Model Context Protocol · 7 channels · write access

Your reviews, right inside Claude.

Audienca is an MCP server, and one with write access. Connect it once to Claude, ChatGPT or Cursor: your assistant reads open reviews, drafts replies from your company knowledge, and once you have edited and approved them they appear on Google, Instagram, Facebook or YouTube. No tab switching, no copy-paste.

Start free trial→Set up in Claude Code

Included in every plan, trial included. 14 days free, no credit card.

Inside the Claude window
You

Show me the open Google reviews for Hotel Seeblick and draft answers for the critical ones.

Claude, through the Audienca MCP

Three reviews are open, two of them critical. For the 2-star review about breakfast I drafted a reply that mentions your refurbishment, which your knowledge base says finished in March. Shall I submit it for approval?

Illustration of a real flow: list_inquiries, get_project_config and create_inquiry working together.

  • 51 tools
    Work, analyse, channels, knowledge, automation
  • OAuth with scopes
    audienca:read and audienca:write kept apart
  • Hosted in Germany
    sense.AI.tion GmbH, GDPR compliant
The question first, the product second

Why a review tool should be an MCP server.

The Model Context Protocol is the standard AI assistants use to talk to software. SDK downloads went from 100,000 a month in late 2024 to 97 million by early 2026. Most connections may only read: Claude's own Google Business Profile access, for instance, can analyse reviews but not answer them. Audienca writes back.

What the MCP access is built for

  • Review incoming feedback inside your assistant, without opening a second window.
  • Draft replies in the conversation and submit them for editing and approval. After that they really go out: to Google, Instagram, Facebook, YouTube, Trusted Shops, Google Play and the App Store.
  • Maintain the knowledge base: upload FAQs, product docs and shipping rules so answers hold up.
  • Create and configure client projects — for agencies, the whole per-client routine.
  • Read, change and switch off automation rules without going to the dashboard.

What it is not for

  • Publishing without review. Audienca can write, but only once you have approved. The four-eyes step stays, over MCP as well.
  • Replacing the dashboard. Analytics and team management stay there.
  • Access without permissions. Every call carries your account and the scopes you granted.
  • A side project. The server runs on the same system as your production account.

What that actually saves

Industry reports put a custom MCP server at 50,000 to 150,000 dollars per integration per year, and write access to Google's API carries a waiting list of more than two months. With Audienca both are done: reading and writing, on every plan, at no extra cost.

The next sections show what that looks like day to day.

Where things stand in 2026

Why this matters right now.

MCP went from proposal to standard in eighteen months. The numbers are unusually clear-cut.

97M
SDK downloads per month

Early 2026, up from 100,000 in November 2024.

Industry reports 2026
$50–150k
per integration per year

Cost of building and running your own MCP server.

Market analyses 2026
51
tools in the Audienca MCP

Work inquiries through to sending, analyse, bulk changes, channels, knowledge base, automation.

audienca-sidekick 2.0
1
connection for every client

Claude, ChatGPT, Cursor and any other MCP-capable assistant.

Model Context Protocol

MCP access is included in every plan, from the free trial through Enterprise.

Start free trial→
Three steps

How to connect your assistant.

No SDK, no server of your own, no maintenance. The endpoint speaks streamable HTTP and is stateless.

Create access

Generate an OAuth client in the dashboard and pick the scopes. audienca:read alone is enough if the assistant should only read.

Add the endpoint

Register the MCP server address in Claude, ChatGPT or Cursor. The server advertises its tools and schemas itself.

Start working

From then on your assistant knows your projects, reviews and knowledge base. Approvals still run through you.

Common questions

What teams want to know before connecting

No, and that is a setting, not a technical limit. Audienca is allowed to write on Google and Meta, but only does so once you have approved the draft. The four-eyes step sits exactly where it does in the dashboard. If you want an even tighter leash, grant only the audienca:read scope.
Any that speaks the Model Context Protocol. Tested with Claude and Cursor; the standard is open and client-agnostic. One server, every client.
On the same servers as your Audienca account, in Germany. The MCP endpoint is not a second system, it is another way into your existing one. No copies are created.
Nothing extra. MCP access is included in every plan, from Solo through Enterprise, and during the 14-day trial. There is no separate activation.
The token becomes invalid immediately and running connections drop. There are no cached permissions, because the server is stateless.
Not to set it up. You create an OAuth client in the dashboard and paste the address into your assistant. If you want to build your own flows, every tool ships a complete JSON schema.

Answer reviews without switching windows.

MCP access is included in every plan. Connect your assistant and try it.

Start free trial→
  • Included in every plan
  • 14 days free, no credit card
  • Hosted in Germany

About the names on this page

Claude and Claude Code are trademarks of Anthropic. ChatGPT is a trademark of OpenAI. Cursor is a trademark of Anysphere. None of them belong to us, and none of these companies endorse or sponsor Audienca. We name them because our server works with their products — that is all the connection there is.