Workshops
Agentic Coding: Developing an app with AI – from idea to specification to finished app in 1 day | GitHub Copilot & SpecKit Workshop
Fast app development is crucial today: requirements change, functions grow, the codebase becomes complex. Traditional programming often reaches its limits – too much manual work, too little repeatability. With AI tools, AI agents and a clear process, apps can be developed more easily: from the idea to specification and design to the finished application on your platform. In the Agentic Coding Workshop, you will learn how to develop apps with AI – in a structured, secure and scalable way. We will show you how to use GitHub Copilot, Claude and a spec-driven framework (e.g. SpecKit) to translate requirements into code, professionalize your development workflow and accelerate your development.
Satisfied customers from SMEs and corporations
What the Developing an app with AI workshop offers you
Our workshop Developing an app with AI offers you clear benefits and practical results – beyond buzzwords and hype:
Your highlights at a glance:
- Idea to app in hours: From specification and design to a functioning mini-app with backend and frontend
- AI agent process: clear workflow for requirements, architecture, coding, reviews and releases
- Higher quality: structured specs, tests, guardrails and clean repo/artifact templates
- Use AI tools sensibly: Copilot, Claude and agent workflows with context instead of ad hoc prompts
Content and process - Developing an app with AI in modular steps
Each workshop is individually tailored to your goals, existing tools and previous knowledge – always with a focus on added value. Each module ends with a concrete result that you can use directly in your development.
Module 01
Developing an app with AI: Agentic Coding: Role, tools and first steps
- What does agentic coding mean? Change of perspective: AI agents as active development partners
We explain what is meant by agentic coding and how the perspective changes when AI agents are included as active partners in the development process. - Clear allocation of roles between developers and autonomous agents
It describes which tasks should be retained by humans and which tasks can be delegated to autonomous agents. - Overview of Copilot, Claude and Custom Agents in the development process
We present the tools mentioned, compare their typical areas of application and outline how they fit into existing workflows. - Basics of structured and targeted prompting
Basic principles for precise prompting are explained so that agents can deliver consistent and usable results.
- Setting up Copilot and/or Claude in Visual Studio Code
Practical tips for integrating the agents in VS Code and for basic configuration for daily use. - Setting up an AI-enabled project structure
Concrete suggestions for folder structure, artifacts and conventions that facilitate collaboration with agents. - Creating and configuring the first custom agents
Instructions for creating simple custom agents, including basic settings and role assignment. - Formulating, testing and optimizing structured prompts
Exercises for writing prompts, testing them and iteratively improving them until reliable results are achieved.
- Clearly defined target image for agentic coding in your own project context
You will receive a concrete idea of how agentic coding can work in your project. - Functional setup with ready-to-use AI agents
The end result is a ready-to-use setup with the first configured agents.
Module 02
Developing an app with AI: Context, structure and token efficiency
- Why high-quality context is crucial for AI results
We explain how relevant context improves the quality of answers and reduces misinterpretations. - Understanding tokens: limits, costs and
Performance effects
Basic relationships between token limits, cost factors and response behavior are explained. - Every Token Counts” principle – quality and precision in input
Focus on precision and relevance of input to achieve better and more efficient results. - Strategies for the targeted selection of relevant content per task
Methods for selecting and providing the context that is really necessary for a task.
- Optimization of the project structure for AI-supported development
Practical adjustments to the structure and processes in order to integrate agents efficiently. - Structuring code, documentation and specifications
Concrete recommendations for structuring code and accompanying documents so that they can serve as context. - Task-specific provision of context
Procedure for compiling and providing the exact context required for individual tasks. - Application of concrete techniques for token reduction
Simple techniques for shortening and condensing context without losing important information.
- AI-optimized project and documentation structure
A structured basis that facilitates collaboration with agents. - Ability to provide context efficiently, precisely and purposefully
You will learn how to deliver only what you need – efficiently and effectively.
Module 03
Developing apps with AI: Spec-driven development with SpecKit
- Principles and added value of Spec-Driven Development
We explain the basic idea behind specification-driven development and the benefits it brings. - From natural language to structured, implementable specifications
Ways in which free descriptions are transformed into clear, machine-usable specifications. - Cooperation between Copilot and SpecKit as a coordinated agent team
Illustration of how different agents can work together to convert specifications into code. - Potentials and limitations of automated code generation
Realistic assessment of what automated generation can achieve and where human control remains necessary.
- Formulation of features in precise natural language
Exercises on how to describe requirements in such a way that they are usable for SpecKit. - Creation of structured, AI-usable specifications
Step-by-step creation of specifications that can be processed automatically. - Automated code generation with SpecKit
Use SpecKit to generate output code based on the specifications. - Systematic review and iterative quality improvement
Procedure for review loops, tests and gradual refinement of the generated code.
- Established spec-to-code workflow
A functioning process from specification to initial code. - Measurably accelerated development through structured AI collaboration
Improved efficiency through clear role allocation and structured specifications.
Module 04
Developing an app with AI: Live hacking: developing a mini-app
- Architecture decisions in AI-supported development
Criteria that influence architecture decisions when agents are involved. - Effective distribution of tasks between humans and agents
Guidelines on which parts of the project are better handled by humans and which by agents. - Quality assurance, review strategies and control mechanisms
Methods for ensuring quality and incorporating control points into the development process.
- Development of a mini-app with frontend and backend
Joint implementation of a small project as a practical case for working with agents. - Combined use of Copilot and SpecKit in the live scenario
Coordination of the tools in the real-time workflow to control architecture, implementation and tests. - Management of architecture, code generation and reviews
Active moderation of the steps from design and generation through to review and integration. - Iterative testing, refactoring and optimization
Cyclical improvement of the code through tests, refactoring and targeted optimizations.
- Running mini-app as a reference project
A concrete, functioning result that serves as an example for other projects. - Proven expertise in the coordinated use of multiple AI agents
Acquired know-how for managing and controlling multiple agents in the development process.
Module 05
Developing an app with AI: Reflection, best practices and transfer
- Best practices for sustainable and scalable agentic coding
Summary of best practices for long-term, scalable use of agents. - Typical sources of error and systematic optimization approaches
Common pitfalls and pragmatic approaches to avoid or eliminate them. - Integration into existing development processes and teams
Information on how agentic coding can be embedded into existing processes and team structures.
- Structured reflection on own results
Guided reflection to evaluate own work and the results achieved. - Derive specific guidelines for your own team
Develop practicable rules and recommendations for teamwork with agents. - Adaptation of templates, prompts and workflows for everyday use
Practical adaptations to ensure that templates and workflows work in day-to-day operations.
- Concrete action plan for the transfer into practice
An actionable plan on how the findings can be transferred into the project. - Sustainably applicable spec-to-code workflow in a real development environment
A tested workflow that can be used and further developed in everyday life.
Who is the workshop Developing an app with AI suitable for?
- Software engineers with and without agentic coding experience Teams that
- want to integrate AI tools into their app development and delivery pipeline
- Tech leads who want to test and scale Spec-Driven Development for companies
Workshop formats and participation according to your requirements
You decide whether you want to start with a compact introduction or establish the complete end-to-end process for software development with AI.
- Suitable for: Teams and developers who want to implement software with AI tools (Copilot, Claude, SpecKit) and understand the agentic coding workflow.
- Example focal points:
- Basics of agentic coding, setup in VS Code, tool integration
- Context management, specification (Spec-Driven Development) and initial code generation
- Mini app with core functions (frontend/backend light), simple tests and reviews
- Typical results/deliverables:
- Executable setup, repo/artifact templates, simple spec-to-code pipeline
- Functional prototype (mini-app) with basic tests and documentation
- “Developing an app with AI” checklist for daily use
- Prerequisites:
- Basic knowledge of development/repos helpful, but not mandatory
- GitHub Copilot, Claude, Codex Abo
- Others depending on individual requirements
- Format & Duration:
- 1 day, online or in-house (DE/EN), up to 10 participants
- Suitable for:
- Teams with initial tool experience, tech leads and organizations looking to integrate, scale and secure Agentic Coding into their delivery pipeline.
- Focal points:
- Spec-to-Delivery End-to-End: Requirements → Architecture → Implementation → Tests → Release
- Quality and safety standards: review patterns, guardrails, CI/CD integration
- Roles, handovers, metrics; scaling via services/teams, governance
- Typical results/deliverables:
- Team-capable agentic coding workflow (documented), context/prompt libraries
- Validated app prototype with tests, release criteria and CI/CD hooks
- Short roadmap (next steps, KPIs) for team-wide introduction and scaling
- Prerequisites:
- Practice with repos/branding and basic understanding of modern development; initial use of tools an advantage
- Format & Duration:
- 2 days, Munich on site or remote/hybrid (DE/EN), up to 20 participants
What you can develop directly after the workshop App with AI
Immediately afterwards, you will have the process, templates and knowledge you need to reliably accelerate your app development with AI tools. An excerpt:
Your key takeaways:
- Setting up the Agentic coding workflow: Development environment, standards and team handoffs
- Applying Spec-to-Code: Systematically convert specifications into executable code
- Create a mini app: Program frontend and backend, integrate tests and reviews
- Ensure structure and quality: repo/artifact templates, guardrails, CI/CD connection
- Understanding the way we work: Differences between classic programming and AI-supported development
Developing an app with AI Workshop: group size, location & format, costs
To ensure that the effect, tempo and organization match, we coordinate the framework and depth with you.
- Group size: Up to 10 people
- Location & format: On site at your premises, remote or hybrid; on request in our office in Munich
- Costs:Depending on duration, scope and group size; on request
- Additional services: Optionally bookable - follow-ups, prototyping, implementation, scaling support
About Ventum
With over 20 years of consulting experience, we combine in-depth expertise in the introduction of digital innovations such as artificial intelligence with tried-and-tested methods.
01
Industry and size expertise
We know SMEs and enterprises - from lean pilots to structured implementation across several areas.
02
Effect before theory
Our workshop delivers tangible results: prioritized use cases, validated MVPs, roadmap and clear responsibilities.
03
Strategy connection
Role model fits your organization, governance and roadmap.
04
Designed to be scalable
We plan data, processes and systems from the outset so that solutions are viable from pilot to rollout.
05
Over 20 years of experience
Digital projects, data-driven decisions and effective change - best practices that have proven their worth.
06
Support until success
On request, we provide support with review, implementation, scaled introduction and sustainable optimization.
Our references and projects in AI and data
Competence in AI Security Assessment - Your expert
Request a non-binding appointment now
- Field-tested expertise: Over 20 years of experience in digital innovation and scalable app development for companies
- Individual workshops: Content and formats tailored to your goals, tools, codebase and platform
- Experienced coaches: daily use of AI tools and agents such as Copilot and Claude in real software setups
- Immediately applicable knowledge: Methods, best practices, templates and checklists for measurable success from day one
TISAX and ISO certification for the Munich office only
Your message
FAQ - Frequently asked questions about the Agentic Coding Workshop
No. We pick up your team where they are – from getting started with AI tools to expanding existing workflows. The depth and pace of the workshop are tailored to your prior knowledge and goals.
Ideally, yes. This way, results have an immediate effect. Alternatively, we use representative examples. We coordinate data access, security and necessary resources with you in advance.
Typical examples are VS Code, GitHub Copilot, Claude and SpecKit (Spec-Driven Development). We work tool-agnostically: We integrate your existing tools and platforms (e.g. GitHub, GitLab, Azure). Selection criteria are benefits, connectivity to your stack, security and scalability. On request, we evaluate supplementary agents/frameworks and set up a minimally viable setup.
Through binding standards: Spec and architecture reviews, test patterns, static analyses, dependency checks, security guidelines and defined release points (Definition of Done/Ready). Guardrails, checklists and CI/CD integration ensure that agent code enters your pipeline in a reproducible, auditable and secure manner.
Yes – on-site and remote. A hybrid approach is possible, but not optimal in our experience.
On request, we can support review, implementation and scaling: follow-up sessions, prototyping/feature expansion, CI/CD hardening, metrics/monitoring and organizational anchoring. The goal is a sustainable process from spec to release.
German or English. Mixed-language teams are possible. We provide documentation and templates in the desired language.
Ideally, there should be 5-10 participants per session.
None formal. Basic knowledge of development/repos is helpful. We tailor the depth and pace to your prior knowledge and set up a minimal setup for “Developing an app with AI” together in advance.