One API key gives your agents three production tools. Results come back with their sources and a record of what changed, so your code can check them before it acts.
# normalize a CRM export curl https://patchwell.io/v1/clean_data \ -H "Authorization: Bearer $PATCHWELL_KEY" \ -d '{"url": "https://example.com/customers.csv"}'
{
"stats": {
"input_rows": 48210, "clean_records": 41877,
"duplicates_merged": 5112, "values_changed": 63094
},
"changes": [{ "row": 14, "column": "Signup Date",
"before": "3/4/2019", "after": "2019-03-04" }, ...],
"recipe": { /* replay it on next month's file */ }
}curl https://patchwell.io/v1/research \ -H "Authorization: Bearer $PATCHWELL_KEY" \ -d '{"ask": "When did JWST launch and what are its instruments?"}'
{
"answer": "The James Webb Space Telescope launched on
December 25, 2021 [science.nasa.gov/...]. Its main
instruments are NIRCam, NIRSpec, NIRISS and MIRI
[science.nasa.gov/...].",
"gaps": [], "confidence": "medium",
"billing": { "price": 0.15 }
}curl https://patchwell.io/v1/transcribe \ -H "Authorization: Bearer $PATCHWELL_KEY" \ -d '{"url": "https://example.com/interview.mp4"}'
{
"language": "en", "duration_s": 2626.2,
"words": [{ "w": "Behnken.", "s": 2.31, "e": 2.88, "p": 0.97 }, ...],
"srt": "1\n00:00:00,000 --> 00:00:02,880\n...",
"flags": [], // anything the guards removed, with times
"billing": { "price": 0.35 }
}Each tool runs on GPUs that start when a request arrives and stop when it finishes. You're billed for completed work.
Speech to text with timing for each word, from any audio or video link.
Messy spreadsheets and exports returned as consistent, typed records.
Answers to questions about markets, products, companies and topics, from the live web.
Responses include the evidence behind the result, in fields your agent can read.
Data cleaning returns each edit as row, column, before, after and rule, plus a row count that reconciles input to output.
Research answers attach a source URL to each statement. Points that couldn't be confirmed are returned in gaps.
A declined request returns a category such as contact_details, so your code can tell a policy refusal from an error.
Each response includes a billing object. Long jobs hold their maximum price and release the difference on completion.
Choose an amount and pay through Stripe. Your API key is displayed once after payment.
Call the REST endpoints with JSON, or add the MCP server URL to any MCP-compatible client.
Requests that run past 100 seconds return a job_id. Poll it until status is done.
// your MCP client config { "mcpServers": { "patchwell": { "type": "http", "url": "https://patchwell.io/mcp/", "headers": { "Authorization": "Bearer YOUR_KEY" } } } } // tools: transcribe, clean_data, research, get_job_result
import httpx, time API = "https://patchwell.io" H = {"Authorization": "Bearer YOUR_KEY"} r = httpx.post(f"{API}/v1/transcribe", headers=H, timeout=150, json={"url": "https://example.com/talk.mp4"}).json() while r.get("status") == "running": # long files become a job time.sleep(20) r = httpx.get(f"{API}/v1/jobs/{r['job_id']}", headers=H).json() print(r["srt"][:200], r["billing"])
curl https://patchwell.io/v1/research \ -H "Authorization: Bearer YOUR_KEY" \ -H "Content-Type: application/json" \ -d '{"ask": "Which open-source licenses allow commercial use?"}' # check your balance curl https://patchwell.io/v1/credits/balance -H "Authorization: Bearer YOUR_KEY"
Prepaid credit, deducted per call. No subscription, no minimum, and credit doesn't expire.
| Tool | Price | Charged when | Free |
|---|---|---|---|
| Transcribe | $0.008 / audio minute | The transcript is delivered | Failed downloads, unreadable files |
| Clean data | $2.00 / 1,000 rows (min $0.02) | The cleaned file is delivered | Files we refuse or can't read |
| Research | $0.15 / answered question | A sourced answer is delivered | Refusals and empty answers |
The research tool doesn't answer questions about specific people: their location, contact details, relationships or records. These requests are declined with a category and aren't billed. Keys that repeatedly send them are paused.
No. Add any amount of credit from $10 to $500, and each call is deducted from your balance. Credit doesn't expire.
No. Declined requests, research questions without a sourced answer, and files that can't be read aren't billed. Each response shows its cost in the billing field.
Unused credit is refundable within 30 days of purchase. Email us from the address you paid with.
No. Files are processed in a temporary container that is discarded afterwards. Results of long jobs are held briefly so you can retrieve them. We keep billing records (tool, size, price and time).
Any client that supports remote MCP servers over HTTP. Add the server URL and your API key as a bearer token.
Data cleaning completes in seconds. Research typically takes 30 to 90 seconds. Transcription runs many times faster than the audio's length. Requests longer than 100 seconds return a job ID.
We don't want the service used to profile or locate individuals, so those questions are declined. Questions about companies, products, markets and public topics are supported.