Give the agent a useful brief
What context to give, what to leave out, and why a plan before any file changes saves you the most time.
FREE LIVE WEBINAR FOR BACKEND ENGINEERS · 17 NOV · 17:00 CET
Ship AI-written code
you’d put your name on.
Coding agents write plausible code in seconds. Checking it is now the slow part, and you’re the one who signs off.
Watch a senior engineer take one real backend bug from issue to review-ready PR, live, including the moment the agent gets it wrong.
Tue 17 Nov · 17:00–18:00 CET / Free · Live online · In English
✧ ISSUE → CONTEXT → DIFF → REVIEWAn illustrated workflow
THE TASK ↗
Fix a bug.
Understand the code.
An issue, a stack trace, tests and repo rules. Reproduce the bug first.
≡ bug-report.md
✧AI + your contextYou stay in control
THE RESULT ✓
A change
you can defend
− return cache[key]
+ return cache.get(key, fallback)- A small diff you can explain
- A test that catches the original bug
- Your own review checklist
✓ The final review is yours
- 01 Context
- 02 Change
- 03 Tests
- 04 Review
Illustration · not a live AI model
Real repo, real terminalThe agent’s mistake, caught on screenA checklist you can use on Monday
FREE LIVE WEBINAR · TUESDAY 17 NOVEMBER 2026
60 minutes. One real bug.
Every step on screen.
No slides about the future of AI. Alexander opens a backend repository and works one bug end to end with a coding agent: the brief, the plan, the diff, the tests and the review. You see where the agent helps, where it quietly goes wrong, and the checks that catch it before a reviewer does.
Keep the diff small
Task, plan, change, check. How to stop the agent from “improving” code you didn’t ask it to touch.
Catch the convincing mistake
Green tests aren’t proof. See a test that passes on the broken code, and the quick check that exposes it.
Most of the hour is live work, then open Q&A. You leave with the repo-instruction template and PR review checklist used in the session. At the end, a short intro to the paid cohort; no obligation.
SOUND FAMILIAR?
AI made writing code cheap.
Checking it is now the job.
The diff looks right.
You’re not sure it is.
The agent writes a convincing fix in seconds. Then you spend longer checking it than you would have spent writing it, and your name is on the PR.
Everyone’s figuring
it out alone.
Your team uses AI five different ways. Nobody has time to look at how you work with it, so you don’t know what you’re missing.
Demo tricks don’t
survive your codebase.
Tips from videos work on fresh toy apps. Your service has history, constraints and tests. You need a workflow that respects them.
For working backend engineers who use Git and tests and have already tried a coding agent (Claude Code, Cursor, Copilot, Codex or similar). Bring the tool you use; the workflow is about context, review and tests, not a specific product.
AFTER THE WEBINAR · THE FOUR-WEEK COHORT · STARTS 27 NOVEMBER
Then do it on your work,
with a senior engineer reviewing.
Watching a demo isn’t the same as doing it. In the cohort you run the workflow on your own backend tasks each week and submit the result. Alexander reviews every stage in writing, down to specific lines of your diff, and tells you what to change and how to check it.
27 Nov – 24 DecFour weeks · 2026
4 written reviewsOf your brief, PR, tests and results
€1,000For the full programme
VIDEO COURSES
Generic demos on fresh projects. Nobody looks at your work.
FIGURING IT OUT AT WORK
Habits formed by trial and error. Colleagues are busy figuring it out too.
THIS COHORT
Your task, your diff, your tests. A senior engineer’s written review, four times.
WEEK 01Brief the task. Set the baseline.
Choose a repeatable backend task. Record how you handle it today, then prepare repo context, constraints and a plan for the coding agent. Alexander reviews your task definition and measurement approach.
You keep: a task brief, baseline and repo instructions · Personal review 1 of 4WEEK 02Build a diff you can explain.
Reproduce the problem and use AI to prepare a small PR. Alexander reviews your diff, calls out weak assumptions and unnecessary changes, and explains how to improve it.
You keep: your PR, line-level feedback and revisions · Personal review 2 of 4WEEK 03Prove the tests catch the bug.
Check that your regression test fails on the original bug and passes on the fix. Work through edge cases and review the agent’s assumptions. Alexander reviews your tests and checking process.
You keep: verified regression tests and a review checklist · Personal review 3 of 4WEEK 04Repeat it. Decide what earns its place.
Run the workflow on a comparable task. Track active time, rework and model costs. Alexander reviews the results and helps you decide what to keep, change or do yourself.
You keep: a second PR, your workflow handbook and a cost comparison · Personal review 4 of 4THE DIFFERENCE IS IN YOUR REPOSITORY
Your code. Alexander’s review.
Submit one stage each week: the brief, PR, tests and final report. Alexander personally writes feedback on what to change, why it matters and how to check it. The PR review refers to your actual diff and specific lines of code.
You learn alongside a cohort with the same start and weekly milestones. The work and written reviews happen online, so engineers in different countries can take part. Submission dates and review turnaround are confirmed before payment.
NOVEMBER COHORTFour weeks.
Four weeks.
A workflow that’s yours.
Starts 27 November · In English · Four personal reviews
€1,000Full four-week programme
Start with the free webinar Enrolment happens after the webinar, once we’ve checked the cohort fits you. Nothing is charged on this page. Your company can pay from its learning budget; an invoice is available. AI tool subscriptions and API usage are not included.
THE PRACTICAL DETAILS
Before you join.
Who is this for?
Backend engineers who already use Git, run tests and have tried AI for coding. This is practical mentoring on AI-assisted software development. It assumes you can read and change code in your chosen language.
Is the webinar just a sales pitch?
No. Most of the hour is live work on a real bug, followed by open Q&A. The cohort gets a short introduction at the end. If you never buy anything, you still leave with the template and checklist from the session.
I already use AI. What will be different?
The feedback is on your work: your task definition, diff, tests and results. You build a process you can repeat and get personal comments at four stages. Start with the webinar to judge the depth for yourself.
When are the webinar and cohort?
The free webinar is Tuesday, November 17, 2026, 17:00–18:00 CET (Germany time; 16:00–17:00 UTC). The four-week cohort begins Friday, November 27 and runs through December 24, 2026. The cohort uses weekly submissions and written feedback; deadlines and review turnaround are agreed before payment.
Can I join from another country?
Yes. The webinar and programme are in English, online. The webinar is live at 17:00 CET; your local time is shown in the registration section. Cohort submissions and written reviews let you work around your own schedule.
Can my employer pay?
Yes. Many engineers use their company learning budget. Alexander can invoice your employer; ask for the details after the webinar or email s@perederey.com.
Can I use my work repository?
Start on a public or teaching repository. Use company code only with permission and tools allowed by your employer. Keep secrets and confidential data out of submissions.
Will I code faster?
We measure active work, rework and model costs, and account for differences between tasks. You leave with your own evidence about where AI helps. No particular speed-up is guaranteed.
Does the webinar commit me to buying?
No. It is free. At the end Alexander introduces the €1,000 cohort. You choose whether to discuss enrollment afterward; registration here does not take payment or enroll you in the paid programme.
What happens to my registration details?
Alexander Perederey uses your name and email to manage this webinar, send joining details and follow up about the November cohort. Registration details, your contact permission and campaign tags from the link are stored in a private Cloudflare D1 database in the EU. They are not published or added to a general newsletter. This page uses no advertising cookies or third-party analytics. To cancel, withdraw permission or request deletion, email s@perederey.com.
YOUR NEXT STEP · FREE LIVE WEBINAR
Watch the agent
get it wrong.
Learn how to catch it.
One backend bug, live, start to finish.
Bring your questions: the last part is open Q&A.
Tuesday, 17 November 202617:00–18:00 CET · Germany time
Live onlineIn EnglishFreeTemplate + checklist included