New startup ideas · AI and software · Vertical AI agents

startup idea

Permitrun

Machine-run permitting and interconnection review for the agencies that hold up infrastructure.

Permitrun is a subscription system of record for public bodies that review infrastructure: state environmental agencies, utility commissions and grid operators running interconnection studies and NEPA-style reviews.

Software subscription
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

$2,000 · 6 weeks · 15 prospects

Prove for $2,000 in 6 weeks that two state utility commissions will sign a dated $9,500 pilot MOU for machine-drafted, staff-edited small-generator interconnection determinations.

Riskiest assumption · A state utility commission's engineering staff and general counsel will approve, in writing and within one budget cycle, a pilot in which a model drafts legally challengeable interconnection determinations that staff edit and sign.

1Focus group: who and where

The engineering division director or lead interconnection engineer at a state public utility commission whose small-generator interconnection docket shows a pending queue measured in months, together with the commission general counsel who must bless any drafted determination before staff can sign it.

where to find 15 · State commission staff directories published on every PUC website (names, titles and emails of engineering and legal staff are public record); the DOE i2X Interconnection Innovation e-Xchange community calls, where commission staff and utility interconnection engineers already meet; IREC's interconnection working groups and BATRIES materials, which the same staff use for study-process reform; NARUC's Staff Subcommittee on Electricity roster for warm second contacts.

2Sell first, build later

A 90-day pilot on the commission's small-generator interconnection queue: Permitrun (founders hand-in-the-loop) drafts every Level 1 and Level 2 determination from the commission's own precedent and study math, staff edits and signs each one, cycle time measured against the trailing 12 months, starting within 60 days of signature.

the ask · $9,500 per 90-day pilot, invoiced at the dated start, sized to fit under most states' small-purchase authority so no competitive bid is required.

a real yes · A signed pilot MOU or purchase order with a dated start and a named queue is a yes; letters of support, innovation-team enthusiasm, and invitations to present to the commissioners 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 staff calls on drafted determinations
$200 · 10 days
Pull the 15 commissions with the longest small-generator queues from their public dockets, email the lead interconnection engineer directly, and run 20-minute calls with a fixed script: walk their current determination workflow, then ask whether they would edit and sign a model-drafted determination and what their counsel would need to allow it. One founder runs all calls; log every answer in a shared sheet.
keep going if · 8 of 15 staff say they would edit and sign a model draft, and at least 5 describe a counsel-approvable path that keeps staff signature and needs no rulemaking
2. Hand-built draft on a real docket
$300 · 12 days
Take one completed small-generator interconnection case from a public e-docket in a target state and hand-write the drafted determination Permitrun would have produced, citing the commission's own prior decisions and study math. Show it side by side with the actual order to the 6 warmest engineers from experiment 1 and ask what they would have changed and how long the blank-page version took them.
keep going if · 4 of 6 reviewers say the draft would have saved them 2 or more hours on that case and forward it to a colleague or supervisor unprompted
3. Counsel gate and pilot MOU
$1,500 · 20 days
Ask the 3 warmest commissions for a 45-minute meeting that includes the general counsel or a hearing examiner, present the staff-edits-and-signs workflow with the experiment 2 artifact, and put a one-page 90-day pilot MOU on the table: Permitrun drafts every determination in the small-generator queue, staff edits and signs, $9,500 priced to fit under common state small-purchase thresholds. Travel to two commissions in person.
keep going if · 2 of 3 general counsel identify an approvable path in the meeting, and 2 commissions sign the pilot MOU with a dated start

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 · No cash deposit: a state commission cannot credibly prepay a vendor before a procurement instrument exists, so collect a pilot MOU signed by the executive director or chief of staff under the commission's small-purchase authority, naming the queue, the staff lead, and the start date, with the $9,500 fee invoiced at start. set up: Stripe Invoicing ↗

what it reserves · One of two Q4 2026 pilot slots, a founder on site with the review staff for the first two weeks, and the commission's own precedent corpus loaded first.

refund · Either party may cancel with 15 days written notice before the dated start, and no fee is owed if the pilot never begins.

target · 2 signed pilot MOUs with dated starts from 15 commissions within 6 weeks

before taking money · This is government procurement: follow each state's small-purchase and gift rules, and keep signature authority on every determination with commission staff until counsel approves anything more.

Go: build it if

2 commissions sign the $9,500 pilot MOU with a dated start, and at least 8 of 15 staff interviews plus 2 of 3 general counsel affirm a staff-edited, staff-signed workflow needs no rulemaking.

Kill: stop if

0 signed MOUs after all 15 commissions are reached, or all 3 general counsel say a model-drafted determination cannot proceed without rulemaking or legislation.

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

outreach message

I have been reading your small-generator interconnection docket and the pending queue behind it. I am building software that drafts each determination from your commission's own past decisions and study math, so your reviewers edit a draft instead of starting from a blank docket - staff keeps the signature on every order. Before writing production code I am talking with engineering staff at 15 commissions about what counsel would need to allow this. Could I get 20 minutes with you this week?

landing page

Drafted interconnection determinations your staff edits, signs, and stands behind. $9,500 for a 90-day pilot on your small-generator queue, priced under small-purchase authority. Book a 45-minute review with your engineering lead and general counsel.

deposit terms

No payment is collected today: your signature on this pilot MOU reserves one of two Q4 2026 pilot slots for your small-generator queue, with the $9,500 fee invoiced only at the dated start. Either party may cancel with 15 days written notice before the start date. The pilot begins within 60 days of signature and runs 90 days, with staff editing and signing every determination.

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 37.5 x 1.61

Warm

competition: more crowded than 20% of ideas · headwind x0.90

+4.1

government priorities, secondary (58 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)23
Rounds announced 2025+ in the sector25
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)0
Someone already pays (similar companies, recent / all-time)25
Operator judge: real pain25

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 judge75

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 machine-run, permitting, interconnection, review, agencies, hold, infrastructure, subscription.

The idea in full

What: Permitrun is a subscription system of record for public bodies that review infrastructure: state environmental agencies, utility commissions and grid operators running interconnection studies and NEPA-style reviews. It trains on decades of decisions, comment records and engineering studies, and runs learned surrogate models that pre-compute study outcomes so a reviewer edits a drafted determination instead of starting from a blank docket. Developers file into the same system, so both sides of a review work off one live record rather than PDFs.
Why now: Operating regulated physical systems with AI is where money is going right now - Edviro (yc S26) is building AI that operates energy infrastructure and Skymerse (yc S26) is doing autopilot for flight operations - and federal science funding is backing the underlying knowledge machinery, including an NSF I-Corps award for an AI-powered knowledge management system for nuclear fusion; meanwhile no YC Fall 2026 RFS theme covers this cluster, so the public-sector side is uncrowded.
Wedge: first customer and entry point: One state public utility commission's small-generator interconnection queue, where the backlog is measurable, the study math is standardised, and one founder can sit with the review staff for a quarter.
Path to 100x: Infrastructure review and public capital program administration in the US is a $10-100B spend, and this creates a category that does not exist today: permitting operations as software rather than case management. Each completed review adds precedent and calibration data, and once developers file into the same record across states the two-sided pull makes a second system pointless, which is a real network moat in a market that normally has none.
Ceiling: Agency budgets and multi-year procurement could cap annual contract values low enough that even category leadership is a few hundred million in enterprise value.
Closest real companies, as the generator saw them: Edviro operates the energy asset, not the approval that gates it; Locke works on how companies and governments transact; Chromie helps vendors win contracts. Permitrun differs by being bought by the reviewing body and holding the decision record itself.
Main risk: Agencies will not let a model draft a legally challengeable determination, and every adoption stalls in general counsel review.

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
Holding the decision record for both sides of interconnection is a real two-sided moat, but agency budgets and multi-year procurement cap contract values.
Bootstrapper
2/5
Software margins are fine, but selling a legally challengeable determination engine to a state utility commission is a multi-year procurement slog.
Operator
2/5
Agencies will not let a model draft a legally challengeable determination, so adoption stalls in general counsel review despite a measurable interconnection backlog.
Technologist
4/5
Learned surrogate models that pre-compute interconnection study outcomes from decades of engineering dockets is deep expertise no generalist agent vendor can assemble.
Risk
2/5
One state utility commission wedge in a market where general counsel can veto a model drafting a legally challengeable determination.
trends
2/5
Interconnection backlogs were just as painful in 2023, and the card itself concedes no current YC RFS covers the public-sector reviewer.

Similar startups in the directory

Companies whose pitch matches most of the idea's terms (machine-run, permitting, interconnection, review, agencies, hold, infrastructure, subscription): 3 all-time, 1 from the last two years. Same matching as Idea Check.

Bountyspeedrun SR007 · 2026 · Commerce and marketplacesalive
Bounty lets founders and small business owners hire AI agents and specialised teams, paying for completed work instead of tool subscriptions.
Sparkyc W24 · 2024 · Climate and energyalive
AI-powered workflows for energy infrastructure
Pacesyc S22 · 2022 · Climate and energyalive
Agentic AI for Power Projects and AI Infrastructure

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 Vertical AI agents

$135M

awarded in the sector, tracked

I-Corps: Translation Potential of Edge-Artificial Intelligence (AI) Personalized Risk Prediction for International Travelersawardhigh relevance
National Science Foundation · I-Corps · $50K · posted 2026-08-19
I-Corps: Translation Potential of Safer, Longer-Life Lithium-Ion Energy Storage for Critical Infrastructureawardhigh relevance
National Science Foundation · I-Corps · $50K · posted 2026-08-19
I-Corps: Translation Potential of a Resilient, Low-Cost Internet-of-Things (IoT) Monitoring Platform for Precision Agricultureawardhigh relevance
National Science Foundation · I-Corps · $50K · posted 2026-08-19
SBIR Phase I: Wireless Sidestep Network: Harnessing Sub-terahertz to Enable New Degrees of Freedom in AI Data Center Architecturesawardhigh relevance
National Science Foundation · SBIR Phase I · $305K · posted 2026-08-18
I-Corps: Translation potential of a Light Detection and Ranging (LiDAR)-based bridge monitoring and inspection support technologyawardhigh relevance
National Science Foundation · I-Corps · $50K · posted 2026-07-24
SBIR Phase II: Air-Breathing Electric Propulsionawardhigh relevance
National Science Foundation · SBIR Phase II · $1M · posted 2026-07-21
VINES: Track 1: U.S-Ireland R&D Partnership: Synchronized Mobile and Stationary Sensor Networks for Real-Time Bridge Health Assessment (SYN-Bridge)awardmedium relevance
National Science Foundation · Use-Inspired NextG, GVF - Global Venture Fund · $700K · posted 2026-07-30
Learning and Optimization-Powered Quickest Change Detection: Theory, Algorithms and Applications for Power Systemsawardmedium relevance
National Science Foundation · EPCL: Energy, Power, Control, · $600K · posted 2026-07-10

All public money by sector →

Market signal

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

Vertical AI agents · 159 → 261 → 364 → 411 → 353 new companies 2022 → 2026 · 93% aliveYC F26 live: 25 in this cluster, 20% of the batch (was 25% in S26)Since February, of 698 YC companies here: 19 acquired, 9 shut down, 219 rewrote their pitch

Vertical AI agents: 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: Software subscription
Path to 100x: Creates a new category
Market size: $10-100B market
Capital intensity: Capital-light (software margins)
Speed to revenue: Revenue in 1-3 years
Technical depth: Deep tech: ML, hardware, bio
Go-to-market: Founder-led sales
Moat: Network effects
Geography: US first
Regulation: Heavily regulated
Vibe: Hot space