Agentic AI for Research: Build Custom Workflows & Synthetic Users

No. It's recommended for researchers new to AI research tools, but not required. Course 1 builds foundation in the enterprise SaaS research stack (Listen Labs, Marvin, Coloop, Vurvey) and tool evaluation. If you've been using AI for research and are comfortable evaluating outputs, you're ready for Course 2. Many students take only Course 2.

Plan for 6-8 hours per week total: 2 hours for live sessions (or watching recordings), 2-3 hours for building agents and projects, 2-3 hours for setup guides and async materials. Building custom systems takes more iteration than using pre-built tools — that's where the learning happens.

You don't need coding skills — John isn't a coder either. But you do need willingness to learn new technical interfaces (an IDE and a terminal). We provide onboarding designed for non-technical researchers. The course emphasizes validation, ethics, and governance alongside building — so you don't just build agents, you build agents you can defend.

Required: a pro AI subscription. Any major plan works — Claude (Pro or Max), GitHub Copilot, Gemini, or ChatGPT Plus. This is what runs Claude Code or Codex, the tools you'll use to build agents. We'll guide setup in Week 1. We provide limited tech support — we're research experts, not IT — but you'll learn troubleshooting habits.

Yes. A 30-minute 1:1 with John for personalized guidance on your workflows. Weekly office hours where you dig into builds with John and classmates — students consistently call these one of the most valuable parts of the course. Direct access during live sessions and the community Slack.

We'll teach you to get help from the AI itself — the same way most developers do now. You'll learn to debug errors and interpret error messages using Claude or ChatGPT. We also provide step-by-step setup guides, weekly office hours, and a community Slack as backstops. This approach makes you self-sufficient rather than dependent on IT support.

You'll build a real system every week — Interview Analyzer, reliable pipeline, multi-agent research system, synthetic experts and users, and a strategic foresight capstone. In class, we build core components everyone can use; you then customize and extend each one for your specific context.

Core focus of the course. Week 2 (Reliable Agents) covers validation: Glass Box for inspecting outputs, evals for quality, Inspector Loops where one LLM checks another, and the "Beyond Vibes" framework. Week 4 applies these to synthetic users — known-response testing and confidence assessment. You'll create audit trails so your work is defensible.

Absolutely. If you're ready to build with AI for research, your role doesn't matter. We have PMs, designers, and founders building agentic systems alongside researchers. The course assumes no coding background — just curiosity and willingness to learn.

Yes. Many companies provide yearly professional development budgets. Check with your employer, manager, or HR about L&D funding. We're happy to provide course details to support your business case — emphasize the ROI of scaling research capacity by building custom systems that fit your team's specific workflows.

CARS is a two-course credential. This course earns Step 2 — the advanced builder credential. Pair with "AI for Customer Research" (Course 1) to earn the full CARS designation. You don't need Course 1 to take this course; many students take Course 2 alone. More at brilliantexperience.com/training.

Yes — Maven emails a certificate of completion after the course ends. You can share it on LinkedIn or with your employer. You can also optionally earn Step 2 of the Certified Agentic Research Specialist (CARS) credential — pair with Course 1 to earn the full designation.