iSAQB AGENTA module

What is agent-based architectural work?

Generative AI, large language models and coding agents are fundamentally transforming software development. Today, designs, prototypes and production-ready code are being produced at a pace that increasingly challenges traditional development and review processes. This does not make architectural work any less important – on the contrary: it is becoming a crucial steering force.

At the same time, software architects can utilise agent-based tools themselves to gather architectural information, explore alternatives more quickly and prepare well-founded decisions.

In the iSAQB AGENTA training course, you will learn how to deploy AI specifically for your architectural work, how to prepare architectural knowledge in a format suitable for agents, and how to permanently align coding agents with your architectural goals using appropriate guardrails.

Key topics include context and harness engineering, AI-supported architectural decision-making, the extraction of architectural information from existing systems, as well as governance, security and traceability.

Why is agent-based software development transforming architecture work?

When code is produced faster than humans can review it, an effective architecture determines whether a viable system is created – or whether new technical debt and emergent structures are generated at an ever-increasing rate.

Coding agents work quickly and with increasing autonomy. However, without the right context and appropriate guidelines, they are unaware of your quality objectives, architectural principles or the reasons behind existing decisions. Agent-based software development therefore presents new challenges for architectural work:

• Preventing architectural drift at an early stage: changes occur faster than traditional review cycles can detect them.
• Making architectural knowledge accessible to agents: Decisions, rules and quality objectives must be understandable not only to humans but also to machines.
• Enforce architectural rules automatically: architectural tests, fitness functions and guardrails complement traditional documentation and reviews.
• Using AI for architectural decisions: Variants can be explored more quickly, hypotheses tested and decisions prepared on a more informed basis.
• Ensure traceability: AI-generated results must remain traceable for reviews, audits and analyses.
• Take security and compliance into account: prompt injection, data protection, licensing issues and the EU AI Act bring new requirements.
• Preserving human judgement: Even with increasing autonomy, responsibility and architectural assessment remain with humans.

With the iSAQB AGENTA training, you will learn to systematically tackle these challenges and embed architecture as an effective governing body within AI-supported development processes – whilst simultaneously using agents to improve your own architectural work.

iSAQB Advanced Level – AGENTA Certification

The iSAQB AGENTA training course is an accredited module of the iSAQB Advanced Level and equips you with in-depth expertise in the field of agent-supported architectural work.

As part of the training , you will earn 30 credit points, comprising 10 technical and 20 methodological points, which  can be counted towards the ‘Certified Professional for Software Architecture – Advanced Level (CPSA-A)’ certification .

Secure double the benefits now

Use the code ‘Team20’ to save 20 per cent on the same training for every participant from your company from the second person onwards – and if you register early, you’ll also benefit from our attractive early-bird price. Register together now and get more out of it!

Alternatively, we also offer bespoke in-house training courses for your organisation – at our premises, at yours or remotely. Please contact us directly for a personalised quote.

Credits: Technology 10 | Methodology 20 | Communication –

Maximum number of participants per session: 12 people