How do I actually get an AI agent to finish a coding task and hand me something I can merge? In this module, you will give Devin a small backend task and receive a pull request that Devin opened, tested, and described on its own. You will learn what happens between the prompt and the pull request: how a session moves through its stages, when Devin pauses to ask for input, how it verifies its work, and how you take over its changes. You will launch a session the right way by choosing the repository, the capability level, and the context Devin needs, learn to steer Devin mid-task with interactive planning and chat messages, and apply a prompt structure built on context, goal, constraints, and a clear definition of done. The module is designed for software developers, tech leads, and engineering managers who are new to AI coding agents. You need basic familiarity with Git and pull requests, but no previous experience with Devin or any other agent. In the hands-on lab, you will choose a small backend feature, such as a version endpoint, hand it to Devin, steer the session, and review the resulting pull request. You will leave with a reusable task-prompt template and a completed pull request that proves you can take a request from a single message to reviewable code.

How can I understand a large codebase I have never seen without spending days reading files? In this module, you will produce a one-page architecture brief that answers five onboarding questions, each backed by a source-linked citation. You will learn how repository indexing turns code into a generated wiki, how Ask Devin uses that wiki and code search to give grounded answers, and how curated Knowledge differs from generated documentation. You will browse an auto-generated DeepWiki for a public repository, ask Devin where a feature is implemented, check its citations, and launch a work session straight from an answer. You will also use AI to generate and maintain developer documentation, steer wiki content with a configuration file, and check generated pages for invented details. The module is designed for developers joining a new team, reviewers who need context quickly, and tech leads who onboard others. You need to be able to read code in a common language; no prior use of Devin is assumed. In the lab, you will question a sample microservice about data flow, authentication, configuration, error handling, and deployment. You will leave with a cited architecture brief and a repeatable question set you can run against any repository before you start changing it.

Which engineering tasks should I hand to an AI agent, and how do I make sure the result is actually correct? In this module, you will turn a stakeholder's bug report into a scoped ticket and get a Devin pull request containing a fix plus a regression test that fails on the original code and passes on the fix. You will learn to size work at a junior-engineer level, write acceptance criteria, and sort tasks with a delegate, pair, or keep decision grid. You will watch Devin split a long function into smaller units and review the diff, save a dependency upgrade as a reusable playbook, and use AI to raise test coverage without producing brittle tests. In a role play, you will question a product manager who reports a vague bug in Slack and agree on a clear ticket. The module is intended for developers, tech leads, and engineering managers who want to offload routine bug fixes, tests, refactors, and upgrades. You should know how to read a diff and run a test suite; no previous Devin experience is required. In the lab, you will delegate a seeded date-parsing bug and prove the new test catches it. You will leave with a ticket template, a playbook pattern, and a verified, merge-ready pull request.

How do I let an autonomous agent touch our codebase without creating security or quality risks? In this module, you will get sign-off on a Devin usage policy that covers secrets, permissions, review gates, and task-size limits. You will study where autonomous agents fail, including ambiguous requests, oversized scope, invented APIs, and silent shortcuts, and learn the early signals that tell you to stop a session. You will store secrets in Devin without pasting them into prompts, ask Devin to address pull request review comments on GitHub, and use AI reviewers such as Devin Review alongside branch protection and required CI checks. You will also compare team entry points, including Slack, Linear and Jira, the CLI, and the API, and choose metrics such as merge rate, rework, and time saved to track before scaling. The module is designed for tech leads, engineering managers, and security-minded developers responsible for adopting AI agents. You should understand pull request workflows and basic access control; no prior Devin experience is needed. In the role play, you will defend your written policy to a skeptical engineering manager. You will leave with an approved governance policy, a human review checklist, and a short metrics plan for a team rollout.