Idea Lab
Pick a vertical. The radar reads everything it tracks there - momentum by year, crowded subniches, dead companies, fresh rounds, grant money, open YC RFS themes - and proposes concrete startups someone should build. The companies below are fictional; the data behind them is not.
Fintech
407
companies, last 2 years
7519
tracked all-time
83%
of checked sites alive
$1470M
across 7 recent rounds
Most crowded right now: prediction markets (4) · data centers (5)
concept 1
Standline
Read-only downtime attribution for factory lines, installed in a day on the plant's own cloud.
A hosted service that ingests existing MES, historian and PLC-tag exports and answers one question per shift: which asset, changeover or material shortage cost the line the most minutes, and what it cost in money. Buyer is the plant IT manager or operations director at a mid-size discrete manufacturer who already has data but no analyst to read it. Sold as a fixed monthly fee per line, no new hardware, no write access to control systems.
- why now
- The robotics/physical cluster added 143 companies in two years and 78 in 2026 alone, with YC S26 funding Dawn Industries (diagnosis and fix for automation cells), Neuron Industries (industrial controllers for AI) and Control Seat (control and monitoring of industrial operations) — investor attention is on the plant floor, while the crowded phrase is 'data centers' (5 companies), not shopfloor analytics. The NIST Manufacturing Extension Partnership Center State Competition (closes 2026-08-21) and NSF 'Manufacturing Systems Integration' program show public money pushing mid-size plants toward exactly this integration work.
- wedge
- One read-only connector for the single most common historian/MES export format he already knows, plus a weekly PDF 'lost minutes and lost euros' report for one line at one friendly plant, priced at a few hundred a month. No dashboards, no real-time, no write path.
- closest real companies
- Dawn Industries and Control Seat from the brief are the nearest: both aim at automated diagnosis and control of automation cells. This differs by refusing control-plane access and selling a shift-level cost attribution report to IT/finance rather than an automation product to engineering, which is why it can be installed without a plant trial.
- main risk
- Sales cycles at mid-size manufacturers run longer than the 6-month runway and nobody on the team can shorten them.
concept 2
Freightdelta
We audit freight invoices against contracts and take a cut of what we recover.
A service that pulls a shipper's carrier invoices, rate sheets and shipment records, detects overcharges — wrong accessorials, duplicate billing, missed dimensional weight rules, service failures owed as credits — and files the claims. Buyer is the logistics or finance manager at a mid-market shipper spending $2M-$50M a year on freight. Priced purely as a percentage of recovered money, so no budget approval is needed.
- why now
- The commerce cluster shows freight software being funded now — Ekho Labs (decisions for disrupted freight), Haladir (operational superintelligence for global logistics), Tarifflo (supply chain compliance and cost savings), chrt — while ClearJet raised $25M in August 2026 for cargo logistics. The graveyard shows the opposite shape failing: KokkanLogis's freight brokerage platform went inactive in 2024, so the money is in software and recovery layers on top of existing carriers, not in re-intermediating them.
- wedge
- One shipper, one carrier, one 90-day back-audit done semi-manually with scripts, invoiced as a share of the recovery. Recovery money arrives in weeks, not after a procurement cycle, which fits the 6-month runway.
- closest real companies
- Tarifflo (supply chain compliance and cost savings) is the closest tracked company; it sells compliance software up front, while this sells recovered cash with zero risk to the buyer. Haladir and Ekho Labs target operations decisions, not billing accuracy.
- main risk
- Recovery rates on a given shipper may be too thin to fund the work once the easy first-audit backlog is cleared.
concept 3
Tallyrail
Closes the books for multi-channel merchants by reconciling every payout to every order.
Software that ingests payouts and fee statements from payment processors and marketplaces, matches them to orders and refunds, and produces a reconciled ledger plus a journal export the merchant's accountant can post. Buyer is the finance lead or outsourced accountant of a $5M-$100M online merchant that sells across several channels and closes the month by hand in spreadsheets. It never touches or holds money, so no licence is required.
- why now
- Fintech remains the deepest cluster in the brief with 118 new companies in two years, and YC S26 funded Luca IQ, an API-first tax engine for CPAs — accountant-facing money plumbing is live. The graveyard argues for the specific shape: Integration Labs' unified API for business financial data went inactive in 2022 and Ruby Card's e-commerce neobank died in 2023, so sell the finished close, not an API, and stay out of regulated banking.
- wedge
- Support exactly one processor plus one marketplace and one accounting export, run one merchant's month-end close as a paid pilot, and price per month per channel. Manual fallback allowed on any unmatched line.
- closest real companies
- Luca IQ (API-first tax engine for CPAs) is adjacent but sells tax calculation to CPAs; ZeroSettle handles in-app billing, not multi-channel settlement reconciliation. Integration Labs, in the graveyard, shows the pure-API version does not survive.
- main risk
- Merchants tolerate their spreadsheet close, so the pain may not clear the willingness-to-pay bar without an accounting-firm champion.
concept 4
Shelfmap
Makes mid-market catalogs readable and buyable by AI shopping agents, and proves it monthly.
A hosted layer that turns a merchant's existing product feed into structured, agent-consumable endpoints (clean attributes, stock, price, policies) and tracks whether AI assistants actually retrieve and recommend those products. Buyer is the head of e-commerce or growth at a mid-market brand watching organic search traffic erode. Sold self-serve plus a monthly visibility report.
- why now
- The radar's own measurement is the wedge: average AI-visibility across tracked commerce companies is 45/100 and 44-46/100 across clusters, meaning most catalogs are invisible to model-mediated discovery. Funding is moving to agentic buying rails — Agentcard (debit cards for AI agents) and Uno Wallet in fintech, Wildcard (AEO/GEO for e-commerce and retail) in commerce — while the commerce cluster added only 13 companies in 2026, so the tooling side is thin.
- wedge
- Free scan: point at a merchant's domain, return a one-page report on how agents see their top 50 products, then charge monthly to host the corrected feed. Self-serve conversion covers the missing sales function.
- closest real companies
- Wildcard (AEO/GEO for e-commerce and retail) is the direct tracked competitor; this differs by shipping the machine-readable commerce feed itself and by charging for hosted infrastructure rather than optimization advice. Quissly (e-commerce search) sits on the shopper side, not the agent side.
- main risk
- Shopify, Google or the model vendors standardize agent feeds for free and the layer becomes worthless overnight.
concept 5
Kernwerk
Fixed-price integration pods that connect legacy fintech and plant systems to modern cloud.
A small productized delivery outfit that takes one narrowly scoped integration — core banking or ERP to a cloud service, plant system to a data platform, marketplace to an accounting system — and delivers it at a fixed price in four to six weeks, with the connectors accumulating into a reusable library. Buyer is a CTO or IT director who has budget but cannot hire. Cash-first, product-later.
- why now
- The brief shows steady demand pressure and a survivor bias toward integration layers: 85% of tracked fintech sites are still alive and 83% in commerce, while API-only plays like Integration Labs (2022) and Pier's credit compliance layer (2023) sit in the graveyard — buyers now pay for delivered outcomes rather than a platform promise. Cross River Bank's $50M in April 2026 and 9fin's $170M Series C in March 2026 signal continued spend inside the financial plumbing that needs this work.
- wedge
- Sell one fixed-price integration to one warm contact within the next 60 days, invoiced 50% up front, and keep every connector under a reusable internal framework so the third engagement takes half the effort of the first.
- closest real companies
- None tracked in the brief — the closest funded shapes are Torus (Legora for physical engineering firms) and Reframe, both product companies rather than delivery firms, so this competes with ordinary consultancies rather than tracked startups.
- main risk
- It stays a consultancy: revenue tracks Alex's attention, no asset compounds, and the pulled engineers go back to their day jobs.
Fictional concepts generated 2026-08-18 by claude-opus-5 from MarkosWeb data (accelerator portfolios, funding news, US federal grants, YC RFS). Numbers reflect tracked sources only; treat every concept as a research prompt, not a plan.