Products & Tools

Liner launches an MCP connector for Claude and ChatGPT

Liner’s free MCP connector links Claude and ChatGPT to more than 1,100 tools. Its API and claimed scale arrive alongside a $36.1 million round and more than 14 million users.

Liner launches connector for Claude & ChatGPT tasks
Liner launches connector for Claude & ChatGPT tasksDaniil Komov / unsplash
By James Calloway6 min read

Why it matters

  • Liner Actions MCP links agents including Claude and ChatGPT to more than 1,100 applications through one integration.
  • Liner says it tested more than 1,100 verified tools across multiple environments.
  • Liner Actions MCP and Liner Actions API are currently free and support Korean and English.
  • Liner disclosed a $36.1 million Series C round and reports more than 14 million users across 220 countries.
  • Liner cited Similarweb data ranking it 30th among global AI services in late July 2026.

By James Calloway

According to Liner, Liner Actions MCP links agents including Claude and ChatGPT to more than 1,100 applications through one currently free integration. The South Korea-based company launched the Model Context Protocol (MCP) connector for AI agents (software that acts on a user's behalf) working across third-party services.

Liner says users can “connect once” and then ask an agent to search for information or complete tasks across Slack, Notion, GitHub and calendar applications. The company says users do not need to configure every application in advance. Liner says the release targets administrative work that has grown as agents connect to more outside services.

MCP has emerged as a common way to connect AI agents to external tools, but the growing number of servers has added setup and maintenance work for users. The setup has included installing, configuring and managing a separate API key (a software credential) for every application.

Liner says its relay removes that setup burden. The announcement gives no figures for time saved, failure rates or maintenance avoided, so the claimed reduction remains unmeasured.

Takeaway: Liner is selling fewer connection steps, not another AI model.

How does Liner’s relay reduce setup?

Liner Actions MCP sits between an AI agent and each service's MCP server. When access is first needed, Liner links the account through a single login, rather than requiring preconfigured credentials for every application. The announcement does not specify how access permissions are scoped.

The catalogue contains more than 1,100 verified tools. Liner says it tested each one across multiple environments to check that the integrations worked as expected. That testing claim distinguishes the catalogue from a generic directory, although Liner has not published the environments, test cases, pass rates or underlying results.

The announcement alternates among applications, tools and services when describing scale. It does not explain whether every application contributes one tool, whether one service can expose several tools, or whether the count changes as Liner adds integrations. The headline total is clear, but the unit behind it is not.

Takeaway: The relay can cut setup work, but Liner leaves the catalogue’s measurement and test evidence unstated.

How does an agent use several apps in one prompt?

Liner says tools load only when needed, rather than all at once at the start of a request. For an agent, context means the instructions, retrieved information and tool details it can process at once. On-demand loading reduces that burden during a multi-step request, according to Liner, and helps an agent manage work spanning several applications from one prompt.

One example starts with a request to summarize a week of discussion in a Slack marketing channel and save the result in Notion. In Liner's described workflow, the agent identifies the two required services, retrieves the information, completes the summary and writes it to Notion. The user does not configure each connection in advance.

That example shows the intended benefit, but it is not an independent performance test. Liner's announcement provides no completion rate, latency measurement or failure data for cross-service work. It also does not explain what the agent should do when a required service rejects the request or returns incomplete information. Those omissions matter because completing a task covers several actions beyond sending a prompt to a model.

Takeaway: On-demand tools support useful multi-app workflows, though the announcement documents the happy path only.

What does the API add?

Alongside the connector, Liner offers the Liner Actions API, an application programming interface for companies building their own AI agent products. The API reaches the same catalogue of about 1,100 services. Liner says its own AI writing tool, Liner Write, already uses connector functions built on this interface.

Liner Actions MCP and Liner Actions API are currently free to use. Both support Korean and English. Liner says it plans to add more verified tools in response to user feedback, which gives prospective users a route to shape the catalogue but supplies no timetable for expansion.

The announcement does not specify usage limits, service-level commitments, a paid tier or a pricing schedule. It also does not detail the permissions an organization must grant when employees connect company accounts. Those omissions leave the operating terms for a production agent workflow undefined.

Takeaway: Developers can reuse Liner’s connector layer now, but its production terms remain unspecified.

How large is Liner?

The South Korea-based company says the launch follows its disclosure of a $36.1 million Series C funding round. It reports more than 14 million registered users across 220 countries, including the US, Europe, India and Southeast Asia. Those figures describe company reach, but registered-user totals do not show how many people use Liner Actions MCP or the API.

Liner cited Similarweb web traffic data showing that it ranked 30th globally among AI services in late July 2026. The company said this placed it ahead of services including Poe, Mistral AI and Glean.

The announcement does not identify the ranking's traffic metric, measurement window or comparison set. The order therefore does not by itself show direct product competition with those services.

Liner also said it has appeared on Andreessen Horowitz's Top 100 GenAI Consumer Apps list for four consecutive editions since 2024. The list concerns consumer apps, so it does not show enterprise adoption of Liner Actions MCP or Liner Actions API.

Liner says it has expanded beyond consumer products into business software by drawing on user data and an academic database containing more than 500 million records. The announcement does not identify the academic database, its licensing, update cycle or how those records enter Liner's products.

Takeaway: Liner has a large consumer footprint, but the announcement does not measure demand for its agent products.

What does Liner say about user data?

Liner says its system “does not store user data.” It says operational logs are de-identified (stripped of identifying details) and designed to prevent the company from viewing details of users' activity. The statement gives the connector a specific privacy posture, but the announcement provides no independent audit, retention schedule or technical description of the de-identification process.

The privacy claim concerns operation of the system. Elsewhere, Liner describes using user data and a database of more than 500 million academic records as inputs to its expansion into business software. The announcement does not explain how those broader uses relate to the no-storage statement or which products it covers.

Takeaway: Liner offers a specific privacy promise, but its scope and verification remain unclear.

What has to work next?

Liner says user feedback will drive additions to the verified-tool catalogue. The immediate test is whether the free MCP connector and API attract developers while the company expands beyond the current total of roughly 1,100 services. It must preserve single-login operation and on-demand loading as the integration count rises, because those features carry Liner's pitch to users and developers.

The announcement offers no launch targets, adoption metrics or schedule for further tools. Its next claim to test is operational: whether agents can complete cross-service tasks reliably as the catalogue expands beyond roughly 1,100 services. Liner says user feedback will guide tool additions, so it must show that more integrations do not weaken its verification, single-login or privacy claims.

Takeaway: Liner must turn connector breadth into dependable agent actions without weakening its privacy or verification claims.

Source: itbrief.com.au

Share this article:

More from James Calloway

James Calloway

Show full bio

News editor covering people, launches and day-to-day practice in artificial intelligence and machine learning at AI In Context.

722 articles

Related articles

  1. Adobe Opens Creative Cloud to Outside AI Agents via Expanded MCP
  2. Adobe will open Creative Cloud desktop apps to outside AI agents via MCP
  3. Islo Raises $15M to Scale Its Enterprise Software Factory
  4. Figma Adds 20 MCP Partners While Its CEO Calls the Protocol Flawed
  5. Adobe Opens Photoshop and Premiere to Rival AI Agents in 2027 Beta

« Previous articleNext article »