# Automation with project memory.

> A delivery team that wanted AI to help with support and projects, with context and control.

URL: https://www.talkit.pt/en/cases/ai-operations/
Language: English · Português: https://www.talkit.pt/pt/cases/ai-operations/
Case study · AI operations

## The challenge

Tickets, repositories and databases held the knowledge, but assistants could not reach it safely and every answer started from zero.

## Our approach

1. **Give AI context** — MCP tooling connects assistants to tickets, repositories and databases with scoped access.
2. **Automate the pipeline** — Omnichannel support tickets are classified, enriched and routed automatically.
3. **See every step** — Observability shows what each agent did and why.

## What changed

- Assistants answer with project history, not guesses.
- Support tickets arrive triaged and enriched.
- Secure, auditable data access for AI.

## Stack

MCP, AI agents, PostgreSQL, SQL Server, Observability, TypeScript

## Next case

[Finance & credit: Credit journeys, end to end.](https://www.talkit.pt/en/cases/finance-credit/) · [Start a project](https://www.talkit.pt/en/contact/)