Connecting Traditional Transport to the Agentic Future
How AI agents and MCP redefine mobility platforms
Pre-acquisition, defining the next phase of Meep. Author and strategic lead. Scope: product, data, platform architecture, and positioning.
Why this mattered
Mobility platforms had already solved aggregation. The next shift is not more integrations — it's intelligence on top of those integrations. AI agents introduce a new interface paradigm: from user-driven to agent-driven, from reactive to proactive, from planning to orchestration. Transport operators face a risk: AI systems access their data (often via scraping), without control, context, or monetization.
"How does mobility evolve when agents, not users, become the primary interface?"
Core ideas
- Mobility platforms evolve from aggregation layers → orchestration layers → intelligence layers
- AI agents shift interaction from apps to autonomous decision-making systems
- MCP becomes the bridge for agents to discover, understand, and interact with transport systems dynamically
- Winning platforms translate intent into execution across fragmented systems
- Transport operators must move from passive data providers to active participants in the agent ecosystem
Strategic insight
The key shift is architectural, not technical. AI agents handle the 'what' (user intent); platforms like Meep handle the 'how' — positioning mobility platforms as orchestration engines for agent-driven ecosystems.
What this influenced
- Repositioning Meep from MaaS platform to orchestration layer for AI agents
- Product direction toward agent-compatible infrastructure (MCP, real-time systems)
- Data strategy focused on context, not just availability
- Strategic narrative for partners, operators, and stakeholders
What I still believe
- Long-term value sits in the orchestration layer, not the interface
- AI agents will make apps less relevant as primary interaction surfaces
- Platforms managing complexity (like mobility) are positioned to become agent backbones
- Intelligence compounds when systems are connected — not when new tools are added
What I would update today
- AI lowers building costs but increases importance of control over data and execution layers
- Interface shift happening faster than expected (LLMs as default interaction layer)
- Speed of iteration is now a strategic advantage, not just operational
- Risk of disintermediation via scraping and uncontrolled AI access is higher than most operators realize