Building a private AI search workflow for internal operations.

The Information Tax was killing us. The ops team was drowning in "where do I find X?" questions. SOPs were scattered across Google Drive, Notion, and Slack threads. Every time a tiered support ticket came in, someone had to manually search for the right protocol.

The solution is rarely another wiki page. The useful pattern is a workflow system: a retrieval layer that finds relevant source material, drafts an answer, and shows what it used.

Here is how I would structure a private AI search workflow for internal operations.

1. Design & Infrastructure

For teams already on Google Cloud, Vertex AI Search can be a practical starting point because it handles retrieval without requiring a custom vector stack on day one. For other teams, the same pattern can be built with Postgres, pgvector, and a small application layer.

The Setup:

• GCP Project: Dedicated project for the search infrastructure to isolate billing and access.
• Data Sources: Connected Google Drive (for SOPs) and a dump of resolved Jira tickets (as CSVs) to the Vertex Data Store.
• Interfaces: Test a search view and a chat-style view before deciding where the workflow should live.

2. The System Architecture

The core of the workflow system is the retrieval loop. We didn't just want a search bar; we wanted answers delivered where the work happens (Slack).

The Workflow:

1. Trigger: New question in the #ops-support Slack channel.
2. Orchestration (Cloud Run): A lightweight Python service receives the webhook.
3. Retrieval: The service calls the Vertex AI Search API with the user's query.
4. Grounding: The LLM (Gemini via Vertex) summarizes the search results, citing the specific source documents.
5. Response: The answer is posted back to the Slack thread with links to the original Drive docs.

Critical Feature: "Not in Corpus" Behavior. We tuned the system to explicitly say "I cannot find that in the SOPs" if the confidence score was low. This prevents hallucinations and flags gaps in our documentation.

3. Outcomes & Metrics

The metrics worth tracking are practical, not theatrical.

• Search time: How long it takes a team member to find the right answer.
• Review rate: How often the system routes unclear questions to a person.
• Documentation gaps: Which “not in corpus” questions reveal missing SOPs.