New startup ideas · AI and software · Horizontal AI assistants

startup idea

Provenant

An agent runtime that turns every AI action into regulator-admissible evidence.

Provenant is an assistant for licensed professionals in supervised industries (broker-dealers, insurers, pharma commercial teams) that drafts, sends and acts, with each step passed through a jurisdiction rule pack and written to Provenant's own write-once retention infrastructure with cryptographic attestation.

AI agent as a service
Enterprise
$1-10B market
Creates a new category
Global from day one

3/5

venture judge

similar startups, last 2 years (9 all-time)

81%

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

$1,000 · 5 weeks · 30 prospects

Proves for $1,000 in five weeks that broker-dealer compliance officers will pay $2,000 up front to pilot AI email drafting that runs through their own supervision queue and existing archive.

Riskiest assumption · A broker-dealer compliance officer will approve and pay this quarter for AI-drafted client communications because the supervision record makes them safe, without waiting for Provenant to own certified retention infrastructure, since the pilot journals into the archive the firm already trusts.

1Focus group: who and where

The Chief Compliance Officer at an independent US broker-dealer with 25-200 registered reps, whose reps are already pasting client emails into ChatGPT and whose next FINRA exam will ask how those communications were supervised and retained

where to find 30 · FINRA's public firm directory and Form BD filings filtered to independent broker-dealers with 25-200 reps (pull 30 CCO names), the National Society of Compliance Professionals (NSCP) member community and its fall National Conference, and the Financial Services Institute (FSI) member firm list of independent broker-dealers

2Sell first, build later

A 90-day pilot for five producing reps: AI-drafted client emails that cannot send until they clear the firm's supervision queue, with every draft journaled to the firm's existing Smarsh or Global Relay archive, live within 30 days of signing.

the ask · $2,000 per pilot, invoiced up front and credited against a $125 per seat per month subscription at conversion.

a real yes · Money received against the pilot invoice plus a start date on the calendar counts as yes; compliments, 'our reps would love this', unpaid trials, and requests to reconnect after exam season do not.

3Small experiments

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

1. Fifteen CCO problem calls
$200 · 10 days
Pull 30 CCO names from the FINRA firm directory and FSI member list. One founder emails and cold-calls until 15 conversations are booked, asking one question: what happens today when a rep wants to use AI on a client email, and who has to approve it. Log whether AI drafting is banned, tolerated unsupervised, or already archived.
keep going if · 8 of 15 CCOs say AI drafting is currently banned or happening unsupervised at their firm and ask to see the supervision queue
2. Mock exam-production packet demo
$500 · 10 days
Build a clickable supervision-queue demo in Figma and a 10-page sample exam-production packet: flagged AI drafts, reviewer actions, timestamps, and journaling confirmations into a Smarsh archive. Walk 10 CCOs from experiment 1 through both on a 20-minute screen share and ask what their examiner would say to the packet.
keep going if · 5 of 10 CCOs forward the packet to outside counsel or ask for pilot terms
3. Close two paid pilots
$300 · 14 days
Email written pilot terms to every CCO who asked: $2,000 up front, 90 days, five producing reps, AI drafts that cannot send until cleared in the queue, everything journaled to the firm's existing archive, live within 30 days. E-sign the two-page agreement and invoice by ACH the same day.
keep going if · 2 pilot agreements signed with invoices paid and start dates on the calendar

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.

$2,000

per prospect, refundable

how · A paid design-partner pilot invoiced up front: the CCO signs a two-page pilot agreement by e-signature and the firm pays the $2,000 invoice by ACH before onboarding starts; at this size an invoice is the one instrument a broker-dealer can pay without triggering a full procurement cycle. set up: Stripe Invoicing ↗

what it reserves · One of five October pilot slots, five-rep onboarding, and the $125 per seat price locked for the first year.

refund · Fully refunded if the pilot has not started within 30 days of signing or if the firm's counsel declines after reviewing the agreement.

target · 2 paid pilots from 15 CCO conversations within 35 days

before taking money · Have a securities attorney review the pilot agreement, never represent the pilot as satisfying SEC Rule 17a-4 or FINRA Rule 3110 on its own, and route every draft through the firm's existing approved archive until Provenant's retention estate is certified.

Go: build it if

2 pilots paid at $2,000 with start dates inside 30 days, and at least 8 of 15 CCOs confirming AI drafting is blocked or unsupervised today for supervision reasons.

Kill: stop if

After 15 CCO conversations and 10 packet walkthroughs, zero paid pilots and fewer than 3 requests for written pilot terms.

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

outreach message

Your reps are already pasting client emails into ChatGPT, and none of it reaches your review queue or your archive. I am building a drafting assistant where every AI draft lands in your supervision queue and journals to your existing Smarsh or Global Relay archive before it can send. I am running five paid 90-day pilots this quarter. Do you have 20 minutes this week to see the queue and a sample exam-production packet?

landing page

AI drafting your reps can use and your examiners will accept $2,000 reserves a 90-day, five-rep pilot journaled to your existing archive, live in 30 days Book a 20-minute walkthrough and reserve one of five pilot slots

deposit terms

$2,000 reserves one of five design-partner pilot slots starting in October 2026: five reps, AI email drafting under your supervision queue, every draft journaled to your current archive, and the $125 per seat price locked for year one. Fully refunded if the pilot has not started within 30 days of signing or if your counsel declines after review. The fee is credited against your first-year subscription.

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 25.2 x 1.61

Warm

competition: more crowded than 32% of ideas · headwind x0.84

+3.3

government priorities, secondary (26 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)0
Rounds announced 2025+ in the sector0
Sector direction (live batch)0
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)75
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 axis33
Moat axis80
Neighbours still alive1
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 runtime, action, regulator-admissible, evidence, assistant, licensed, professionals, supervised.

The idea in full

What: Provenant is an assistant for licensed professionals in supervised industries (broker-dealers, insurers, pharma commercial teams) that drafts, sends and acts, with each step passed through a jurisdiction rule pack and written to Provenant's own write-once retention infrastructure with cryptographic attestation. A rep signs up per seat and works normally; the compliance officer gets a supervision queue and an export a regulator will accept without argument. Provenant owns the retention estate in each jurisdiction it serves, which is why it can attest rather than promise.
Why now: NSF is funding 'Collaborative Research: RI: Conditioning Uncertainty on Epistemic Context in Human-AI Communication' at ~$941k, which is the calibration problem a supervision queue depends on. Meanwhile AirCaps (yc F25) and COACH (yc S26) are building conversation copilots for sales reps with no supervisory record at all, and Solari AI died in 2024 giving non-technical staff agents that 'work right out of the box' with nothing accountable behind them.
Wedge: first customer and entry point: US broker-dealer branch offices, one workflow: client email and message drafting under a supervision review queue, sold seat by seat to producing reps whose firm already retains communications.
Path to 100x: Supervision and regulated-communications tooling is a $2-8B market that expands as agents start acting rather than suggesting, and Provenant prices per licensed professional in a population of millions. Distribution is the compounding asset: seats spread bottom-up inside a firm, firm-wide policy then makes the runtime mandatory, and OEM deals put it under other regulated desktop tools until it is the default place regulated AI actions get written.
Ceiling: Regulators publish a simple logging schema every vendor implements natively, and evidence-grade recording becomes a checkbox feature rather than a runtime anyone pays for.
Closest real companies, as the generator saw them: AirCaps and COACH assist the same conversations but for unregulated sales floors, with no retention, no rule pack and no attestation. Caretta and Nomi do realtime sales call assistance and stop at the call. Provenant's product is the record, not the suggestion.
Main risk: Building certified retention and attestation infrastructure across jurisdictions takes three years, and a compliance archiving incumbent bolts a rule pack onto an existing estate in one.

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
Per-seat pricing across millions of licensed professionals is real, but a $2-8B market and three years of retention buildout limit the power-law upside.
Bootstrapper
1/5
Owning certified retention infrastructure in every jurisdiction is a three-year, eight-figure build before the first broker-dealer seat pays.
Operator
4/5
Broker-dealers already pay for supervision and retention, and the compliance officer owns that budget, though the three-year attestation build invites a faster archiving incumbent.
Technologist
4/5
Owning certified write-once retention plus per-jurisdiction rule packs is three years of unglamorous infrastructure no conversation copilot like AirCaps can bolt on.
Risk
4/5
Owns certified retention estate per jurisdiction and sells seat by seat to producing reps, so no single firm or channel gates it.
trends
2/5
A three-year R&D build into certified retention races bolt-on archiving incumbents, and its why-now leans on a $941k NSF grant, secondary at best.

Similar startups in the directory

Companies whose pitch matches most of the idea's terms (runtime, action, regulator-admissible, evidence, assistant, licensed, professionals, supervised): 9 all-time, 0 from the last two years. Same matching as Idea Check.

FutureClinicyc F24 · 2024 · Healthcare and bioalive
Personal doctors through chat
Freestyleyc F24 · 2024 · Horizontal AI assistantsalive
The AI Assistant for Gmail
Asteroidef EF London 2024 · 2024 · Healthcare and bioalive
Runtime supervision for AI agents.
DirectShiftsyc S19 · 2019 · Healthcare and bioalive
Workforce platform to recruit, license, credential & manage clinicians
Modern Healthyc W18 · 2018 · Healthcare and bioalive
A mental health benefits platform for employers.
OncoBrainplugandplay · Healthcare and biounchecked
The Only Oncology Clinical Reasoning Engine. Turning intelligence into expertise at the speed of care.
株式会社mediVRplugandplay · Healthcare and biounchecked
mediVR is a Medical equipment which applies the video technology of Virtual Reality, etc.
korzoplugandplay · Fintechunchecked
betterment + mint all within a single app.
Rocket Doctorplugandplay · Healthcare and bioacquired
Rocket Doctor is a telemedicine marketplace that enables physicians to bring an advanced doctor's office directly to patients’ homes.

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 Horizontal AI assistants

$72M

awarded in the sector, tracked

opportunities open now in the sector

SBIR Fast-Track: Scalable Robotic Skill Learning via Generative Embodied Twins and Vision-Language-Action Models for Task-Oriented Autonomyawardhigh relevance
National Science Foundation · SBIR Fast-Track · $2M · posted 2026-09-09
Addressing Rising Mental Health Concerns in Youth: Development of an Digital Evidence-Based App for Providers and Teensawardhigh relevance
NIH / NIMH · SBIR phase I · $314K · posted 2026-09-01
SBIR Phase II: Ai-mediated Neurotechnology for Enhanced Access to Virtual and Physical Realityawardhigh relevance
National Science Foundation · SBIR Phase II · $1M · posted 2026-08-04
III: Building Reliable Retrieval-Augmented Reasoning Agents for Health Applicationsawardhigh relevance
National Science Foundation · Info Integration & Informatics · $706K · posted 2026-07-20
SBIR Phase II: Advancing MentorAI to improve academic and workforce outcomesawardhigh relevance
National Science Foundation · SBIR Phase II · $1M · posted 2026-07-16
SBIR Phase I: Creating Immersive Courseware for Calculus Pedagogyawardhigh relevance
National Science Foundation · SBIR Phase I · $304K · posted 2026-07-08
Developing a Learning Platform to Support Minimally Verbal Individuals with Autismawardhigh relevance
NIH / NIDCD · STTR phase I · $313K · posted 2026-07-01
The Development of a Personalized Feedback Program for Substance Use Disorder Riskawardhigh relevance
NIH / NIDA · STTR phase I · $305K · posted 2026-06-01

All public money by sector →

Market signal

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

Horizontal AI assistants · 33 → 78 → 85 → 58 → 41 new companies 2022 → 2026 · 89% aliveYC F26 live: 6 in this cluster, 5% of the batch (was 3% in S26)Since February, of 127 YC companies here: 4 acquired, 3 shut down, 31 rewrote their pitch

Horizontal AI assistants: 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: Enterprise
Business model: AI agent as a service
Path to 100x: Creates a new category
Market size: $1-10B market
Capital intensity: Capital-heavy (hardware, bio, infra)
Speed to revenue: R&D first, revenue after 3 years
Technical depth: Deep tech: ML, hardware, bio
Go-to-market: Self-serve
Moat: Distribution moat
Geography: Global from day one
Regulation: Heavily regulated
Vibe: Boring business