🧜‍♀️ The AI Collective Hampton Roads

Notice: This meeting will be recorded. By joining and remaining in this meeting, you consent to being recorded. If you do not consent, please do not join.

This event is capped at 500 attendees. Register now to secure your spot.

Build Your Own AI Agent: The Claude Agent SDK -

You've been using Claude Code interactively. Type a prompt, watch it work, review the output. That's great for your workflow. But what happens when you need Claude running inside your application? Processing uploads at 3 AM? Reviewing PRs automatically? Analyzing logs as a service?

The Claude Agent SDK is Claude Code packaged as a Python and TypeScript library. One function call, one prompt, and you have a fully autonomous agent with 10+ built-in tools — file reading, code editing, shell commands, web search, MCP connections — all handled autonomously. No tool loop. No orchestration code. The SDK is free and open source. You pay for tokens only.

This is a live coding workshop. We'll build 10 real agents from scratch, starting with a 10-line bug finder and ending with a multi-agent coordinator that spawns subagents for parallel work.

What we'll cover:

🧠 Why the Agent SDK Changes Everything The difference between the Client SDK (you build the tool loop), Claude Code CLI (you interact manually), and the Agent SDK (Claude handles tools autonomously in your code).

⚡ 10 Agents We'll Build Live:

1. Bug Finder & Fixer — reads code, finds bugs, fixes them. 10 lines of Python.
2. Code Documentation Generator — docstrings for every function in a directory
3. Test Writer — writes tests, runs them, fixes failures, iterates until green
4. File Processor Pipeline — watch a folder, process every new file automatically
5. API Health Monitor — hits endpoints, reports status, opens issues on failure
6. Log Analyzer — reads logs, groups errors by pattern, produces a digest
7. Security Auditor — scans for hardcoded secrets, SQL injection, missing validation
8. Data Report Generator — reads a database or CSV, produces formatted reports
9. PR Review Bot — triggered by webhook, reviews against your checklist, posts comments
10. Multi-Agent Coordinator — spawns subagents for parallel research + writing + review

🏗️ Managed Agents Preview The hosted version: sandboxed containers, persistent state, credential vaulting. $0.08/session-hour + tokens. Public beta since April 8, 2026. When to use Agent SDK (your infra) vs Managed Agents (Anthropic's infra).

Who this is for: Python or TypeScript developers who want to build autonomous AI agents. If you've been using Claude Code and want to embed its capabilities into your own applications, services, or pipelines — this is the session. Prior experience with Claude Code is helpful but not required.

Who this is NOT for: Non-developers. This is a coding workshop. Every demo involves writing and running Python or TypeScript code. If you're looking for no-code AI workflows, see our earlier events in the series.

What you'll leave with: ✅ The Agent SDK installed and a working agent running on your machine ✅ A free homework doc with all 10 agent templates in Python and TypeScript ✅ Understanding of built-in tools, permission modes, MCP integration, and subagents ✅ The architecture for building production agent services

Prerequisites: — Python 3.10+ or Node.js 18+ — An Anthropic API key (sign up at console.anthropic.com) — A code editor and terminal — Basic Python or TypeScript proficiency

🎙️ Your Speaker Andrew Stafford Co-Lead of The AI Collective Hampton Roads. 25+ years across technology, cloud architecture, and AI — including work with Amazon, NASA, healthcare platforms, and enterprise systems. Author of AI for Business. Currently building at the intersection of AI tooling, agentic workflows, and practical business automation.

New AI tools are dropping every week. Bookmarks are piling up. “To-be-studied” lists are growing faster than we can keep up with.
Sound familiar?

Peninsula Builders Study Group is a local initiative for students and developers of the Peninsula area who want to stop meaning to learn and start actually doing it—together.
There’s no pressure to be an expert, and no bootcamp intensity. Just a consistent habit, a room full of curious people, and the accountability that comes from showing up.

How it works:
We meet every two weeks on Tuesdays from 6:00–7:30 PM at CNU. Sessions are a mix of focused study and group discussion.

Things to bring:

• Laptop
• Earphones/headphones

#### About the challenge

Welcome to the AI for Learning & Development Build Challenge with the AI Collective Hampton Roads Chapter to be held at the Regent University Library in Virginia Beach.

This is a local build challenge with an in-person kickoff, followed by a two-week build phase, then an in-person presentation & judging event.

You'll have two weeks to build an AI application based on the problem statement shared at the kickoff event. This AI Build Challenge is open to adult participants of all skill levels, from experienced developers to beginner using no or low code tools. You're encouraged to use AI tools to further your development. After the build phase, you'll deliver an in-person presentation to judges and other attendees. The winning projects will receive prizes (TBA).

Kickoff on Thursday, June 4th from 6:30–8:00pm.
We encourage you to join us in person in the Regent Library Auditorum, but the kickoff will be also be recorded if you're unable to attend.

Build Phase from June 5th through 18th. You can work individually or in teams.

Presentations & Judging on Thursday, June 18th from 6:30–9:00pm. Return to the Regent Library Auditorium to share your work! Presentations must be delivered in person to be considered for judging and prizes.

#### GET STARTED

The first thing you should do is register through Devpost before the June 4th Kickoff.

You can work individually or in teams, but registration is important because space is limited. Then plan to come out to the Kickoff event to discover the problem statement and receive inspiration and development guidance from our AI Collective experts.

If you're unable to attend the in-person Kickoff, that's okay because we'll share a recording after the event with those who pre-register.

This is a beginner-friendly event. If you're never participated in a build challenge or hackathon, this is your opportunity! The Kickoff will include some initial guidance regarding topics and tools, and we'll make additional online resources available during the two week build phase.

### Requirements

#### WHAT TO BUILD

You'll build a project that uses AI to enhance learning and development (with more details unveiled at the Kickoff event). Projects will be evaluated on potential impact, not technical complexity.

#### WHAT TO SUBMIT

Prior to the in-person Presentations & Judging event, please submit the following in Devpost:

• Project Overview (e.g., PDF one-sheet, slidedeck, etc.)
• Code Repository (e.g., Github) and/or Project Link (e.g. website)

At the Presentations & Judging event, please prepare for the following:

• 3-5 minute presentation and/or demo
• 2-3 minute Q&A with judges
• In-person delivery (at least one team member in attendance)

### Judging Criteria

• Impact and Relevance
  Does the solution address a real learning or development need? How meaningful is the potential impact on learners, educators, or organizations?
• Innovation & Creativity
  How original is the approach? Does the team use AI in a novel or surprising way to solve the problem?
• Functionality & Demo
  Does the solution work as demonstrated? Is the core concept functional, even if not fully polished?
• User Experience
  Is the solution intuitive and accessible? Would the target user actually want to use it?
• Presentation Quality
  Is the team’s pitch clear, compelling, and well-structured? Do they effectively communicate the problem, solution, and value?

New AI tools are dropping every week. Bookmarks are piling up. “To-be-studied” lists are growing faster than we can keep up with.
Sound familiar?

Peninsula Builders Study Group is a local initiative for students and developers of the Peninsula area who want to stop meaning to learn and start actually doing it—together.
There’s no pressure to be an expert, and no bootcamp intensity. Just a consistent habit, a room full of curious people, and the accountability that comes from showing up.

How it works:
We meet every two weeks on Tuesdays from 6:00–7:30 PM at CNU. Sessions are a mix of focused study and group discussion.

Things to bring:

• Laptop
• Earphones/headphones