Background – Why Now
There’s an active global discussion about the design of AI agents.
In January 2025, Nvidia CEO Jensen Huang stated that “we are entering an era where AI no longer just produces text or images, but acts autonomously.”
At the same time, Andrej Karpathy, who introduced the concept of vibe coding, reminds us that humans should always verify the decisions made by AI. These two schools of thought define the frontier of the future — where organizations must decide when AI should act independently and when it should only assist.
This training helps your company take concrete steps toward the Software 3.0 era, where business logic is written in human language and implemented through agent-based solutions.
Which school of thought is right? Now is the perfect time to learn what Jensen and Andrej are actually talking about — in practice.
The Problem – The Era of Manual Information Management Is Over
Corporate information work is still largely manual and fragmented.
For example, ERP master data — such as customer, supplier, and product data — is often full of errors, duplicates, and missing information.
This leads to waste, inaccurate reporting, and frustrated employees.
Example from the data field:
An AI agent can now read, understand, and correct data autonomously.
It can validate, merge, and enrich supplier data without manual effort.
This is a tangible example of what the agents of the future will do — and exactly the kind of skills you’ll build in this training.
The Solution – Learn to Build Your Own AI Agents
In this training, you’ll build Agentic AI solutions that:
- Read and update data in ERP systems (via API interfaces)
- Retrieve and process data from file systems and databases
- Use web browsers to operate in SaaS systems like a real employee
- Generate and verify outputs through intuitive UI/UX workflows
We’ll use leading open-source libraries and modern LLM/LVM interfaces, enabling you to build your own AI agents around your own processes and data.
This isn’t theory — it’s fully hands-on implementation.
In just two days, you’ll build an agent that actually works: it reads data, executes tasks, and requests human verification through an elegant UI/UX.
Who Is This Training For
This training is ideal for:
- Consultants and architects designing AI roadmaps and business cases
- Leaders and decision-makers seeking to understand the business impact of agents
- Developers and data professionals who want Agentic AI superpowers
The program can also be tailored specifically for your company — using your environment and data.
Training Overview
Day 1 – The Agent Architect’s Core Toolkit
- Balancing Software 1.0, 2.0, and 3.0
- UI/UX design using TypeScript (Software 1.0)
- Optimizing AI memory as the main bottleneck (Software 2.0)
- “System prompting” for business logic (Software 3.0)
- NoSQL databases supporting AI generation (Software 1.0)
- ERP integrations using REST APIs and RPA tools (Software 1.0)
- Security, continuous improvement, DevOps, and test automation
Day 2 – Build Your Own Agent
- Browser-based agent design
- Hands-on project: from idea to working application
- Final presentations and “Best Agent” voting
Why This Trainer
M.Sc. (Tech.) Mikko Valtonen is one of Finland’s most respected trainers in AI and data solutions.
Mikko’s AI Agent and AI Assistant portfolio includes:
He has been:
- Named SAP Trainer of the Year
- Designer of the largest data warehouse architecture in the Nordics
- Creator of one of the most advanced chatbots in Finland
- Trainer of over 100 AI professionals
- Certified Azure expert — and father of eight children
Mikko’s training combines deep technical expertise, business insight, and an inspiring, hands-on approach.
Results and Impact
After completing this training, you will:
- Be able to build a functional AI agent using your own data
- Understand agent architecture and secure deployment
- Communicate about AI initiatives in business terms
- Identify the best use cases for agents across different organizations
We also offer AI agent development and implementation as a service.
For more information, contact Johannes Hovi — [email protected], +358 50 3670809, or via LinkedIn.
