Skip to main content
Developer training

Developer training – AI-assisted development that actually speeds the team up

Your developers are most likely already using AI tools. The question is whether those tools make them faster, or quietly add code to the repository that nobody really understands at review time. We teach Claude Code, Cursor and GitHub Copilot on your own repository: how to set up context, how to plan before implementing, how to read a diff you did not write, and what not to merge. Six modules of 2 to 4 hours each combine into a half day, a full day, or a 50-hour programme.

Training formats

Custom workshop

Duration
4 to 8 hours
Capacity
up to 12 participants
Funding
Paid by the company. We set the price by calculation after a free consultation.

Developer training, 50-hour programme

Duration
50 hours, spread across several weeks
Capacity
up to 15 participants per run
Funding
Funded from the Czech Labour Office (ÚP ČR) subsidy. The training contribution of 260.73 CZK per person-hour goes to the provider as the course fee, the wage contribution of 217.46 CZK per person-hour goes to the employer as wage compensation. The programme is delivered by PECKA.
PROBLEM

The team uses AI, but nobody knows what it does to the codebase

Copilot autocompletes, the agent generates, pull requests arrive faster than before. But writing speed is not delivery speed. The share of code nobody really understands at review time grows, tests get written after the implementation, and nowhere is it written down what may go into a prompt and what may not. The usual reaction is to ban the tools, after which the team keeps using them anyway, just quietly and without rules.

SOLUTION

We teach the team to direct the agent, not to trust it

We train on your repository, not on a to-do app from a tutorial. Developers set up context (CLAUDE.md, AGENTS.md), learn to plan before implementing, to write the test before the agent writes the code, and to reject output that does not belong in production. We add rules for security, secrets and licensing of generated code, and give an honest read on where AI makes sense in a product and where it only makes operations more expensive. Six modules of 2 to 4 hours make up a half day, a full day, or a 50-hour programme eligible for the Czech employee training subsidy.

BENEFITS

What you get

Training on your repository

We do not work on a practice project. Bring your own code, open tickets and a real pull request. The team leaves with context configured, its own working practices, and one concrete change finished during the session.

Six modules, programme per team

AI-assisted development, context, quality and review, security, modern stack, AI in the product. Each module runs 2 to 4 hours. A junior team starts somewhere different than a team that has been tuning agents for a month.

Review, not just generation

The hardest part of working with an agent starts once it finishes generating. We cover TDD with an agent, how to read a diff you did not write, and which rules belong in code review so that nothing slips through.

Security and governance for dev teams

What must never go into a prompt, how to handle secrets, where licensing of generated code stands, and when a self-hosted model is worth it. You leave with rules you can post for the team, not with a general warning.

A stack AI can work with

Next.js 15, strict TypeScript, Drizzle. Opinionated tools with strong types raise an agent's success rate, because the compiler catches the mistake before a human does. We explain why and what it means for your architecture.

50 hours and a training subsidy

The modules stack into 50 hours of instruction, the minimum scope of the Czech subsidy programme for employee training. A single run takes at most 15 participants. We prepare the syllabus, the schedule and the certificates of completion.

FAQ

Frequently asked questions

For engineering teams that already use AI tools or want to adopt them properly. We handle juniors and senior teams alike; only the starting point differs, sometimes at the basics of working with an agent, sometimes straight at review, security and shipping AI in a product. We agree on the level and on the repository we will work with before the session.

Six modules of 2 to 4 hours each. 1) AI-assisted development in practice: Claude Code, Cursor and GitHub Copilot, where they speed things up and where they quietly do damage. 2) Context beats prompting: CLAUDE.md and AGENTS.md, planning before implementation, why agents fail on large repositories. 3) Quality and review: TDD with an agent, reviewing AI output, what not to merge. 4) Security and governance for dev teams: what must never go into a prompt, secrets, licensing of generated code, self-hosted models. 5) A modern stack AI can work with: Next.js 15, strict TypeScript, Drizzle. 6) AI in the product: APIs, RAG, agents, real costs and token limits.

Under the Czech subsidy programme for employee training, yes. The programme requires at least 50 hours of instruction and allows at most 15 participants per run, so we combine and extend the modules into a programme that meets those conditions. That does not make the training free of charge: the subsidy covers part of the cost, the company files the application, and we supply the syllabus, the schedule and the certificates of completion. Conditions and a worked example are on the employee training subsidy page.

The base unit is a module of 2 to 4 hours. The most common bookings are a half day (two modules) or a full day (three to four modules). For the complete programme of all six modules we recommend splitting it across two days one to two weeks apart, so the team can try the practices on its own work in between. The subsidised variant is extended with practice and consulting to reach 50 hours.

Both. On site we come to you, or we meet in Prague. Online we run the training over Google Meet or Microsoft Teams with screen sharing and pair programming. For developer training the online format works well, because everyone works in their own environment with their own setup.

Claude Code, Cursor and GitHub Copilot, plus the agent modes in VS Code and JetBrains. If your company has standardised on a specific tool or has licensing constraints, we build the training around it. Tools change fast, so we mostly teach principles: working with context, planning, review and security. Those survive a change of editor.
LET'S GET STARTED

Interested in this service?

Book a free, no-obligation consultation. We will discuss your needs and design a tailored solution.

Webforte Technologies s.r.o. · Reg. No. 23364343 · Prague · we reply within 1 business day.

Free consultation