Training

Training catalog: engineering and business tracks for teams working with AI

Every course has a code, a level, a duration in days and a defined audience, so you know what you are buying before you talk to us. Courses run on-site, remote or hybrid, in Polish or English, for one team or several. All of them are hands-on: participants work on their own code, their own processes and their own tooling wherever the format allows.

In short

23Minds runs a catalog of eight AI courses split into an engineering track and a business track. The engineering track covers AI-assisted development with coding assistants, reviewing AI-generated code, building and operating agentic workflows, secure coding with AI assistants, and LLM observability with cost control. The business track covers AI literacy for business teams, designing AI use cases that reach production, and AI governance with EU AI Act essentials. Courses run from half a day to three days, at foundation, practitioner or advanced level, on-site, remote or hybrid. All courses may be eligible for BUR/KFS co-funding subject to operator rules.

Tracks

Engineering track

ENG-101

AI-Assisted Development with Coding Assistants

Level: foundationDuration: 2 daysFormat: hybrid

For: Software engineers and tech leads who use or are about to use a coding assistant in daily work.

You leave with

Set up a coding assistant against a real repository, including project context files and conventions the assistant reads.
Break a task down so an assistant produces reviewable changes instead of large unverifiable ones.
Recognise the failure patterns specific to assistant output and know which prompt to fix versus which code to rewrite.
Apply the team rules for what an assistant may touch, what it may not, and what must be disclosed in a pull request.
Measure the effect on your own work with cycle time and rework, rather than by impression.

Prerequisites: Daily work in at least one programming language and comfort with Git.

Funding: Eligible for BUR/KFS co-funding subject to operator rules.

Ask about this course

ENG-201

Reviewing AI-Generated Code

Level: practitionerDuration: 1 dayFormat: remote

For: Reviewers, tech leads and QA engineers who approve changes written wholly or partly by AI.

You leave with

Identify the defect classes that AI-generated code produces most often, working on real examples.
Review a large machine-authored change without reading every line, by targeting the parts where the risk actually sits.
Judge whether a test suite protects the requirement or merely restates the implementation.
Configure an automated review tool so its findings are worth acting on, and define what blocks a merge.
Write a review policy for the team that survives a busy release week.

Prerequisites: Regular experience reviewing pull requests. ENG-101 helps but is not required.

Funding: Eligible for BUR/KFS co-funding subject to operator rules.

Ask about this course

ENG-301

Building and Operating Agentic Workflows

Level: advancedDuration: 3 daysFormat: on-site

For: Senior engineers and architects designing systems where a model calls tools and acts on business systems.

You leave with

Decide which parts of a process justify an agent and which should stay a deterministic script.
Design tool interfaces and permission scopes so a misbehaving agent cannot exceed its mandate.
Implement MCP-based integrations and treat the server as the privileged boundary it is.
Build stop conditions, budgets, retries and audit logging into the workflow from the start.
Take an agent from pilot to production with evaluation, monitoring and a rollback path in place.

Prerequisites: Practical experience integrating LLM APIs, plus solid backend and API design skills.

Funding: Eligible for BUR/KFS co-funding subject to operator rules.

Ask about this course

SEC-201

Secure Coding with AI Assistants

Level: practitionerDuration: 2 daysFormat: hybrid

For: Engineers and tech leads whose teams ship code written with AI coding assistants

You leave with

Spot the security flaws coding assistants introduce most often: unsafe defaults, missing authorization checks, secrets left in code, over-permissive dependencies
Review a prompt-building path in an application and find where user input reaches the model unfiltered
Set permission boundaries for an in-app agent so it cannot call what it was never meant to reach
Write review rules your team applies to AI-written pull requests, and wire the automatable checks into CI
Handle model output safely before it reaches a user, a database or another system

Prerequisites: Working knowledge of web application security fundamentals and hands-on experience with an LLM API.

Funding: Eligible for BUR/KFS co-funding subject to operator rules.

Ask about this course

OPS-201

LLM Observability, Evaluation and Cost Control

Level: practitionerDuration: 2 daysFormat: remote

For: Platform, SRE and backend engineers running LLM features that real users depend on.

You leave with

Instrument an LLM application end to end: traces, prompts, tool calls, tokens, latency and cost per request.
Build an evaluation pipeline from a curated set and run it on every prompt, model or retrieval change.
Detect quality drift with a stable baseline and alert on it before users report it.
Cut model spend with caching, routing and prompt compression, and prove the saving with numbers.
Define the dashboard and alert set that an on-call engineer can actually act on at three in the morning.

Prerequisites: Experience operating a production service and familiarity with your observability stack.

Funding: Eligible for BUR/KFS co-funding subject to operator rules.

Ask about this course

Tracks

Business track

BIZ-101

AI Literacy for Business Teams

Level: foundationDuration: 0.5 daysFormat: on-site

For: Office, operations, sales, marketing and administrative staff using AI tools in daily work. No technical background required.

You leave with

Explain in plain terms what a language model does, what it cannot do, and why it sounds equally confident either way.
Write a usable prompt for a real work task and improve it based on the result.
Recognise which company data must never go into an external tool, and what the alternative is.
Spot hallucinations and check output before it reaches a client or a decision.
Meet the AI literacy expectation for your role, with attendance documented.

Prerequisites: None.

Funding: Eligible for BUR/KFS co-funding subject to operator rules.

Ask about this course

BIZ-201

Designing AI Use Cases That Reach Production

Level: practitionerDuration: 1 dayFormat: hybrid

For: Product owners, process owners, analysts and managers responsible for deciding what gets built.

You leave with

Qualify a use case against volume, rule density, error tolerance and the cost of being wrong.
Estimate value before building, and define the metric that will show whether it worked.
Recognise the use cases that reliably fail, and say no to them with an argument rather than a feeling.
Write a brief an engineering team can act on, including data, constraints and acceptance criteria.
Plan a pilot with a defined scope, a defined end date and a decision to be made at the end of it.

Prerequisites: Ownership of a real process or product. No technical background required.

Funding: Eligible for BUR/KFS co-funding subject to operator rules.

Ask about this course

GOV-101

AI Governance and EU AI Act Essentials

Level: foundationDuration: 1 dayFormat: remote

For: Managers, compliance and risk staff, and technical leads who have to answer for how AI is used in the organisation.

You leave with

Classify your own AI systems against the risk tiers in the EU AI Act and document the reasoning.
Build an inventory of AI systems with owner, purpose, data flows and risk class.
Understand which obligations are engineering work and which are documentation, and who owns each.
Design human oversight that a person can actually exercise, rather than an approval button.
Draft a usage policy and a risk register that hold up in a supplier questionnaire.

Prerequisites: None. Legal background not required; this is the technical and organisational side, and we work with legal counsel rather than replacing it.

Funding: Eligible for BUR/KFS co-funding subject to operator rules.

Ask about this course

FAQ

Questions teams ask.

All of these courses may be eligible for co-funding through Polish instruments such as BUR and KFS, but eligibility, the level of co-funding and the paperwork are decided by the operator handling your application, not by us. We provide the course card, the scope and the documentation the operator needs. Confirm your own eligibility with the operator before you plan a budget around it.

Yes, and it is the format we recommend. Exercises on a real repository or a real process are worth considerably more than generic examples, and the output is usable on the Monday after. It requires a short scoping conversation beforehand and, where relevant, an NDA.

Foundation assumes no prior experience with the topic and builds the vocabulary and the first working habits. Practitioner assumes you already do the work and focuses on doing it well under real conditions. Advanced assumes you design systems others depend on and concentrates on architecture, failure modes and operations.

Hands-on courses work best between six and twelve participants, because everyone needs to be working rather than watching. AI literacy sessions run larger. Above roughly sixteen people we recommend splitting the group rather than lowering the amount of exercise time.

Yes. Each participant receives a certificate listing the course code, name, level, duration and date, which is what a co-funding operator or an internal AI literacy record needs. Our engagements run under an ISO 9001:2015 certified quality management system, so attendance and delivery records are kept as a documented process.

That is the usual case. A typical combination is BIZ-101 across the wider organisation, ENG-101 and ENG-201 for the engineering team, and GOV-101 for whoever has to answer questions about how AI is used. We put the sequence and the timing together with you during scoping.

Talk to an engineer.

Describe where you are with AI training. A reply within one business day.

Talk to an engineer

Szczecin - ul. Wawrzyniaka 6WWarszawaZielona GóraKraków