Logly's AI Ad Ops Framework 'mureo' Supports Claude Desktop App, Enabling Anyone to Build an 'AI Ad Ops Crew'

Logly Inc. (Headquarters: Shibuya-ku, Tokyo, CEO: Hirokazu Yoshinaga, Ticker: 6579) announces that its open-source AI ad operations framework, 'mureo' (https://mureo.io), now supports use within Anthropic's Claude Desktop app.

Example screen of mureo's /daily-check running on Claude Code within the Claude Desktop app. It diagnoses campaign health across Google Ads and Meta Ads (demo using synthetic data).

With Meta and Google successively releasing official Model Context Protocol (MCP) for their ad APIs, mureo's role as a 'higher layer above official MCP', envisioned since its initial design, has become clearer. mureo is provided not as an API connection itself, but as an AI agent team handling ad ops strategy, safety mechanisms, auditing, and learning. Under the concept of 'your local-first AI ad ops crew', it offers three implementation routes: Demo mode, BYOD (Bring Your Own Device) mode, and Live mode. This provides an environment where even non-engineering executives, marketers, operations managers, and agency representatives can launch an 'AI ad ops crew' in just 5 minutes.

■ Background

In October 2025, Google released the Google Ads MCP, followed by Meta announcing the Meta Ads MCP in April 2026. With major ad platforms sequentially establishing official MCPs, the function of 'AI directly hitting ad APIs' is expected to become commoditized.

mureo was developed from the beginning with a design philosophy to operate in the layer above, anticipating that each ad platform would provide an official MCP (it initially included API wrappers because official APIs were non-existent or access was limited). The release of official MCPs by these companies has ushered mureo into a phase where its intended role—not just connecting to APIs, but serving as a 'higher layer' where AI understands business strategy to judge, execute, and audit ad operations—can be more clearly communicated.

■ Concept

mureo's concept is 'your local-first AI ad ops crew'. It operates not as a single tool, but as an AI team sharing the multiple roles required for ad operations.

Strategy Formulation: Judgments based on STRATEGY.md (persona, USP, brand voice, goals)
Delivery Diagnosis: Cross-monitoring via /daily-check
Budget Allocation: Cross-platform optimization via /budget-rebalance
Creative Proposals: Ad text and LP alignment via /creative-refresh
Emergency Response: Handling performance degradation via /rescue
Weekly Reporting: Summaries for management via /weekly-report
Learning Accumulation: Persisting account-specific knowledge via /learn

By operating these within a single framework, users can leverage AI with the sensation of 'forming a team' rather than 'adopting a tool'.

■ Three Operating Environments (Modes × Hosts)

In addition to the Live mode announced at launch, mureo offers implementation routes through combinations of 3 modes and 3 hosts.

[3 Modes (by Data Source)]
Demo Mode: Check mureo's operation instantly using pre-prepared scenarios (no OAuth or data required)
BYOD Mode: Analyze simply by uploading Google Ads / Meta Ads XLSX files (no OAuth or Developer Token required, initial diagnosis in 5 minutes)
Live Mode: Full support for production operations and execution of changes via OAuth authentication

[3 Hosts (by Execution Environment)]
Claude Code (Both within Claude Desktop app and CLI): CLI for technically proficient users, fully supporting MCP and skill features
Chat (Within Claude Desktop app): Operable via natural language, accessible to non-engineering operators
Cowork (Within Claude Desktop app): Sandbox operations referencing folders in addition to chat

Notably, with the newly added local web UI for setup, 'mureo configure', even operators, agency representatives, and executives previously unfamiliar with terminals or command lines can now implement it in minutes by following the on-screen browser instructions.