New startup ideas · AI and software · Horizontal AI assistants
startup idea
Backline
Flat-rate AI back office for small businesses, running on our own GPUs.
Backline sells small businesses a set of working agents - phone answering, scheduling, invoicing, collections, vendor follow-up - for one flat monthly price with a self-serve signup that takes fifteen minutes.
- AI agent as a service
- Small business
- $100B+ market
- Collapses a cost 10x
- US first
4/5
venture judge
27
similar startups, last 2 years (119 all-time)
95%
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
$700 · 3 weeks · 30 prospects
For $700 in three weeks, prove Phoenix home-services owners will prepay a flat monthly fee for an AI agent that answers their after-5pm calls.
Riskiest assumption · Owners of small home-services shops who miss after-hours calls will prepay a flat monthly fee to an unknown vendor for an AI phone agent instead of staying with voicemail or a human answering service.
1Focus group: who and where
Owner-operator of a 2-10 person plumbing, HVAC, or electrical shop in metro Phoenix that runs calls to the owner's cell, misses inbound calls after 5pm and on weekends, and either pays part-time admin help or watches those jobs go to whoever answered.
where to find 30 · PHCC of Arizona chapter membership and meetings (association); the Plumbing Business Owners and HVAC Business Owners groups on Facebook (community); a Google Maps and Yelp list of Phoenix plumbing and HVAC companies, including their review sections mentioning missed calls (directory); Service World Expo in Las Vegas this October (event).
2Sell first, build later
An AI phone agent that answers every call after 5pm and on weekends, books the job into the shop's calendar, and texts the owner a summary, live for 10 founding Phoenix shops by October 1, 2026.
the ask · $299 per month flat, no per-minute fees; first month $199 for founding members, rate locked for 12 months.
a real yes · A real yes is a prepaid $199 first month through the payment link or checkout. Compliments, requests for a free trial, and call me when it is live are not.
3Small experiments
The first one attacks the riskiest assumption; each ends with a number that says whether to run the next.
1. Owner calls ending in prepay
$150 · 12 days
Book 20 calls with Phoenix shop owners recruited through PHCC of Arizona, the Facebook owner groups, and direct calls off the Google Maps list. Quantify missed after-hours calls per week and current admin spend, then make the founding offer live on the call: $199 first month prepaid, $299 per month locked for a year, agent live by October 1. The founder runs every call and sends the payment link before hanging up.
keep going if · 5 of 20 calls end with a prepaid $199 first month
2. After-hours mystery call audit
$50 · 3 days
Call 40 Phoenix plumbing and HVAC companies from the Google Maps list at 6:30pm on weekdays posing as a homeowner with a water heater leak. Record whether a human answers, a service takes a message, or it goes to voicemail. This doubles as the outreach target list.
keep going if · 24 of 40 calls reach voicemail or a message-only service
3. Priced landing page test
$500 · 10 days
Put up a one-page site with the founding offer and a checkout for the $199 prepaid first month, then spend $400 on Google Ads against Phoenix searches like answering service for plumbers and after hours HVAC calls. Track visits, checkout starts, and completed prepays.
keep going if · 3 prepays or 15 checkout starts from the first 200 visits
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.
$199
per prospect, refundable
how · First month prepaid at the founding price through the landing-page checkout or a payment link sent on the call; chosen because a home-services owner already prepays for software monthly and the first month is the natural unit, no contract to sign. set up: Stripe Payment Links ↗
what it reserves · One of 10 founding slots in Phoenix, the $299 per month rate locked for 12 months, and the October 1 go-live date.
refund · Refunded in full within 5 business days if the agent is not answering the shop's calls by October 1.
target · 8 prepays within 21 days, 5 from 20 owner calls and 3 from the landing page
Go: build it if
8 or more prepaid first months within 21 days across calls and the landing page, with the audit confirming at least 24 of 40 shops miss after-hours calls.
Kill: stop if
Fewer than 3 prepays after 20 owner conversations and 200 landing visits, or fewer than 12 of 40 audited shops actually missing after-hours calls.
5 Scripts to run itoutreach message, landing copy, deposit terms · click to open
outreach message
You run a plumbing or HVAC shop, and every after-5pm call that hits voicemail is a job booked with whoever picked up. I am building Backline, an AI phone agent that answers every after-hours call, books the job on your calendar, and texts you the details. Flat $299 a month, no per-minute fees, and I am taking 10 founding Phoenix shops live by October 1. Do you have 20 minutes this week to see if your call volume fits?
landing page
Your phones answered after 5pm, every job booked $299 a month flat; founding shops pay $199 for the first month, live by October 1 Prepay your first month to claim 1 of 10 founding slots
deposit terms
Prepay $199 for your first month to reserve one of 10 founding slots in Phoenix, with your rate locked at $299 a month for the next 12 months. If your agent is not answering calls by October 1, we refund the $199 in full within 5 business days. No contract, cancel any time.
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.
40
Idea Score, 0-100 · raw 24.7 x 1.61
Crowded
competition: more crowded than 87% of ideas · headwind x0.56
+3.0
government priorities, secondary (20 matching grants)
Trend
11
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)20
- Rounds announced 2025+ in the sector0
- Sector direction (live batch)0
- 2026 trend analyst25
Demand
68
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)100
- Operator judge: real pain75
100x potential
73
Can it return a fund? The venture judge (double weight), market-size and moat axes, neighbours still alive, the technologist judge.
- Venture judge75
- Market size axis100
- Moat axis100
- Neighbours still alive41
- Technologist judge50
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 flat-rate, back, office, gpus, sells, set, working, phone.
The idea in full
- What
- Backline sells small businesses a set of working agents - phone answering, scheduling, invoicing, collections, vendor follow-up - for one flat monthly price with a self-serve signup that takes fifteen minutes. It hits that price by owning its own GPU clusters and serving tuned open-weight models, putting per-task cost roughly 10x below competitors who resell API tokens. Every business on Backline exposes its agent to every other: a plumber's agent books the bookkeeper's agent, requests quotes, and settles invoices without a human on either side.
- Why now
- 91 of the 329 tracked companies in this cluster launched in the last two years, but almost all resell frontier-model tokens at API prices; Typework (Antler Singapore 2026) is already pitching agents that find deals by talking to each other, validating agent-to-agent commerce as the next layer.
- Wedge: first customer and entry point
- Home-services businesses in one US metro that miss inbound calls after 5pm; a 3-person team ships the phone-and-scheduling agent first and lets the invoicing and agent-to-agent features pull in the rest.
- Path to 100x
- US small-business back-office labor is a $100B+ annual spend, and a flat-fee product that does the work for a tenth of a part-time hire's cost can convert labor budget into software revenue at scale. The agent-to-agent transaction graph is the winner-takes-most layer: once enough local businesses transact through Backline, joining it becomes the default way to trade with them, and the network compounds.
- Ceiling
- If the inter-business network never forms, Backline is a well-priced but substitutable admin tool competing on cost alone, and gets capped near a few hundred million in revenue by churn and price pressure.
- Closest real companies, as the generator saw them
- Typework builds bespoke agent teams sold as a service; Cohesive AI and Tasklet target knowledge workers with per-seat tools; none own inference hardware or connect customers' agents to each other.
- Main risk
- Frontier API prices fall faster than owned-hardware economics improve, erasing the 10x cost gap before the agent-to-agent network reaches critical mass.
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
4/5
$100B+ SMB back-office labor with a flat fee and an agent-to-agent graph; owned GPUs are the bet, not a services drag.
Bootstrapper
3/5
Home-services businesses already pay for missed calls and revenue starts within a year, but owning GPU clusters blows the capital budget.
Operator
4/5
Home-services owners lose booked jobs on missed after-5pm calls and already pay part-time admin help; flat fee, fifteen-minute signup.
Technologist
3/5
Owning GPUs and tuning open-weight models is real work, yet the 10x rests on frontier token prices staying high for the plumber-agent tier.
Risk
3/5
Owned GPU clusters remove API-platform dependence, but the entire 10x pitch dies if frontier token prices fall faster than the hardware amortizes.
trends
2/5
Flat-fee back office on owned GPUs cites only one Antler company; nothing here depends on a dated 2025-2026 shift from the brief.
Similar startups in the directory
Companies whose pitch matches most of the idea's terms (flat-rate, back, office, gpus, sells, set, working, phone): 119 all-time, 27 from the last two years. Same matching as Idea Check.
All-In-One AI Growth Partner. Built for Trades.
The AI sales agent for CPG brands
Our SaaS is the only solution that enables banks and fintech to focus on scale while automating the burdensome and manual back-office tasks related to Payment/Card Operations and Compliance.
AI to engage employees, predict attrition and automate HR/IT support
Modern, on-demand lawn care service. Get a quote in two minutes.
Build everyday back-office agents on one live model of your books
Context layer for enterprise AI surveying employees with voice agents
We run benefits, payroll & compliance for startups, for free
AI-native OS for insurance brokers
AI for Construction's Back Office
AI For The Skilled Workforce
Healthcare AI
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.
8
grants and programs matching the idea
20
startup-relevant grants in Horizontal AI assistants
$72M
awarded in the sector, tracked
1
opportunities open now in the sector
- IRB-Assist: AI-powered decision support platform that enhances the speed, consistency, and accuracy of Institutional Review Board reviewsawardhigh relevance
NIH / NCATS · SBIR phase I · $314K · posted 2026-09-01
NIH / NHLBI · SBIR phase I · $406K · posted 2026-09-01
- Collaborative Research: CPS: Medium: Designing a Closed-loop Unvoiced Interaction Paradigmawardhigh relevance
National Science Foundation · CPS-Cyber-Physical Systems · $497K · posted 2026-08-07
- Reducing Hypertension Among Men in Manual and Trade Labor: A Mobile Stress Management and Medication Adherence Interventionawardhigh relevance
NIH / NHLBI · SBIR phase I · $305K · posted 2026-08-01
- Empowering Direct Care Workers with AI-Integrated Wearable Technology to Prevent Back Injuriesawardhigh relevance
NIH / NINR · STTR phase I · $307K · posted 2025-09-09
- Collaborative Research: CPS-CIR: Community-maintained Resilient AI Infrastructure for Local Food Security in Hawaiiawardmedium relevance
National Science Foundation · CPS-Cyber-Physical Systems · $642K · posted 2026-08-21
- Collaborative Research: CPS-CIR: Community-maintained Resilient AI Infrastructure for Local Food Security in Hawaiiawardmedium relevance
National Science Foundation · CPS-Cyber-Physical Systems · $358K · posted 2026-08-21
- Collaborative Research: Fast and sound coalescent-based inference of semi-directed phylogenetic networks from genome-scale dataawardmedium relevance
National Science Foundation · MATHEMATICAL BIOLOGY · $126K · posted 2026-08-20
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
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
- Small business
- Business model
- AI agent as a service
- Path to 100x
- Collapses a cost 10x
- Market size
- $100B+ market
- Capital intensity
- Capital-heavy (hardware, bio, infra)
- Speed to revenue
- Revenue within a year
- Technical depth
- Real engineering
- Go-to-market
- Self-serve
- Moat
- Network effects
- Geography
- US first
- Regulation
- Unregulated
- Vibe
- Hot space
Listed under
An idea sits in its own sector and in any sector its text clearly touches.
More ideas like this
AI and software · Horizontal AI assistants
Subroclaim
Agents that negotiate insurer-to-insurer claim recoveries on both sides of the file.
Subroclaim runs subrogation and inter-carrier recovery as an agent service for US property and casualty insurers: the agent reads the loss file, builds the demand with liability evidence, and then negotiates directly with the other carrier's Subroclaim agent, escalating to a human adjuster only on disputed liability or amounts above a set threshold.
AI and software · Horizontal AI assistants
Guildhall
The exchange where enterprises hire proven AI agents like contractors.
Guildhall is a self-serve marketplace where enterprises browse, trial, and deploy third-party AI agents - sales research agents, ops agents, finance-close agents - each listed with standardized, independently run performance telemetry instead of vendor claims.
AI and software · Horizontal AI assistants
Piecework
A marketplace where enterprises hire third-party AI agents and pay per verified task.
Independent builders list agents with a declared task type, a price per completed unit and a verification rule; enterprise teams connect their systems through no-code connectors and buy agent labour without a procurement cycle.
AI and software · Horizontal AI assistants
Procura
API that gives any AI agent legally binding authority to act for a small business.
Procura is the mandate layer agent builders call before their agent signs, pays or files anything: the small business owner verifies identity once, grants a scoped, revocable power of attorney, and Procura issues a cryptographic credential that binds each agent action to that mandate with a qualified electronic seal and an immutable audit trail.
AI and software · Horizontal AI assistants
Registrum
Federated agent network for public agencies: every resolved case trains every jurisdiction.
Registrum runs caseworker agents for public agencies (permit review, benefits eligibility triage, procurement document checks) inside a sovereign compute enclave the agency controls.
AI and software · Horizontal AI assistants
Bursar
A personal agent with a license to move your money.
Bursar is a consumer financial agent that does not just read your accounts - it pays bills, disputes charges, cancels subscriptions, negotiates rates, and switches providers, because the company holds money transmitter licenses and runs its own payment rails.
Fictional company written 2026-08-26 from MarkosWeb data; the companies, grants and numbers around it are real and tracked. Treat the idea as a research prompt, not a plan.