Show HN: Routing24 – free route optimization agent for Claude Cowork/WebMCP

What Happened

A developer has launched Routing24, a free route optimization and planning application for businesses, and is now actively building integrations with Claude AI via Cowork and WebMCP. The core innovation is enabling browser-based AI agents to interact with Routing24's route optimization engine—addressing a gap the developer identified: the inability of browser agents to work effectively with their existing web application. The tool processes natural language tasks like data ingestion, route planning, and optimization, allowing Claude to act as an intelligent dispatcher rather than just a chat interface.

Why It Matters

This integration represents a practical step toward making AI agents genuinely useful for operational business workflows. Route optimization is a classic "dull, dirty, and dangerous" task for human dispatchers—it involves repetitive data entry, complex constraint satisfaction, and real-time adjustments. By connecting Claude's reasoning capabilities to Routing24's optimization engine, the developer is demonstrating a pattern that could apply broadly across logistics, field service, and delivery industries.

The significance lies in the agentic nature of the integration. Rather than simply querying a database or generating text, Claude is being positioned to ingest unstructured data (customer addresses, time windows, vehicle capacities), transform it into structured optimization problems, execute the routing algorithm, and return actionable plans. This is a concrete example of AI moving from "chat" to "action" in a business context.

For AI practitioners, this highlights a growing trend: the value of AI agents is not in their general knowledge but in their ability to interface with specialized tools. Claude alone cannot solve a traveling salesman problem with 200 stops and 15 vehicle constraints—but Claude plus Routing24 can. The key architectural insight is that the agent acts as a translator between natural language and domain-specific APIs.

Implications for AI Practitioners

First, this case underscores the importance of tool-use architectures over monolithic AI systems. Developers building AI agents should focus on creating clean, well-documented APIs that agents can call, rather than trying to embed all functionality into the model itself. Routing24's approach—exposing route optimization as a service that Claude can invoke—is a textbook example of the "model as orchestrator" pattern.

Second, the free tier strategy is notable. By making the core optimization engine free, the developer is lowering the barrier for experimentation. This is a smart move for building an ecosystem around agentic workflows, as it allows developers and businesses to test the integration without financial risk. For AI practitioners, this suggests that pricing models for agent-ready tools may need to shift toward usage-based or freemium models to encourage adoption.

Third, the focus on browser-based agents (via WebMCP) addresses a real pain point. Many businesses want to use AI agents to automate web-based workflows but struggle with the fragility of browser automation. Routing24's integration suggests that providing structured APIs specifically designed for agent consumption—rather than scraping or simulating human clicks—is a more robust path forward.

Finally, this development signals a maturation of the AI agent ecosystem. We are moving beyond toy demos toward production-grade integrations where AI agents handle real business constraints: time windows, vehicle capacities, driver hours, and customer preferences. Practitioners should watch for similar integrations in adjacent domains like inventory management, workforce scheduling, and supply chain coordination.

Key Takeaways

• Agent-tool integration is the critical enabler for practical AI automation in logistics; Claude acts as orchestrator, Routing24 as domain expert.
• Free access to specialized APIs lowers the barrier for experimentation and ecosystem growth, a model worth emulating for agent-ready tools.
• Browser-based agent workflows benefit from purpose-built APIs rather than fragile web scraping or simulated clicks.
• Real-world constraints (time windows, vehicle capacities) are now being handled by AI agents, marking a shift from demos to production deployments.