · posted 2026-07-01
POSE: Phase II: Implementing an Open-Source Ecosystem for the Interdependent Networked Community Resilience Modeling Environment (IN-CORE)awardmedium relevance
National Science Foundation · POSE · $2M · posted 2026-05-28
NeTS: Technological Management of Computer Networksawardmedium relevance
National Science Foundation · Networking Technology and Syst · $675K · posted 2026-05-25
Category II: Transitioning the National Science Data Fabric Pilot into a National Operational Cyberinfrastructure and Service for Democratized, AI-Driven Scientific Discoveryawardmedium relevance
National Science Foundation · NAIRR-Nat AI Research Resource · $9M · posted 2026-05-22

All public money by sector →

Market signal

What the radar sees in Agent infrastructure: new companies by cohort year, the forming YC batch, and outcomes since the February snapshot.

Agent infrastructure · 28 → 64 → 108 → 150 → 171 new companies 2022 → 2026 · 94% aliveYC F26 live: 22 in this cluster, 18% of the batch (was 15% in S26)Since February, of 239 YC companies here: 9 acquired, 4 shut down, 111 rewrote their pitch

Agent infrastructure: companies, trend and grants →

Design attributes

The card is one cell of a designed set: every axis below was chosen before the text was written, and the text had to realize it.

Buyer: Government and public sector
Business model: AI agent as a service
Path to 100x: Creates a new category
Market size: $10-100B market
Capital intensity: Capital-medium (ops, field teams)
Speed to revenue: Revenue in 1-3 years
Technical depth: Real engineering
Go-to-market: Founder-led sales
Moat: Network effects
Geography: US first
Regulation: Heavily regulated
Vibe: Boring business