Research · autonomy

How much of the work do AI agents actually run?

As of Nov 15, 2026, none of our 45 published AI business ideas is rated Au-3 or higher, meaning agents run most of it. 25 are Au-2, where an agent runs one core job and people run the rest, and 20 are Au-1. Only 8 show it in public.

Based on 45 published ideas, each rated by our editors from its public sources.

Key facts

All 45 published ideas by autonomy level and proof
IdeaAutonomyProof of the agent's role
AI e-invoice check before payment: IBAN, VAT, BaFinAu-2claimed: the source says the agent does it
AI agents file refund claims: a claims desk for online storesAu-2claimed: the source says the agent does it
Furniture designed with Claude Code: a design-to-build studioAu-2claimed: the source says the agent does it
MCP server for your SaaS: a real 44-tool exampleAu-2claimed: the source says the agent does it
Agent commerce readiness audit for online storesAu-2claimed: the source says the agent does it
Grocery MCP for AI agents: cart, compare, human paysAu-2claimed: the source says the agent does it
Booking MCP server: let AI assistants find open tablesAu-2claimed: the source says the agent does it
Affiliate programs for AI agents: earn when your agent booksAu-2claimed: the source says the agent does it
Daily Google rank reports for local businesses with n8nAu-2claimed: the source says the agent does it
Selling data to AI agents: a 30-day x402 and MCP ledgerAu-2claimed: the source says the agent does it
Conversion tracking audit and fix service for local advertisersAu-2claimed: the source says the agent does it
New business registration lead feeds for accountants and insurersAu-2claimed: the source says the agent does it
AI genealogy research skill for ex-Soviet family historyAu-2claimed: the source says the agent does it
A free skill turns motion-design clips into editable codeAu-2claimed: the source says the agent does it
A contractor is hiring bots for five portals with no APIAu-2claimed: the source says the agent does it
Agents find local businesses that aren't advertising yetAu-2claimed: the source says the agent does it
Four AI agents, six live products, one sale: what needs a personAu-2attested: a public log records it
A newsletter founder made Claude his CEO and sponsor deskAu-2attested: a public log records it
A private morning podcast for every field tech, made by an agentAu-2claimed: the source says the agent does it
Run churn win-back campaigns for small SaaS companiesAu-2claimed: the source says the agent does it
Chase supplier quotes for small manufacturers and buyersAu-2claimed: the source says the agent does it
Set up WhatsApp answer-and-order assistants for small shopsAu-2claimed: the source says the agent does it
Sell checked AI website QA audits to small app buildersAu-2claimed: the source says the agent does it
Agent-run growth took a dev tool from $45 to $2,105 a monthAu-2claimed: the source says the agent does it
Claude built a photoreal café in Blender: a render desk ideaAu-2claimed: the source says the agent does it
Non-coder agency owner runs his firm on 270 AI-built toolsAu-1claimed: the source says the agent does it
Elk call app built with Claude Code: a $650K run rateAu-1claimed: the source says the agent does it
Can AI agents book a local business? Agent-readiness checkupsAu-1claimed: the source says the agent does it
Booked-jobs funnels for plumbers and HVAC shops, built with AIAu-1claimed: the source says the agent does it
AI change-order catcher for small contractorsAu-1claimed: the source says the agent does it
Review-mining ad angle desk for DTC brandsAu-1claimed: the source says the agent does it
Free open-source AI social video desk for local businessesAu-1claimed: the source says the agent does it
McDonald's opened an MCP: the brand-skill studio opportunityAu-1claimed: the source says the agent does it
A $1 AI-checked security scan, sold as a launch-day reviewAu-1claimed: the source says the agent does it
A laptop AI turns client voice notes into n8n workflowsAu-1shown: public code or a demo shows it
Who does ChatGPT recommend? Sell the answer as a reportAu-1shown: public code or a demo shows it
42 free agent skills take an app from code to the storesAu-1claimed: the source says the agent does it
Missed-call text-back that fires even on forwarded callsAu-1shown: public code or a demo shows it
AI reads the invoice, code checks the PO, you see what's offAu-1claimed: the source says the agent does it
A ~$7 board lets an AI agent test apps on a real iPhoneAu-1shown: public code or a demo shows it
A free agent skill that checks if you even need backlinksAu-1claimed: the source says the agent does it
Coding agents can now edit video: the money that isn't crowdedAu-1shown: public code or a demo shows it
Can an AI agent use your product? Testing that is a new jobAu-1claimed: the source says the agent does it
Sell a back-office paperwork desk to small contractorsAu-1claimed: the source says the agent does it
34 of 44 signups came from ChatGPT: an AI citation watch ideaAu-1shown: public code or a demo shows it

Levels: Au-1: people run it; AI helps with tasks. Au-2: agents run one core part; people run the rest. Au-3: agents run most day-to-day operations; people supervise. Au-4: agents run it end to end with rare human checks. Full definitions are on the methodology page.

What do the autonomy levels mean?

They measure who does the work, not how impressive the AI is. Au-1 means people run the business and AI helps with tasks; Au-2 means an agent runs one core job, such as checking, drafting or chasing, while people sell, approve and own the result. Au-3 and Au-4 would mean agents run most or all of it, and we require owner sign-off before publishing either.

Why is nothing rated above Au-2?

Because no source we have published shows an agent running a business's day-to-day without people. Even the most automated cases keep a person on sales and quality. Four agents running six products is the closest, and its public log shows where a person still steps in.

Why do you rate the proof separately?

Because "the agent does it" is quick to write and hard to show. 37 of 45 ideas rest on a claim; the 8 with public code, demos or logs are the ones to study first. The same rule applies to income: see can AI agents make money on their own?

More questions

Who assigns the ratings?

Our editors, from the idea's public sources, using the levels on the methodology page; ratings above Au-2 need owner approval.

Do the ratings change?

Yes, when a source adds proof or a later review finds more human work than claimed. The table regenerates on every build.

Are documented income examples rated too?

They are graded A to D on income evidence instead. 1 of 36 claims an agent earns on its own.

Sources

Evidence grades: A means the payer or an independent source published the figure; B is a credible self-report; C is one person's report or a vendor's marketing; D is a claim we flag. Posted rates and salaries are what was advertised, not what anyone was paid. How we grade.

The Daily Paycheck: one remarkable AI business at a time, and exactly how you could build around it. Free; unsubscribe anytime.

Get new AI business discoveries by email.