VANCOUVER, BC, March 25, 2026 /PRNewswire/ -- Lanbow, an enterprise growth decision system built by Sandwich Lab, today announced the open-source release of Lanbow Claw Skill — a callable capability unit in the OpenClaw agent framework — that packages Meta ad execution logic and validation rules into an inspectable, reusable decision loop. The Skill is now live on ClawHub and GitHub.

With this release, Lanbow open-sources its single-campaign execution capabilities while directing its full resources toward the enterprise decision layer: cross-campaign capital allocation, multi-market strategy coordination, and continuously evolving decision models.

"We built this execution layer through hundreds of real campaigns and millions in ad spend. Now that Skills and CLI make it possible to ship that as a single callable unit, we're opening it up," said Dr. Zhenyu Guo, founder and CEO of Sandwich Lab. "Our focus going forward is the decision layer above it — where capital allocation, strategy coordination, and institutional learning compound over time. That's what Lanbow Enterprise delivers."

What's Open-Sourced — and What Isn't

Lanbow Claw Skill is the execution layer — made open. It connects four capabilities in one decision loop:

Insight Report — Structures campaign signals into decision-ready insight reports.

Creative Direction — Generates creative directions grounded in strategy inputs.

Meta Ad Launch — Translates strategic direction into executable Meta campaign configurations.

Performance Loop — Returns execution results to the decision layer for the next cycle.

One Skill connects all four stages into a continuous loop — observe, decide, execute, and iterate — each cycle compounding on the last.

Lanbow's enterprise decision layer — cross-campaign capital allocation, multi-market coordination, and reinforcement-based decision models — is available through Lanbow Enterprise.

Technical Proof Points

The Skill is 5,291 lines of TypeScript with 3 runtime dependencies. Zero global state, isolated dependency graphs, no side effects between calls — built for AI agent execution.

Over ~200 campaign sessions, Skill-based orchestration achieved a 92%+ campaign completion rate, up from ~60% in the earlier MCP-only phase.

Costly Meta API edge cases — budget mutual exclusion errors, bid strategy misconfigurations, non-standard video upload authentication — are embedded as local validation rules. Caught before requests are sent, not after money is spent.

Full technical details: lanbow.com/blog/lanbow-claw-skill

Industry Recognition

Sandwich Lab was selected for the AI Accelerate programme run by Microsoft and BLOCK71 (NUS Enterprise) in early 2026. In March, Dr. Guo joined the JUMPSTARTER 2026 Centre Stage panel hosted by Alibaba Entrepreneurs Fund, and Lanbow exhibited at the 2026 AWS Go Global Conference in Shenzhen.

Get Started

• ClawHub: https://clawhub.ai/LanbowAI/lanbow-claw-skill 
• GitHub: https://github.com/sandwichlab-ai/lanbow-claw-skill 
• Lanbow Enterprise: https://get.lanbow.com 

About Lanbow

Lanbow is an enterprise growth decision system built by Sandwich Lab. The company develops decision-layer software that helps enterprises coordinate signals, strategy, and execution across growth operations. Lanbow Claw Skill is Lanbow's open-source ad execution engine. The full enterprise decision layer is available through Lanbow Enterprise.

About Sandwich Lab

Sandwich Lab is an AI technology company founded in Vancouver, Canada, with teams across North America and Asia-Pacific. The company builds enterprise growth decision systems.

Website: https://lanbow.com

Source: Sandwich Lab