Every team building AI products faces the same problem: the models are ready, but the infrastructure around them is not. You need agents that can reason across multiple steps, retrieval pipelines that surface the right context, memory that persists across sessions, and guardrails that keep outputs within policy — all working together reliably at scale.
That's what Liya Engine is. A modular orchestration platform that gives you every layer of the AI application stack as composable, production-hardened building blocks.
What's in the platform
Liya Engine is organised around four core primitives: Agents, Retrieval, Memory, and Guardrails. Each primitive is independently useful, but together they form a cohesive orchestration layer that can power anything from a simple Q&A chatbot to a complex multi-agent workflow.
Domain Packs
On top of the primitives, we've built Domain Packs — pre-configured bundles of prompts, retrieval strategies, output schemas, and guardrails tuned for specific verticals. Our first pack, Talent Intelligence, powers StoryHire and handles everything from resume parsing to candidate-role fit scoring.
We stopped thinking about which model to use and started thinking about what problem we're solving. Liya Engine made that shift possible.
Who it's for
Liya Engine is built for product and engineering teams that need to ship AI features fast, without owning the orchestration complexity. If you've spent months maintaining prompt infrastructure, RAG pipelines, or agent loop logic, Liya Engine collapses that into a single, versioned deployment.
What's next
Over the coming months we're launching additional Domain Packs for compliance, healthcare, legal, and financial services. We're also opening up the platform for external developers to publish their own packs to the Liya Engine marketplace.
The public beta is available now. You can get an API key and start building in minutes.