Agentic Engineering & Claude Code
Most developers already use agents. Agentic engineering is the method around them: specs, guardrails, verification gates, and parallel agents. In six weeks you build it into how your team works.
Format
Results
Graduates self-report a median 5× speedup on their final project.
Curriculum
6 Sessions: Ad Hoc Agents to a Repeatable Method
Curriculum built on Claude Code
6 training days · 9:00–15:30 · Google Meet
Using a different agent? Why Claude Code →
A typical Friday
Watch it done live, then build it yourself
The instructor works in Claude Code on a shared screen. Then you try it on your own project in a breakout room of three. Each day closes with a short quiz and a feedback round; the next Friday opens with what you said.
- Recap and your feedback
- Slides and live demo
- Workshop in breakout rooms
- Teams present
- Next topic, live
- Lunch
- Demo
- Team task in breakout rooms
- Quiz and feedback
Ends at 15:30. Short breathers between blocks.
Harness Context Verification Scale
- Claude Code and the agentic loop: prompts, tokens, skills, subagents, MCP, plans and pricing
- Hooks: observe every tool call, block the dangerous ones
- Set up Claude Code on your own project, or on our sample repo, then /init its CLAUDE.md
- Add a hook that runs your linter or tests on every change; your breakout room then tries to get past it
- Maps of the project: layered summaries, the C4 model, tokens vs the prompt cache
- Design docs, the testing pyramid, validation loops and subagents in depth
- Build a C4-model skill and a DB-schema map, wired into CLAUDE.md
- Prompt engineering, write-then-compare: write yours cold, run it, then run the reference. The delta is the lesson
- Write a spec with acceptance criteria the agent can verify, checked by parallel validation agents
- From validated spec to working code; the two gates every change must pass: e2e and performance
- Loop engineering: turn-based, goal-based, time-based and proactive loops (Anthropic’s “Getting started with loops”)
- Implement the spec in chunks; a validator subagent checks each one, then a /code-review loop runs until no issues remain, with a second review by another vendor’s model
- Build the verification gate for your stack (Lighthouse on a web app, a benchmark elsewhere) with a baseline diff and gated exit codes
- Breakout: one ambitious goal per room. Divide the work, pick a presenter, ship
- Introspection and observability: session logs, /insights, OTEL metrics
- Agent-native architecture, and proactive loops that run without you: log analysis, slow-query triage, payment matching
- Complete 2–5 GitHub issues in parallel, each agent in its own worktree
- A measurable improvement loop: speed up the test suite or improve your performance score, both gates staying green
Your company’s real codebase, the whole workflow, instructors on hand. The weeks participants rate highest.
- A configured agentic setup: CLAUDE.md, skills, hooks, gates
- A real feature shipped through the full workflow, with the spec, ADRs and C4 diagrams
- Final presentation and peer review
Curriculum: Agentic Engineering & Claude Code · field of study: information and communication technology (software development and programming) · volume 50 academic hours (37.5 hours): 42 contact, of which 16 classroom and 26 practical, and 8 independent study
Goal: participants use AI agents in software development as a repeatable method: clear specifications, guardrails, automated checks and parallel agents.
- sets up the Claude Code harness: CLAUDE.md, skills, and hooks that block dangerous commands
- writes a specification with acceptance criteria the agent can check itself
- builds a verification chain that the agent runs until its work passes (tests, verification and validation)
- automates recurring tasks with loops
- runs several agents in parallel in git worktrees and monitors their work
- documents architecture with C4 diagrams and ADRs
Recent Webinar
What 300 Developers Taught Us About Agentic Coding
The patterns, data and harness we teach, drawn from every cohort and hackathon we have run since Claude Code shipped in February 2025, across 20+ Estonian companies.
Built to Stay in Your Codebase
- A CLAUDE.md and architecture docs that make the agent yours
- Reusable skills that scope the work before the spec is written
- Verification gates the agent runs itself: tests, e2e, performance budgets
- A parallel-agent workflow: worktrees, loops, review discipline
Who Should Join
- 2+ years professional development experience
- Comfortable with at least one programming language
- Access to a codebase you can work with during the course
- Available for all 6 full-day sessions (9:00–15:30)
Participants who present a final project receive a Plepic Certificate of Completion: Agentic Engineering & Claude Code
Investment
Training Program
Your own codebase from day one, or our sample repo · 6 Fridays · 37.5 hours · Online via Google Meet
Our first four cohorts sold out. The next cohort starts on 15 January. You can still join it.
+VAT per developer
- For development teams across Europe
- Full 6-session program
- Graduation project on your own codebase
- Certificate of Completion included
Opens a Google Form, 2 minutes
reimbursed, up to €2,500 per employee
Töötukassa reimburses Estonian employers 80% of the training invoice including VAT, capped at €2,500 per employee over three years. If the employer has used up its de minimis aid limit, the rate is 50%. We prepare the application with you. Provider: Plepic koolitus, EHIS majandustegevusteade nr 260952. Apply before training starts →
Join the waitlistOpens a Google Form, 2 minutes
Every cohort we have run finished the full six weeks.
Trained engineers at HELMES · HOLM BANK · DATEL · BITWEB · AKTIVA · DELFI · MERIT · MINDWORKS · RETHINKERS
“The biggest value was a mindset shift: not using AI as a tool, but building architecture around it. An entirely new perspective on applying AI systematically.”
Kristina Krist, Mindworks Industries
“The content was very strong. The biggest value came when I started experimenting on my own project. That’s when real understanding and trust emerged.”
Erich Jagomägis, Bitweb
“It broadens your thinking and shows how to bring AI into your work processes. It gave me a clear direction for where to go next.”
Merilin Sellik, Rethinkers
Bring your whole team
Book a call with Kaido. He explains how Estonian state aid funds the training under the de minimis rules, then shapes a cohort around your team and your codebase.
Talk to KaidoOpens Google Calendar to pick a time
Your Instructors
Kaido co-authored "AI Coding for Beginners"
Start Your Claude Code Course
A 3× promise.3x Time Savings Guarantee: if you don't experience at least 3x time savings in typical development tasks after completing this training, we'll refund your contribution in full. Conditions apply.
Talk to KaidoOpens Google Calendar to pick a time