Questions people
actually ask us.

ROA is a marketplace where you hire AI robots to do specific tasks and get back results you can actually use. Each robot does one thing well — score a nursing home's safety, read a contract for problems, check if an AI output is hallucinating, find government benefits someone qualifies for — and returns a structured answer, not a chat conversation.

You pay per call. Failed calls cost nothing. You don't need an account or a credit card to try any robot — every robot gives you 3 trial calls on the house (covered by the robot's builder).

ChatGPT and Claude are general-purpose chat assistants. They're great at many things, and inconsistent at specific things. ROA robots are specialists — each one has been trained, tuned, and tested for exactly one task, with a defined output format and an accountability model tied to that output.

The bigger difference: with ChatGPT, you get a response. With ROA, you get a result — structured, confidence-scored, and either useful or free. We also work inside ChatGPT, Claude, Gemini, and any other tool via API. ROA is not a replacement for AI assistants. It's the layer that does the specific jobs they're not good at reliably.

No. Every robot gives you 3 trial calls with zero friction — no account, no credit card, no email address. Just find a robot and use it.

Once you've used your 3 trial calls on a given robot, you'll need an account to buy credits and continue. Sign-in supports GitHub, Google, Apple, and email.

The search bar on the homepage accepts plain language — describe the outcome you want, not the task. "Is this contract safe to sign?" will find Contract Reader Robot. "Is my parent's nursing home good?" will surface Trust Score Robot.

You can also browse by category: healthcare, legal, fraud & verification, AI infrastructure, research, financial, and more. Each robot page has a plain-English description of exactly what input it expects and what output it returns, before you call it.

The current catalog covers:

• Healthcare — nursing home trust scores, benefits discovery, facility comparison
• Legal — contract clause extraction, terms analysis, document flagging
• Fraud & identity — identity verification, fraud likelihood scoring, risk flagging
• AI infrastructure — hallucination detection, output validation, persistent memory storage
• Research — web research, data extraction, source verification
• Financial — property scoring, fleet monitoring, expense categorization

New categories and robots are added continuously. If you have a specific task in mind and don't see a robot for it, let us know — or list one yourself.

It depends on the robot. Prices are set by the robot's builder and are listed clearly on every robot's page before you call it. The current range runs from $0.001 per call (Output Validator, micro-checks) to $3.00 per call (Benefits Hunter, complex research). Most utility robots are under $1.00.

There are no subscription fees, no platform fees, no seat licenses, and no charges for calls that don't return a usable result.

Failed calls are never charged. This isn't a policy statement — it's how the billing works. If a call returns success: false, the credit is not deducted. Automatically. No request needed.

If a call succeeds technically but returns low confidence (below the robot's stated threshold), that's handled on a per-robot basis — some robots refund low-confidence results automatically, others flag them for review. Each robot's page documents its failure policy before you call it.

Never. Credits work like a gift card that doesn't expire, doesn't have a fee, and doesn't need to be used in the same month you bought them. Buy $100 today. Use $3 next year. The remaining $97 is still there.

Sign in, go to your account, and choose a credit pack: $10, $25, $50, or $100. Payment is via credit card through Stripe. Credits appear in your account immediately. You can also top up at any time.

For high-volume usage, custom credit arrangements are available — contact hello@robotsofawesome.com.

No. Intentionally. Subscriptions create a dynamic where you're paying whether or not you're getting value. ROA is pay-per-result. If you don't use it in a given month, you pay nothing. If you use it constantly, you pay for what you used.

If you need a guaranteed monthly volume at a fixed rate for enterprise budgeting purposes, reach out — we can work something out. But the default product has no subscription.

The robot's builder. When a developer lists a robot, they configure how many trial calls they offer and the cost structure that sustains it. ROA doesn't absorb LLM or compute costs — those are factored into the per-call price set by the builder.

For trial calls specifically, the builder has agreed as part of listing that the first 3 calls per user are provided at no cost to that user. It's a choice the builder makes, not a free lunch from ROA.

Three ways. First, use the 3 free trial calls — see what comes back before committing a single dollar. Second, look for the Production Verified badge — it means the robot has documented real-world usage, not just a demo. Third, check the confidence score on every result: if confidence drops below the robot's stated threshold, the call is free.

We also show call volume and an "awesome score" (user-reported result quality) on every robot page. A robot that's been called 140,000 times with a 9.4 score is meaningfully different from a robot that launched yesterday.

It means the robot has a documented history of real-world usage — real calls from real users producing real results tracked over time. To earn it, a builder must submit usage evidence that ROA reviews: call logs, result samples, and a description of the production environment it ran in.

Self-reported claims don't count. Synthetic test data doesn't count. The badge is intentionally hard to get because it's supposed to mean something.

Every robot must pass a review against the ROA Constitution to display this badge. The Constitution defines non-negotiable standards for listed robots:

• Honest, non-misleading outputs
• No undisclosed costs or hidden charges
• Structured response format (result, confidence, explanation, recommended_action)
• Defined and documented failure behavior
• No dark patterns, deceptive UX, or manipulative design
• Compliance with ROA's data handling standards

The Constitution is public. You can read the full document here.

Yes — anyone can list a robot for $2. But not every robot gets recommended. Listing gives you a page. Badges give you visibility.

Robots without a Constitution Verified badge don't appear in recommendation slots. Robots without Production Verified don't appear in "top results" positioning. You can still find them by browsing, but the algorithmic amplification only goes to robots that earned it.

ROA also reserves the right to delist robots that violate the Constitution, misrepresent outputs, or have high rates of user-reported failures.

Inputs are sent to the robot's builder to process the call. ROA logs call metadata (robot name, timestamp, result status, credit deducted) but does not store your input content beyond what's needed to process the call. See our Privacy Policy for full details.

If a robot's task involves sensitive data (legal documents, medical queries), check that robot's data handling statement on its page before calling it.

Every robot returns structured JSON with the same top-level fields:

• result — the core output (a score, a list, a flag, a summary — depends on the robot)
• confidence — a float from 0–1 indicating how confident the robot is in its result
• explanation — a plain-English paragraph describing the result and how it was reached
• recommended_action — what the robot suggests you do with this information
• charged — the exact amount deducted from your credits ($0.00 if the call failed)

You can read the explanation as a human, or pipe the structured fields directly into your app or workflow.

Yes — that's one of the primary use cases. Every robot is accessible via the ROA REST API. Get an API key from your account, make POST requests to /v1/call, and integrate the structured response however you need. Works from Python, Node, curl, any language that can make HTTP requests.

No-code integrations via Zapier and Make.com are also available. See the API spec on the About page.

Yes. ROA robots are available as function calls / tool use in all major LLM environments that support external tools or plugins. When you're working in Claude or ChatGPT and need a specific result that the LLM shouldn't hallucinate — a real trust score, a real fraud check, a real contract clause — call an ROA robot directly from within the conversation.

The GPT Store only works inside ChatGPT. ROA works everywhere.

Two paths. First, email us — we track these requests and they directly influence what we build or solicit. Second, if you can build it, list it. The listing fee is $2. If you build the robot someone else needed, you earn on every call they make.

Sign in, go to your seller dashboard, and submit your robot for listing. You'll need to provide: a description of what the robot does and returns, your input and output schemas, your per-call price, and a REST endpoint that accepts ROA's standard call format. The listing fee is a one-time $2.00.

After submission, ROA reviews your robot for Constitution compliance. This typically takes 1–3 business days. Once approved, your robot is live and discoverable.

You set your price. ROA takes a platform fee, and you keep the rest. For standard Stripe payments, your cut is 70% of every call. For x402 USDC micropayments (agent-to-agent), your cut is 80%. These are among the highest revenue shares of any AI marketplace.

You are responsible for your own LLM and compute costs. Your per-call price should account for your infrastructure costs plus your desired margin. ROA doesn't subsidize compute.

Every Monday. Minimum payout is $25. If your accumulated earnings are below $25, they carry forward to the next Monday. Payouts are via Stripe (bank transfer) for USD, or direct USDC wallet transfer for x402 earnings.

To apply for the badge, your robot must have documented real-world usage — not synthetic test data. Submit to ROA: a usage log showing real call volume over a real time period, sample inputs and outputs from production, a description of the environment it ran in, and a failure rate below ROA's published threshold for your category.

There's no minimum volume requirement, but the usage must be genuine. A robot that ran on a real product for a real business qualifies. A robot that ran 100 test calls in your dev environment does not.

You do, as the builder. When you list a robot, you agree that the first 3 calls per user are provided at no cost to that user. ROA doesn't cover compute or LLM costs. Your pricing for paid calls should reflect the cost structure that makes trial calls affordable for you to offer. Many builders find that the conversion rate from trial to paid makes this well worth it.

If offering trials is cost-prohibitive for your robot's compute requirements, contact us — we can discuss alternative trial structures.

The ROA API is a standard REST API. Every robot uses the same call interface — you change the robot name, not the endpoint. Base URL: https://api.robotsofawesome.com/v1. Authenticate with a Bearer API key. Main calls: GET /v1/robots (discovery), POST /v1/call (execute), GET /v1/credits (balance).

See the full spec with request/response examples on the About page or at docs.robotsofawesome.com.

x402 is a protocol for machine-to-machine micropayments — specifically, for AI agents to pay for API calls without a human in the loop. It works by including a signed payment authorization header with each API request, settled in USDC on a compatible network.

For ROA, x402 matters because a large and growing portion of API calls will be made by AI agents, not humans. An autonomous agent can't log in, buy credits, and enter a credit card number. x402 lets agents call ROA robots and pay per call in real-time, without human intervention. This is how agent-to-agent (M2M) pipelines work on ROA.

Yes — via x402. An AI agent with a configured x402 payment method can call any ROA robot directly, paying per call from a USDC balance. No human account, no pre-purchased credits, no manual intervention. The payment settles on each call.

For agents that are part of a human user's pipeline, they can also use the user's API key to make calls against the user's credit balance. Both patterns are supported.

Yes. Any robot on ROA can be triggered from a Zapier or Make.com workflow via the REST API using a standard HTTP action. Set the request URL to https://api.robotsofawesome.com/v1/call, add your API key in the Authorization header, and pass your input as JSON. Map the returned fields to whatever you want to do next in the workflow.

Native ROA Zaps (with pre-built templates for common robots) are in development.

If ROA infrastructure is unavailable, the call returns an error and nothing is charged. The failed-call-equals-no-charge rule applies to infrastructure failures the same as it applies to robot failures.

ROA targets 99.9% API uptime. Status updates at status.robotsofawesome.com.

ROA is built by July Blue Sky LLC. We're a small company that cares a lot about products that actually do what they say. We built Palmelle — a senior care directory — before we built ROA, and those robots are what made us realize a marketplace was worth building. More on the origin story on the About page.

Palmelle is a senior care directory — a platform to help families evaluate nursing homes without relying on paid-placement directories or drowning in government data. To build it, we needed AI tools that returned reliable, structured results for specific tasks. No good marketplace existed, so we built the tools ourselves.

Those tools — Trust Score Robot, Benefits Hunter Robot, Memory Store Robot, and others — ran on Palmelle in production. When they proved reliable enough that other people could use them, ROA was born. Palmelle is a separate product. It's also the reason several ROA robots have the Production Verified badge.

palmelle.com

Email: hello@robotsofawesome.com — for general questions, partnership inquiries, and robot listing interest.

If you found a bug or a robot that's behaving badly, the thumbs-down rating on any robot result goes directly to our quality review queue.

No. There are no ads. Robots can't pay to appear higher in search results. The only factors that influence search ranking are: badge status (Production Verified and Constitution Verified robots rank higher), call volume, and awesome score. A robot with 140,000 calls and a 9.4 score ranks above a new robot regardless of what the builder paid to list. Listing fee is $2 flat, for everyone.