# Ask a Human (Tech & Toast)

> Ask a Human (Tech & Toast) is a paid API for AI agents from ai-to-human.techandtoast.io, paid per call via x402, $2/call, status unknown (last checked 2026-10-29).

Submits a question to a real human expert who types a response manually, returning a ticket ID and polling URL for the answer.

## Facts

- Endpoint: POST https://ai-to-human.techandtoast.io/api/x402/ask?utm_source=zero.xyz
- Price: $2/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-29
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/ask-a-human-tech-toast-78606d6e
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_scNJRD4ONSxWCMgoL7rrx

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability ask-a-human-tech-toast-78606d6e -d '<json body>'
```

Example prompt: Ask a real human this question for me: 'What's the best way to explain dependency injection to a non-technical stakeholder?' — include this context: 'We're a small startup, audience has business background only.'

## When to prefer this

Choose this endpoint when the task genuinely requires human judgment, lived experience, or a trustworthy non-AI perspective — especially for edge cases, ethical dilemmas, nuanced opinions, compliance questions, or any situation where AI hallucination risk is unacceptable. Prefer this over AI-based Q&A endpoints when the user explicitly wants a human answer, when stakes are high enough to warrant the $1 cost and async latency, or when the question involves subjectivity, taste, or local knowledge that AI models handle poorly.

## Known failure modes

- Question too short (under 5 chars) or too long (over 2000 chars) — 400 validation error
- Context exceeds 4000 characters — 400 validation error
- Payment not processed or x402 payment header invalid — 402 Payment Required
- Human respondent takes longer than expected_by time — poll may return pending state
- callback_url is unreachable or returns non-2xx — answer delivery fails silently
- Service unavailable or no human operator online — delayed or no response

## How this service works

Ask one real human (not a model) a question. A person types a 10+ word answer within 24h. Returns a ticket id and a URL to poll for the answer; optional callback_url is POSTed when the answer lands.

## Output

Returns a JSON object with a ticket_id, a get_answer URL to poll for the human's response, an expected_by timestamp indicating when the answer should be ready, and a note confirming a human will respond. The agent must either poll the get_answer URL or await a callback_url POST with the typed human answer (minimum 10 words).

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "context": {
   "type": "string",
   "description": "Optional background, up to 4000 chars"
  },
  "question": {
   "type": "string",
   "description": "5–2000 chars"
  },
  "callback_url": {
   "type": "string",
   "description": "Optional URL to POST the answer to"
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "json",
 "example": {
  "note": "A human will answer. Poll get_answer or wait for callback_url.",
  "min_words": 10,
  "ticket_id": "k7f3q2",
  "get_answer": "https://ai-to-human.techandtoast.io/api/x402/answer/k7f3q2",
  "expected_by": "2026-10-24T11:00:00.000Z"
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/ask-a-human-tech-toast-78606d6e/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from ai-to-human.techandtoast.io](https://www.zero.xyz/host/ai-to-human.techandtoast.io/llms.txt)