From Chat to Agents: How to Use AI & Agentic Workflows

Most teams still use AI as a chat window. In this recorded session Mārcis shows what it takes to go further: giving AI the context, constraints, tools and review process it needs to produce reliable, high-quality development work, and running it as a set of agents that plan, implement, test and review. Mārcis specialises in AI-assisted software architecture, agentic development pipelines and automated multi-agent orchestration, helping development teams and digital businesses integrate AI tools securely, cut out QA back-and-forth and scale their operations.

What you will learn

• How to set up context-first AI development workflows
• How to avoid session hygiene pitfalls and context dilution
• How to follow the six-step agent workflow: plan, review, justify, implement, document, end-to-end test
• How to configure MCP connectors and keep skills lean (about five per project)
• How to deploy multi-agent orchestration and PR-review agents
• How to run headless CLI automation with safety guardrails
• Security best practices: SSH keys, test servers and CI/CD gates

What the session covers

1. Context-first AI and project knowledge
2. The agentic workflow: from plan to evidence
3. MCP connectors and live validation
4. Skills and context hygiene
5. Multi-agent orchestration and PR review
6. Headless automation, guardrails and next steps

Level: Intermediate. Length: 33 minutes. Cost: free. Prefer to take it step by step? The same material is also a six-lesson course: How to Use AI & Agentic Workflows.