1:1 agentic coding workshop

Turn Claude Code into a reliable engineering workflow.

Stop losing time to "almost right" output. Learn a practical system - checkpoints, validation loops, and context hygiene - and apply it directly to your codebase.

10+ years software engineering Agent-first since 2024 Hands-on on your project In-house enablement Hamburg, Germany

Not a philosophy. A system.

This isn't "4 steps to better prompting". It's 50+ skills, sub-agents, and workflows that run an entire multi-client business from a terminal. Real numbers from the last 16 days:

358

messages to AI per day

65

projects in 16 days

50+

production agent skills

What the agent replaces

QA team → browser automation, code reviewer, code simplifier
Bookkeeper → Lexoffice, Clockify, monthly invoices, expense tracking
Assistant → Gmail, Calendar, Slack, task management
DevOps → Ansible, Hetzner, deployment pipelines
PM → Kanban triage, weekly planning, Linear integration
Researcher → web search, YouTube, X/Twitter, deep research

A real day (Feb 6)

→ Built an analytics platform from scratch, deployed it
→ Set up Tailscale private infrastructure with security review
→ Built a lead gen pipeline for a client (Apify → Supabase → Slack)
→ Drafted client emails in German, logged hours to Clockify
→ Reviewed code from 3 angles (security, architecture, simplicity)
→ Deleted 20+ old GitHub repos, built 3 new skills
13 projects · 157 context switches · 529 messages · one terminal
Voice-driven direction, terse execution. Long instructions for context and intent. Then: "done" "push" "test" Think out loud, then delegate.

What you'll learn

This is less "prompting tips" and more "engineering discipline for agentic coding". The goal: shipping faster with fewer regressions.

Validation over vibes

Tests, linters, self-checking loops - so you know it works before moving on.

Never get stuck

Checkpoints, clean restarts, and when to regenerate instead of patching forever.

Context hygiene

How to keep agents sharp over weeks: files, structure, and anti-entropy patterns.

Ownership (skills & tools)

Build small tools you control - so you're not blocked by vendors or brittle setups.

Workshop format (example)

We'll adapt to your situation - but the structure stays the same: understand → apply → ship.

~60 min
Mental models + system
Why agents fail, what "validation loops" mean, and how to design tasks that converge.
~3-4 h
Hands-on on your codebase
Set up the workflow and implement something that matters (feature, automation, integration).
~30 min
Next steps
Concrete plan, reusable templates, and follow-up so the system sticks.

Proof from production

Real metrics from client operations and personal workflow - not demos, not toy projects.

598 tickets auto-solved

In the last 60 days via support automation (Zendesk).

5,700+ agent messages

In 16 days across 5 clients, infra, finance, and content - from one terminal.

Equal time building & compounding

~50% building features, ~50% building skills. Every skill makes the next day faster.

About

Julian Kaiser

Julian Kaiser

Software developer (10+ years). Working almost exclusively with AI agents since 2024 - building tools, automations, and backend systems for consulting clients.

julian@kaiserlich.dev · kaiserlich.dev

FAQ

Do I need to be a developer?

No. Many participants are agency owners or operators. If you can describe the outcome you want, we can build the workflow together. For deeper code changes, basic technical comfort helps.

What do I need beforehand?

A laptop, access to the codebase you want to work on (or a concrete automation problem), and 4-6 hours of focus time. We'll handle the setup during the session.

Is this just slides?

No. The slides are the "map". The value is in applying it to your repo and leaving with something shipped (and a workflow you can repeat).

How does pricing work?

Pricing is on request (depends on scope: 1:1 vs team). Invoiced with a regular EU invoice.