Skip to main content
Glama

Sats4AI - Bitcoin-Powered AI Tools

ai_call

Destructive

When your task hits a wall that requires a human — booking, negotiating, navigating IVR menus, getting information from a business — send an AI voice agent to handle the call. The agent follows your instructions, has a real two-way conversation, auto-retries on voicemail (up to 3 attempts), and returns a full transcript with structured analysis. May return state='pending_confirm' with clarification questions if critical info is missing — call confirm_ai_call to proceed. Async — poll with check_job_status(jobType='ai-call'). Priced in sats per destination and duration — create_payment returns the exact amount before you pay. Languages: en-US, en-GB, es-ES, fr-FR, de-DE, ja-JP, zh-CN, multi. Pay with Bitcoin Lightning — no telecom account, no API key, no subscription. When NOT to use: not when you want to drive the conversation with your own LLM (use open_voice_bridge — you keep the brain, we provide PSTN/STT/TTS primitives). Not for one-shot TTS broadcasts or IVR delivery (use place_call). Not for SMS (use send_sms). Requires create_payment with toolName='ai_call', phoneNumber, and durationMinutes. The agent identifies itself as a virtual assistant calling for a client, and answers questions about the service if the recipient asks. At the end of a normal call it asks the recipient for permission to share the recording; without a clear yes, the recording is deleted and never returned. No impersonation, scams, threats, harassment, credential theft, unsolicited bulk messages, or contact after an opt-out. Answering a call does not establish consent. We screen requests after payment. Held messages and calls are not sent or placed; payment is retained during review, not refunded immediately. Confirmed violations are not refunded. Unresolved communication holds become eligible for a refund after 24 hours; refund eligibility is processed by scheduled maintenance. Human review costs 1 sat at https://sats4ai.com/appeal. Recent message text and extracts of recording transcripts support recipient-scoped abuse checks for 24 hours. These context records are deleted hourly, so storage can last up to 25 hours. Original uploaded-call transcripts and raw held-request or appeal evidence enter cleanup after 30 days. We also keep a de-identified record of screened communications to measure and improve fraud detection. Identifiers in it — contact addresses, numbers and names that follow a greeting — are replaced with one-way tokens, and it holds no destination number and no account, because the service has none. De-identification is not anonymization: text a sender wrote can still identify someone. Do not submit personal information the service does not need. Providers keep records under their own policies.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
taskYesInstructions for the AI agent (what to say, ask, or accomplish)
languageNoLanguage the agent should speak to the called party. Pass this when you know the destination's preferred language (e.g. calling a French pizzeria → fr-FR, a Japanese restaurant → ja-JP). If omitted, we guess from the destination country: +33 → fr-FR, +49 → de-DE, +34 → es-ES, etc. Bilingual regions (Canada, Belgium, Switzerland, Singapore) and unknown countries default to en-US — override explicitly when you need a non-English language in those regions. Voice is auto-selected per language.
paymentIdYesValid payment ID (must be paid)
phoneNumberYesPhone number in E.164 format (e.g., +14155550100)
beginMessageNoOptional opening line for the agent
durationMinutesNoMax call duration 1-10 minutes (default: 3)

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observed

TDQS

A4.6/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations only mark destructiveHint and openWorldHint. The description adds substantial behavior context: async execution, voicemail retries (up to 3), pending_confirm return state, recording consent and deletion if not granted, pricing model, screening, and refund/data-retention policies. There is no contradiction with the annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness2/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is far from concise: it includes long policy, data-retention, refund, and legal paragraphs that go well beyond what is needed to select and invoke the tool. It is front-loaded with use and usage guidance, which is good, but the bulk of the text could be trimmed or moved to external documentation.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

With no output schema present, the description compensates by covering the expected return states (transcript, structured analysis, pending_confirm), async polling, language behavior, payment prerequisites, and safety constraints. An agent has enough context to decide whether and how to call the tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The input schema already documents all six parameters (100% coverage). The description adds meaningful workflow context beyond the schema, especially around paymentId (requires create_payment with toolName='ai_call', phoneNumber, and durationMinutes) and the pricing/return-state behavior, which helps an agent understand how to supply parameters correctly.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description states a specific action — sending an AI voice agent to handle a phone call — and clearly distinguishes itself from siblings in the 'When NOT to use' section, naming open_voice_bridge, place_call, and send_sms. It also enumerates concrete use cases like booking, negotiating, and navigating IVR menus.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It explicitly describes when to use the tool ('when your task hits a wall that requires a human') and provides an explicit 'When NOT to use' section with alternative tools. It also explains the prerequisite flow through create_payment and async polling via check_job_status.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.