# planoai.dev

> AI-optimized mirror of planoai.dev containing 8 pages totalling 4,188 words of clean markdown content, structured data, and semantic HTML. Original source: https://planoai.dev/. Last updated: 2026-04-30T21:53:27.031Z.

Each page is available as HTML (with JSON-LD structured data) and Markdown (text-only, ideal for LLMs and RAG).

## Homepage

- [Delivery Infrastructure for Agentic Apps](/site-root.html): Build agents faster and deliver them reliably to production. Plano is an AI-native proxy and data plane for agent orchestration, LLM routing, guardrails, and observability. (363 words)

## Articles & Blog Posts

- [Research](/research/index.html): Explore Plano's applied AI research focusing on safe and efficient agent delivery. Discover our orchestrator models, benchmarks, and open-source LLMs on Hugging Face. (449 words)
- [DigitalOcean acquires Katanemo Labs, Inc. to accelerate AI development](/contact/index.html): Get in touch with the Plano team. Join our Discord community or contact us for enterprise solutions for your AI agent infrastructure needs. (70 words)
- [Detecting Hallucinations in LLM Function Calling with Entropy (Part 2)](/blog/detecting-hallucinations-in-llm-function-calling-with-entropy-part-2.html): A introduction to detecting hallucinations from LLMs for structured outputs using entropy and varentropy. (1,299 words)
- [Preference-aware routing for Claude Code 2.0](/blog/preference-aware-routing-for-claude-code-2-0/index.html): Preference-aware routing for LLMs for Claude Code 2.0 (697 words)
- [Are prompts the new unit of work in applications?](/blog/are-prompts-the-new-unit-of-work/index.html): In the evolving landscape of AI-native systems, prompts are emerging as the primary unit of work, replacing traditional API calls. This shift from structured requests to open-ended instructions in natural language necessitates a new layer of infrastructure: Layer 8. Unlike Layer 7, which routes based on protocol semantics, Layer 8 routes based on the semantic meaning of prompts—understanding intent, selecting models, and driving observability by goal rather than endpoint status. This paradigm shift is essential for developers building intelligent, prompt-driven applications in today’s AI-driven world. (764 words)
- [Why Katanemo is joining DigitalOcean](/blog/why-katanemo-is-joining-digitalocean/index.html): Katanemo is joining DigitalOcean — bringing the Plano project, our open-source agentic data plane, and our model research into the platform to give developers the infrastructure primitives they need to go from AI prototype to production, without the glue code. (422 words)
- [What's new with Plano](/blog/index.html): Latest insights, tutorials, and updates from Plano. Learn about AI agents, agent orchestration, LLM routing, and building production-ready agentic applications. (124 words)

## Resources

- [Full Page Index](/index.html): Browse all cached pages with rich metadata
- [About This Cache](/content/about.html): Methodology, technical details, and usage guidelines
- [XML Sitemap](/content/sitemap.xml): Machine-readable sitemap for crawler discovery
- [Robots.txt](/content/robots.txt): Crawler directives