Use case

How do you route AI workloads across multiple LLMs without vendor lock-in?

Daena routes AI workloads across 9 LLM runtimes with hot-swap that needs no restart and no workflow rewrite, while every routed call still passes the full 10-stage governance pipeline — so switching models never relaxes policy.

9 LLM runtimes
Hot-swap, no restart
Governance applies to every model
PhiLattice patent-pending

Why route across multiple LLMs

No single model is best at everything, and committing to one provider is a business risk — pricing, availability, and capability all change. Multi-LLM routing lets you match each task to the right model and switch when the landscape shifts, without re-architecting.

The 9 runtimes

Claude · GPT · Gemini · Grok · Ollama · Groq · OpenRouter · Together · Perplexity

Cloud and local models behind one interface. Ollama covers fully local, zero-API-cost execution; the rest cover frontier and specialized models.

Routing and governance together

The point that matters for a business: routing does not bypass policy. Every call, on every model, still traverses the 10-stage governance pipeline and is written to the audit log. You get model flexibility without losing oversight.

Frequently asked questions

What is multi-LLM routing?

Multi-LLM routing sends each task to the model that best fits its cost, latency, capability, and policy requirements, instead of hardwiring one provider. Daena routes across 9 runtimes and can switch models without changing agent logic.

Which LLMs can Daena route across?

Nine runtimes: Claude, GPT, Gemini, Grok, Ollama, Groq, OpenRouter, Together, and Perplexity. Ollama enables fully local execution at no API cost.

Can I swap Claude for GPT without breaking my agents?

Yes. Daena hot-swaps runtimes without a restart and without rewiring workflows, because routing and policy live in the control plane rather than in each agent’s code.

Does switching models change governance?

No. Every routed call passes the same 10-stage governance pipeline and four-tier policy regardless of the model underneath, so changing providers never relaxes enforcement.

How does Daena decide which model to use?

Routing can factor in cost, latency, capability tier, and governance requirements per task. The architecture is covered by USPTO provisional patent #63/877,082 (PhiLattice agent topology).