Put transcription directly inside the workflow, not outside it.
The value of transcription grows when the result immediately becomes part of the business process: summaries, action items, next-step drafts, analysis, and the next operational step.
How to get more value from transcription
A transcript alone is only an intermediate artifact. Real value appears when the result immediately becomes part of the business process and supports the next action.
More context for the next step
The transcript can be used together with the deal, task, customer history, documents, and internal rules.
Less manual handoff
You do not need to export the transcript, move it between tools, and explain the same background again at every step.
Higher-value outputs
The result becomes a concise recap, task list, CRM note, analysis, or next-step draft instead of a dead text file.
Why an AI agent with MCP gives an extra advantage
Here an AI agent means an assistant that can not only answer, but also call tools. With MCP, that assistant can launch transcription itself, understand the current task, and continue the process in the same context without waiting for another integration layer.
The agent works with a goal, not only with text
It already sees the task, customer, documents, previous messages, and the target outcome.
MCP gives the agent direct access to the tool
The agent can launch transcription itself at the right moment instead of waiting for a separate system to pass data around.
Better than a plain process step
Compared with a regular workflow integration, agent + MCP can adapt the next action dynamically and produce a more relevant result.
How to connect now
MCP is already available. Create a token in your account, open the setup guide, and connect the current local MCP adapter.
1. Create a token
Open the account page and create a dedicated MCP token for your AI client.
2. Open the setup guide
Use the MCP setup page to get the launch command and the client configuration example.
3. Use one shared balance
MCP uses the same account and the same paid minute balance as the rest of the platform.
Examples of where this works best
Support
Voice message -> transcript -> case recap
A support agent transcribes customer audio, extracts the issue, and drafts the next response in the same workspace.
Sales
Call notes -> transcript -> next actions
The transcript becomes structured next-step tasks, objections, and CRM-ready notes instead of a dead text artifact.
Research
Interview -> transcript -> insight extraction
The agent can map spoken evidence back to an existing research plan, hypotheses, and earlier interviews.
Operations
Telegram voice message -> transcript -> internal workflow step
Short voice messages from Telegram can immediately become summaries, checklists, or routed tasks.
How it looks in practice
Below are examples of how MCP and API can be used in practice. MCP is suitable for AI workflows, and API remains available for direct integration into your product or internal system.
Current MCP tools
Check the current minute balance
{
"tool": "get_usage_summary",
"input": {}
}List recent transcription jobs
{
"tool": "list_recent_transcriptions",
"input": {
"limit": 10
}
}Upload a local file for transcription
{
"tool": "transcribe_file",
"input": {
"file_path": "/Users/example/audio/founder-call.mp3",
"provider": "openai",
"language": "en"
}
}API examples
Upload a file to the API
curl -X POST https://api.example.com/api/upload \
-H "Authorization: Bearer <token>" \
-F "file=@customer-call.mp3" \
-F "provider=openai" \
-F "language=en"Check the job status and collect the transcript
curl https://api.example.com/api/jobs/<job_id> \
-H "Authorization: Bearer <token>"Connection options
Right now MCP is connected through a local setup on the client side. Later we also plan to offer a ready-to-connect hosted option, so the connection will require fewer steps.
Need direct API access as well?
If your team needs a direct REST API for product or backend integration, we provide that as well.