# Running an agentic AI platform

AI agent platform (Primustech), 2026. Case study by Li ZhenXi. Page: https://portfolio-emsghin2iq-as.a.run.app/work/ai-platform

> Infrastructure for the company's two AI products: an operations agent that answers from live building data, and a document assistant.

- **2**: AI products: an operations agent and a document assistant
- **3**: environments deployed the same way: dev, staging and production
- **2,014**: BMS devices in the demo building I built for the agents to query

## The problem

The agent and RAG services needed repeatable deploys, locked-down secrets and numbers on speed and cost.

## What I did

1. Deployed and operated the multi-agent API and the RAG services on Cloud Run, with models on Vertex AI and OpenRouter.
2. Set up per-environment Cloud Build CI/CD, IAM service accounts, Secret Manager and shared-VPC networking for dev, staging and production.
3. Benchmarked the stack: load and latency tests of the agent API, per-request token cost tracking, and model comparisons on latency and cost.

## Result

Three environments deployed the same way, with building telemetry from the IoT platform feeding the agents.

Tools: Cloud Run, Vertex AI, Cloud Build, Secret Manager, IAM, VPC.

## Go further

- [Follow a request through the platform](https://portfolio-emsghin2iq-as.a.run.app/ai#ai-map)
- [The AI infra stack, layer by layer](https://portfolio-emsghin2iq-as.a.run.app/ai#ai-stack)
- [Inside one traced request](https://portfolio-emsghin2iq-as.a.run.app/ai#ai-trace)