Written by the AgileFever editorial team. Program details were checked against the live program pages in October 2026. We teach the solution architect course mentioned below, so weigh that as a vendor view.
Ask architects what changed in their job over the last two years and most do not say “I use AI to write documents.” They say that every design review now includes an AI question: where does the model sit, what does it cost per thousand requests, what happens when someone pastes a prompt injection into a support form, and how does it talk to the mainframe. Those are architecture skills, and they are the ones that decide whether an AI design gets approved.
The six skills and why reviews turn on them
| Skill | The question a review board asks | What a good answer includes |
|---|---|---|
| Platform and vendor choice | Why this model provider and this cloud? | A comparison on data residency, cost, latency and lock-in, not a brand preference |
| Retrieval design (RAG) | How does the model get company data, and how do we know it is right? | Chunking and indexing choices, access control on retrieved content, an evaluation set |
| Agent design | Does this need an agent at all? | Where a fixed workflow is enough, and where autonomy earns its added failure modes |
| Cost modelling | What will this cost in production? | Tokens per request, expected traffic, caching, a monthly range and a limit that triggers an alert |
| AI security | What happens when someone attacks the prompt? | Prompt injection, data leakage between users, least-privilege tool access, logging |
| Legacy integration | How does it work with what we run today? | Where it plugs into existing APIs, queues and identity, and what fails safely |
Additional discounts are available. Contact us to know more.
A way to check your own gaps
Take a recent proposal that includes an AI component. For each row above, write one sentence on what your design says. The row where you have nothing to write is where to learn first. Most architects find cost modelling and security are thin, because early AI work was prototypes and nobody asked.
How the AgileFever course covers it
GenAI & Agentic AI for Solution Architects runs 16 hours over three consecutive days and has no coding. It has 10 modules: platform and vendor decision framework, architecture documents and RFP automation, RAG architecture, multi-agent architecture, LLMOps and production deployment, FinOps cost modelling, AI security architecture, legacy integration, governance and rollout, and a capstone. The capstone is an architecture package for a real use case at your own company, with the platform choice, cost model, security approach and integration plan.
The course FAQ says about 9 hours go to architecting AI systems and about 2 hours to using AI in your own work, with the rest used for the platform framework and capstone.
Listed hours from the program pages, October 2026
Prerequisites and price
You need architecture fundamentals, comfort with at least one cloud platform, and the 6-hour Python for AI pre-work so you can read code and talk to engineers. The course is $450 in the US. Pre-sales architects and consultants take it too, since the capstone often becomes the first draft of a real proposal.
When a different course fits better
| If you want to | Consider |
|---|---|
| Write RAG pipelines and agents in code yourself | GenAI and Agentic AI BootCamp (74 hours, $2,000) |
| Operate models in production | MLOps, LLMOps and Agentic AI Ops BootCamp (60 hours, $2,000) |
| Set AI strategy and budget rather than design systems | AI Transformation for Business Leaders (16 hours, $1,000) |
Sixteen hours is enough to make sound design decisions. It will not make you hands-on in every framework, so if your role needs both, start with the design course and add a bootcamp later.
