Agent Zero to Hero: Build Your First AI Agent with Grok Build

The problem

Most people's experience of "AI agents" is a chat window and a demo video. Then they point an agent at a real repo and things go wrong:

• It edits the wrong files. You approve a change you didn't read, and now the tests are red.
• You can't tell what it's doing. Tool calls fly past. You don't know which actions need permission, or why.
• It doesn't follow your team's rules. Every session starts from zero, so you repeat the same instructions.
• "Agent" stays a buzzword. You can't explain the difference between a workflow and an agent, or what keeps one from looping forever.
• Cost is a black box. You have no idea what a session costs, or why one run costs more than another.

None of this needs a PhD. It needs one structured day at the keyboard.

The transformation

Before: you've used ChatGPT or Grok chat. You've watched agent demos.

After one day:

• You can explain what makes software agentic (model, tools, loop, stopping conditions) and when a plain workflow is the better choice.
• You can install, sign in to, and run Grok Build in both the full-screen TUI and headless mode.
• You stay in control: plan mode, line-by-line plan comments, diff review, permission prompts, and /rewind (plus why /rewind rewinds the conversation, not your files).
• You've written an AGENTS.md and compared the agent's output on the same task with and without it.
• You've built a tool-calling agent on the xAI Responses API with a step limit, structured logs, and tests.
• You can estimate and track what a session costs with /usage (alias /cost) and grok usage.

Why this, why now: Agentic AI is the fastest-growing AI skill category employers are asking for (Forbes, Aug 2026). Coding agents are the agents professionals use most in daily work (LangChain, State of Agent Engineering, June 2026). Recruiters screen for builds you can show (GSDC). You leave with a repo.

What you'll build

About 75% of the day is hands-on, in real repos with real tests.

Lab 1 · Meet Your Agent (75 min) Install Grok Build, sign in, and run grok inspect. Explore taskflow-api, a small Python FastAPI to-do service. Watch every tool call and permission prompt. Add due-date filtering through plan mode: comment on a plan line, approve, review the diff, run the tests, and practice /rewind. Then write an AGENTS.md and run the same task again to measure the difference.

Lab 2 · Build an Agent From Scratch, With an Agent (60 min) Point Grok Build at a spec and a test suite. It builds a compact Python agent loop on the xAI Responses API with two function tools, a max-steps guard, and JSON logging of every tool call. You annotate the loop line by line in your own words. Then you run Grok Build headless and parse its JSON output with jq.

Capstone · Inbox Triage Agent (80 min build + demos) An agent that reads 25 customer emails, sorts each into bug, billing, feature, or spam, extracts action items through tool calls, and writes tasks.json plus a summary.md digest. It must refuse to act on low-confidence or adversarial emails, including one planted prompt injection. Required: at least 5 passing tests, a step limit, a run log, and a cost estimate. You check accuracy against an answer key, fix one weakness, and give a 2-minute demo.

You leave with: a portfolio-ready repo with a README, tests, a run log, and a cost estimate.

Who it's for

• Developers and software engineers new to AI coding agents
• Technical PMs, data analysts, and IT and ops professionals who want to build, not just prompt
• Founders who need to judge what agents can really do
• Anyone who has used ChatGPT or Grok chat but hasn't built an agent yet

Who it's not for

• Anyone who won't open a terminal. You'll use one all day, along with basic Git.
• People looking for a no-code or chatbot-builder class. Everything here is code, tests, and repos.
• Engineers who already ship tool-calling agents with evals and guardrails. Start with Course 2, 3, or 4.

Live on Zoom, instructor-led, about 75% hands-on. The join link is sent 4 days before class starts. A recording is included for paid attendees.

Logistics

FAQ

Do I need AI or machine learning experience? No. You need to be comfortable in a terminal, know basic Git (clone, commit, branch), and be able to read basic Python. You'll write very little code by hand. Most of the work is directing the agent and checking what it produces.

What do I need to pay for besides the seat? Each student is granted a Grok token certificate worth $25 for use in that class. It covers the Grok tokens and the xAI API-key credit used during the day.

What is Grok Build? Grok Build is the terminal coding agent from SpaceXSI (formerly xAI). It runs as a full-screen TUI, headless for scripts and CI (grok -p), or as an agent server for editors and custom apps. Its source code is public, so in class you can read the real agent loop behind the tool you're using.

Which model will we use? Whichever model Grok Build offers by default on the day. We confirm it in class with /model, and lab steps don't depend on a specific version.

Will this work on Windows? Yes, on Windows 11 with WSL2. macOS 13+ and Linux work too. Linux gives the fullest sandbox support.

What if I run out of Grok usage mid-lab? Usage limits are a real constraint on a full lab day. The instructor has a fallback plan (a shared key with a spend cap or an alternate model) so you can keep building.

Is this course affiliated with xAI or SpaceXSI? No. This is an independent course. See the disclaimer below.

Final CTA

One day. One agent. Built, tested, and understood.

Tuesday, October 27, 2026 · 9:00 AM–5:00 PM ET · $750

Register and pay, $750

Questions? Use the contact form.

Delivered by Vaughn Bullard · BDArgentumAI.

Disclaimer: Grok Build is a product of SpaceXSI (formerly xAI). This course is independent and is not affiliated with, sponsored by, or endorsed by SpaceXSI or xAI. Product names belong to their owners. Grok Build features and subscription plans change often; course content follows the official documentation as of the class date.

Sources for claims on this page: Forbes (Bernard Marr), "The AI Skills Employers Actually Want In 2026" (Aug 31, 2026); LangChain, State of Agent Engineering (survey of 1,340 professionals, June 2026); GSDC, Agentic AI Developer Jobs & Salary Report.

• ✓ Explain what makes software agentic: the model, tools, loop, and stopping conditions, and how an agent differs from a workflow.
• ✓ Install, authenticate, and operate Grok Build in the TUI and in headless mode.
• ✓ Use plan mode, diff review, permissions, and /rewind to stay in control of an AI coding agent.
• ✓ Write an AGENTS.md that measurably improves the agent's output on a real repo.
• ✓ Build a tool-calling agent on the xAI API with explicit step limits, logging, and tests.
• ✓ Estimate and track token use and cost for an agent session.

• ✓ Inbox Triage Agent repo with a README, tests, a run log, and a cost estimate
• ✓ Lab work in taskflow-api and agent-from-scratch
• ✓ A Grok token certificate worth $25 for use in this class
• ✓ Starter repos, sample datasets, lab guide, slides, and a plan-mode and permissions cheat sheet
• ✓ Live Zoom session, 9:00 AM–5:00 PM US Eastern
• ✓ Recording for paid attendees
• ✓ Certificate of completion (issuance criteria confirmed in class)

1. 9:00–12:00 · Morning
   What makes software agentic, then a lab in a small to-do service: plan mode, diff review, and an AGENTS.md you can compare.
2. 12:00–1:00 · Lunch
3. 1:00–2:45 · Afternoon
   The agent loop, then a small tool-calling agent on the xAI API with a step limit and a cost read.
4. 3:00–4:45 · Capstone
   Inbox Triage Agent. Sort 25 emails, refuse the low-confidence and injected ones, and give a short demo.
5. 4:45–5:00 · Wrap
   Takeaways and how the next day connects an agent to real systems.

A certificate of completion is part of the day. Issuance criteria are confirmed in class. Questions go through the contact form.