Essentials · Deep Dive · DACH / CEE · 2026 cohorts open

Stop prompting
Claude.
Start operating it.

A hands-on workshop that turns Claude from a toy your devs tried once into a daily multiplier across your engineering organization. Built for teams of 5–80, run on-site or remote.

Format
4h Essentials
or 2-day Deep Dive
Where
On-site
or remote, EU hours
Built for
Eng teams
5–80 devs
claude-code · ~/repos/payments-api · main
add rate limiting to the /charge endpoint, follow our existing middleware pattern, redis only, no new deps context: CLAUDE.md · src/middleware/* · 4 files loaded   ⏵ entering plan mode   Readsrc/middleware/auth.ts Grep"redis" → 3 matches in src/lib/ Readsrc/lib/redis.ts, tests/middleware.test.ts   Plan: 1. rateLimit.ts in src/middleware/, mirrors auth.ts shape 2. token-bucket via existing redisClient 3. wire into /charge in src/routes/payments.ts 4. tests in tests/middleware/ — mirror auth tests   ✓ approve plan? (y/n) y applying 4 edits · 14 lines · 1 new file   ✓ pnpm test — 47 passed → ready for diff review
Hallucinated APIs
Silent drift
Loop-and-spin
Context rot
Premature victory
One-shot attempt
Context anxiety
Hallucinated APIs
Silent drift
Loop-and-spin
Context rot
Premature victory
One-shot attempt
Context anxiety
01The shift

Same ticket. Same tool. Different week.

The Tourist

Alex

Surface
Web UI · copy-paste in, copy-paste out
Tooling
None. Fights the tab manually.
Context
Whatever fits in a chat box.
Review rounds
3
Merged
Friday afternoon
"It mostly works, I just have to keep correcting it."
vs
The Operator

Sam

Surface
Claude Code in repo. CLAUDE.md + skills.
Tooling
Plan Mode → Auto-accept → diff review
Context
Files, conventions, constraints, tests.
Review rounds
1
Merged
Tuesday before lunch
"It scoped it, planned it, ran the tests, and asked me to look."

The gap is no longer about who uses Claude — almost everyone has. It’s about who knows how to wield it.

Two senior devs. Same ticket. One uses Claude as a chatbot. The other uses it as a tool. The throughput delta is not subtle — and the numbers below are why.
3–5×faster
Throughput delta between an operator and a tourist on the same ticket.
28% → 59%
Anthropic’s own engineering work done with Claude (n=132, 200k transcripts).
+67%
Pull-request throughput on teams with consistent Claude adoption.
9.8 → 21.2
Avg max consecutive tool calls per Claude Code transcript, Feb → Aug 2025.
02What good looks like

Four skills separate a fluent operator from a tourist.

01 / Plan

Plan

Scope work so Claude can execute it. Break it down before it breaks down on you.

02 / Context

Context

Load the right files, docs, and constraints. Context is the new prompt.

03 / Harness

Harness

Give Claude tools and guardrails — agents, skills, and a project structure.

04 / Review

Review

Treat output as a draft, not a verdict. Inspect, correct, steer.

03The agenda

Four hours. Six blocks. Every concept lands on a live demo.

Essentials — 4h teaching budget
~3h45m teaching · 1h lunch · 25m close
0:00
00
Opening & calibration
Story + 6-question poll · the tourist vs. operator frame · the day's contract
0:15
01
Mental model
Slides + "predict what Claude does" · names the 7 failure modes you'll use all day
0:45
02
Fluent prompting & context
Live demo of context starvation · CLAUDE.md anatomy · 15-min hands-on exercise
1:30
Lunch
The single longest break of the day · where 90% of post-workshop questions get answered.
2:30
03
Pitfalls & recovery
Story #2 + 4 micro-demos · audience-as-reviewer exercise
3:15
04
Real workflows
3 workflows · bring-your-own-ticket clinic · skills, plugins, subagents tour
4:00
05–06
Team standards & close
4-week roll-out plan · the three things to do tomorrow · open Q&A
The first three blocks are public. The full second half is shared on the discovery call.
Request the full agenda PDF →
04Do the math, just once

A trained team recovers the seat cost by Tuesday morning.

A Claude seat costs under 1% of a senior dev’s loaded monthly cost.

Trained teams typically see a 15–40% productivity lift. The question isn’t whether the seat pays back — it’s how fast.

Assumes €11k/mo loaded cost on a 160-hour work month. The rest of the month is upside.

“By Tuesday morning.”
Senior dev / month
€11,000
Claude seat / month
€100
Hours/month to break even
~1.5h
→ Net upside
~14–39%
05Two packages

Pick the one that matches where your team is today.

Half-day

Essentials

4 hours · on-site or remote · up to 25 devs

For teams getting started. By the end, every dev in the room can scope, prompt, and steer Claude for real engineering work.

  • The 4-skills frame: Plan · Context · Harness · Review
  • Live demos of every concept — not slides about them
  • 15-min hands-on exercise on a pre-baked sample repo
  • Take-home: scope-and-context template, mode cheat sheet, red-flag checklist
Best for: teams <3 mo in, or with inconsistent usage across devs.
0630 days after

What changes when individual fluency becomes team capability.

i

Faster shipping

Features land in days, not sprints — especially on routine or boilerplate-heavy work.

ii

Smaller review load

PRs arrive with tests, docs, and a self-review already attached.

iii

Less tribal knowledge

Conventions become skills; skills travel with the codebase.

iv

More senior time for real problems

Less shepherding juniors through repetitive work. More architecture, fewer tickets.

07Honest scoping

Who this is for — and who it isn't.

Good fit

  • Eng teams of 5–80 who've tried Claude unevenly
  • Orgs that want a consistent baseline across devs
  • Teams willing to change how they work, not just buy a tool
  • Leaders who want honest ROI, not vibes

Probably not

  • Teams looking for a one-hour magic demo
  • Orgs that want to outsource judgment to Claude entirely
  • Regulated contexts without a path to external tools
  • Leaders expecting AI to replace, rather than amplify, engineers
08Who's running it

Built and taught by one person who's actually shipped with this stuff.

portrait · placeholder

Kamil [Surname]

Engineer · Trainer · Master Claude

I've spent the last two years using Claude every working day — shipping production systems, debugging with it, refactoring at scale, and watching it fail in interesting ways. The workshop is the distilled set of habits that survived contact with reality.

I run sessions for engineering teams across DACH, Poland, and the broader CEE region. Sessions can be in English, Polish, or German.

15+
years shipping
production code
20+
teams trained
since 2024
3
languages
EN · PL · DE
Let's talk

Book a 30-min discovery call.

Tell me about your team — size, stack, where they are with Claude today. I'll tell you whether the Essentials or Deep Dive fits, what to expect 30 days after, and where this won't be the right move.

Book a call →

Tweaks