Building Tools with Tools: Modern AI-Driven Development
summary
Agents sit at the front of the SDLC and the workflow replaces slow manual implementation with a loop of idea→GitHub issue→agent→review→prod, using a text editor, a coding agent and a terminal as the three foundations. Practical moves include running AI review in CI to catch real bugs, using git work trees to run parallel agents, forbidding type ignores and enforcing mypy checks, treating MCP servers as the glue for capabilities, and using repeated evals to measure statistical improvement (about 40% of projects show vulnerabilities).
AI-written summary · about
chapters (17)
- 0:00Introduction and speaker intro
- 1:15About Sentinel One and AI in cybersecurity
- 2:42AI coding assistants landscape
- 4:28Modernizing software development lifecycle
- 6:39Three foundational blocks for AI-driven development
- 8:03Agent personas and parallel workflows
- 9:47Python environment and tooling
- 10:51First tip: CLI as your shell
- 13:17Second tip: Use multiple models
- 14:42Third tip: AI review in CI
- 16:03Avoiding agent misbehavior and hackery
- 18:29Context matters: Documentation and code
- 21:02Managing project dependencies
- 21:43Steering your agents with specs
- 23:49Role of MCP servers in the workflow
- 24:55Evaluations and business case alignment
- 25:59Closing remarks and hiring
related talks
26mAI Agents as Your Co-Developer
27mAgentic Engineering: Working With AI, Not Just Using It
25mFrom AI Chaos to Production Ready: Using AI Agents to Improve Your Code Quality
30mIncorporating AI into your SDLC
29mPairing with the machine: How AI Agents are rewriting the rules of software development
51mBeyond Code Generation: Governing AI Agents Across the DevSecOps Lifecycle
01h 02mSkills for the age of AI developer tools
14mAgent-Driven Software Development