Bland AI MCP. Manage every step of your conversational phone workflow.
Gemini Works with every AI agent you already use
…and any MCP-compatible client
Just plug in your AI agents and start using Vinkius.
Bland AI. Automate phone calls, analyze transcripts, and manage inbound numbers directly from your agent. Send conversational agents to specific numbers or run bulk campaigns.
Retrieve full call logs, pull raw MP3/WAV recordings, and analyze call transcripts to extract goal completion status instantly.
What your AI agents can do
Analyze call
Queries an active recording to determine if a specific goal was completed during the conversation.
Create web call
Sets up a decoupled, internet-based WebRTC signaling socket for browser audio streaming.
End call
Forces an immediate disconnect on a live AI phone call.
Dispatches conversational AI agents to single or multiple phone numbers, or manages the status of large batches of calls.
Interrogates call recordings to extract specific goal completion statuses from the raw transcript data.
Lists purchased PSTN numbers and spawns WebRTC sockets for browser-based audio handling.
Pulls full historical logs of AI calls, detailed transcripts, and raw recording files (MP3/WAV).
Ends an active AI call immediately or retrieves specific data about a single, completed call.
Ask AI about this MCP
Supported MCP Clients
Gemini Waiting for input…
Bland AI MCP Server: 10 Tools for Telephony & Call Flow
These tools let you manage the entire lifecycle of a phone call, from initiating the dial to analyzing the final transcript and pulling raw audio files.
019d755eanalyze call
Queries an active recording to determine if a specific goal was completed during the conversation.
019d755ecreate web call
Sets up a decoupled, internet-based WebRTC signaling socket for browser audio streaming.
019d755eend call
Forces an immediate disconnect on a live AI phone call.
019d755eget batch
Retrieves summary data about the concurrent status of a bulk call batch.
019d755eget call details
Gets the full transcript and specific variables for a call that has already ended.
019d755eget recording
Pulls the raw audio file (MP3 or WAV) for a specific call.
019d755elist calls
Retrieves the full historical record of all AI phone calls made or received.
019d755elist inbound
Shows all purchased PSTN numbers currently available for inbound AI routing.
019d755esend batch
Dispatches multiple AI agents at once to handle large-scale, concurrent calling campaigns.
019d755esend call
Sends a single automated conversational AI agent to a specified phone number.
Choose How to Get Started
Build a custom MCP for your own tools, or connect a ready-made integration from our catalog.
Build Your Own
Turn any API into an MCP. Import a spec, define Agent Skills, or deploy with MCPFusion.
- Import from OpenAPI, Swagger, or YAML specs
- Create Agent Skills with progressive disclosure
- Deploy to edge with MCPFusion framework
- Built in DLP, auth, and compliance on every call
- Real time usage dashboard and cost metering
- Publish to catalog or keep private
Make Your AI Do More
Start with Bland AI, then connect any of our 4,700+ other servers whenever your AI needs more. One click, no limits.
- Use this MCP plus 4,700+ others, all in one place
- Add new capabilities to your AI anytime you want
- Every connection is secured and compliant automatically
- Track usage and costs across all your servers
- Works with Claude, ChatGPT, Cursor, and more
- New servers added to the catalog every week
What you can do with this MCP connector
You can send conversational AI agents to specific phone numbers or run massive campaigns with send_call and send_batch. You'll find your purchased PSTN numbers listing with list_inbound, and you can set up browser audio streaming with create_web_call. You can pull the raw audio files—MP3 or WAV—for any given call using get_recording.
To check the history, you'll get the full record of every call made or received with list_calls. If a call ended, you can grab the full transcript and variables with get_call_details. You can analyze an active recording using analyze_call to see if a specific goal was met during the conversation.
If you need to shut down a live AI call right away, you use end_call. For a specific call that's already done, you can also get the details with get_call_details and pull the raw audio with get_recording.
How Bland AI MCP Works
- 1 Subscribe to the Bland AI server and provide your API key.
- 2 Your AI client calls a tool (e.g.,
send_callorlist_calls) to execute the desired telephony action. - 3 The server executes the action, managing the phone call or returning the requested data (e.g., a list of calls or a recording link).
The bottom line is you get programmatic control over calling and call data, treating the phone line like another API endpoint.
Who Is Bland AI MCP For?
Anyone who deals with high-volume communication—sales, customer support, or operations teams—and whose process involves talking to people over the phone. If your workflow relies on call data or needs to make outbound calls, this is for you.
Runs bulk campaigns using send_batch and then uses analyze_call on the transcripts to immediately determine lead qualification status.
Monitors inbound numbers via list_inbound and uses get_recording to pull raw audio for quality assurance checks on agent performance.
Tests conversational flows by using create_web_call to spawn decoupled WebRTC sockets directly from the IDE for debugging.
What Changes When You Connect
- Scale outbound campaigns instantly. Use
send_batchto dispatch hundreds of AI agents simultaneously, turning a manual campaign into a single tool call. - Stop guessing lead status. After a call, run
analyze_callto extract a definitive goal completion status from the transcript, saving hours of manual review. - Get the full picture of your calling volume.
list_callsgives you the historical log of every AI interaction, whilelist_inboundshows which numbers are ready for calls. - Improve quality assurance. Use
get_recordingto pull native MP3/WAV files for every call, letting you audit performance without leaving your agent environment. - Handle live debugging. If a call goes rogue,
end_calllets your agent client instantly disconnect the session, preventing wasted time and resources. - Build web clients.
create_web_calllets you spawn decoupled WebRTC sockets, meaning you can test phone logic directly in a browser without setting up complex hardware.
Real-World Use Cases
Auditing a failed sales campaign
A manager needs to know why a batch campaign failed. They run get_batch to profile the overall status. Then, they use list_calls to find the specific IDs of the failed calls, and finally use get_call_details on those IDs to pull the exact transcripts for review.
Testing a new web interface
A developer wants to test how their web app handles audio. They use create_web_call to spawn a WebRTC socket. This lets them run the call logic and test the audio stream directly in the IDE, without needing to connect to a physical phone line.
Monitoring incoming support lines
The support team needs to know what numbers are available. They call list_inbound to check all purchased PSTN numbers. If they need to test routing, they can use send_call to send a single test agent to one of those numbers.
Capturing evidence of a lead conversation
A salesperson completes a discovery call. To capture the evidence, they use get_recording to pull the raw MP3/WAV file, and then they use analyze_call on the transcript to get a structured, measurable outcome (e.g., 'High Interest').
The Tradeoffs
Calling the phone system directly
Trying to manually dial a number or use a general-purpose LLM to 'figure out' the call outcome, which results in inconsistent data structures and zero ability to track history.
→
Instead, use the dedicated tools. Run send_call to make the call, then immediately use get_call_details to pull the transcript and analyze_call to structure the outcome.
Ignoring bulk status
Running send_batch and then waiting until the end to see if it worked. You have no way of knowing which calls succeeded and which failed until the process is over.
→
After running send_batch, immediately check the status with get_batch to get a concurrent profile. This gives you a real-time overview of the batch's health.
Debugging in the wrong place
Trying to test WebRTC logic in a backend terminal without a visual interface, leading to complex, hard-to-read logging and broken local setups.
→
Use create_web_call to spawn the signaling socket directly. This gives you a decoupled, visual environment to debug the call flow without needing to leave your IDE.
When It Fits, When It Doesn't
Use this server if your process depends on structured, high-volume voice communication. Specifically, if you need to (1) automate calling, (2) track call outcomes programmatically, or (3) manage real-time call state. Don't use it if your needs are limited to simple message sending (use a messaging tool) or if you only need to analyze text files that were never spoken over the phone. If you need to debug a call flow in a browser, use create_web_call. If you only need call logs, list_calls is enough. If you need to measure performance on a batch of calls, use get_batch after send_batch.
Independent Platform Disclaimer: Vinkius is an independent platform and is not affiliated with, endorsed by, sponsored by, verified by, or otherwise authorized by Bland AI. All third-party trademarks, logos, and brand names are the property of their respective owners. Their use on this website is strictly for informational purposes to identify service compatibility and interoperability.
VINKIUS INFRASTRUCTURE
Cloud Hosted
Managed infra
V8 Isolated
Sandboxed per request
Zero-Trust Proxy
No stored credentials
DLP Enforced
Policy on every call
GDPR Compliant
EU data residency
Token Compression
~60% cost reduction
Works with Claude, ChatGPT, Cursor, and more
The Model Context Protocol standardizes how applications expose capabilities to LLMs. Instead of operating in isolation, your AI gains direct access to external platforms, live data, and real-world actions through secure, standardized connections.
This server provides 10 capabilities that interface natively with Claude, ChatGPT, Cursor, and any MCP client. No middleware. No custom integration required.
Available Capabilities
Manual call logging is a huge time sink.
Today, if a call goes poorly, someone has to manually check the CRM, dig into the call records, and copy the transcript to a spreadsheet. This process takes time—minutes per call—and it's prone to human error. You're spending your morning just aggregating data, not selling.
With this MCP server, the agent handles the data collection. You run `send_call` and then, when the call ends, you use `get_call_details`. The outcome isn't a messy log; it's a structured, queryable data object you can use instantly.
Bland AI MCP Server: Automate the entire call lifecycle.
Previously, to manage a big campaign, you'd have to write a complex, multi-step script involving multiple external APIs just to dispatch and track the calls. The coordination was brittle and difficult to debug.
Now, you use `send_batch` to dispatch the agents, and then you monitor the status with `get_batch`. The system handles the complexity of concurrent telephony arrays, letting you focus on the conversation flow itself.
Common Questions About Bland AI MCP
How do I check the outcome of a completed call using analyze_call? +
You must provide the specific call ID to analyze_call. This tool interrogates the raw transcript and returns a precise, structured status, telling you if the goal was met. It doesn't just summarize; it extracts a boolean status.
Can I make multiple calls at once using send_batch? +
Yes. send_batch is designed for this. It dispatches multiple conversational AI agents concurrently, which is necessary for large-scale, bulk telemarketing campaigns.
What is the difference between get_call_details and get_recording? +
Use get_call_details when you need the structured transcript and variables. Use get_recording when you need the raw audio file (MP3/WAV) for quality assurance or archiving.
How do I test my web application's audio streaming? +
Run create_web_call. This tool spawns a decoupled WebRTC socket, letting you test the audio signaling and client logic in a browser environment without calling a real phone number.
What is the difference between `list_calls` and `get_call_details`? +
list_calls provides a historical index of all your AI phone calls. This log gives you the call ID, date, and basic status. Use get_call_details when you need the full transcript and specific variables for one single call ID.
How do I ensure I can manage live calls using `end_call`? +
You must first have an active call ID to use end_call. This tool sends a synchronous interrupt flag to Bland AI, immediately shutting down the telecom session. It's perfect for handling rogue calls or interruptions.
Can I list my active inbound numbers using `list_inbound`? +
Yes, list_inbound retrieves all your purchased PSTN numbers currently connected to the AI agent's routing logic. It shows you the number and its current status (Active/Inactive).
What is the best way to scale up calling campaigns using `send_batch`? +
send_batch is designed for bulk telecom arrays, allowing you to dispatch multiple conversational agents at once. It handles concurrent calls, making it ideal for large-scale outreach or campaign testing.
Can my agent determine if a sales phone call was successful? +
Yes. By providing the specific Call ID and a structured goal query (e.g., 'Did they buy?'), your AI agent triggers the 'analyze_call' tool, running a targeted LLM pass exactly over that transcript to extract the explicit boolean outcome.
What if a voice agent goes off-script on a live call? +
You can instruct your agent to forcefully disrupt the conversation. Using the 'end_call' tool, it will immediately pass an interrupt flag and drop the active telecom connection without waiting for the other party.
Can I make AI agent calls directly through the browser instead of standard telephony? +
Absolutely. Your agent can execute the 'create_web_call' tool, which bypasses PSTN billing entirely and opens a pure WebRTC signaling socket for direct in-app or browser audio interactions.
Use it with your favorite AI tools
Connect this server to Cursor, Claude, VS Code, and more.
More in this category
Midjourney AI (Generative Image Arts)
Generate professional AI art via Midjourney — use 'imagine' for text-to-image, upscale grids, and perform camera edits.
DJI Developer
Orchestrate DJI drones and payloads — manage flight logs, monitor device health, and handle firmware updates directly from any AI agent.
Tesla Fleet API
Physical actuator proxy mapping explicitly native commands evaluating telemetry tracking active Tesla vehicles cleanly.
You might also like
Alexa Smart Home
Control Alexa-connected smart home devices — lights, thermostats, speakers, and sensors via Alexa Smart Properties API.
LearningSuite
Create and deliver corporate training programs with course authoring, quizzes, and progress tracking for distributed teams.
Clientjoy
Manage your entire client lifecycle from proposals and contracts through invoicing and payments in one streamlined platform.