Hypeless AI: Looking Behind the AI Hype Curtain

Why we started Hypeless AI

If you follow AI news, you have probably noticed that the Internet is full of shallow content about AI. Every new model release is a revolution that completely changes the way we work. AGI will be here any minute. And by the time you finish reading this sentence, there will be another thread explaining why everything you learned last week is already obsolete.

What bothers us most is not the enthusiasm. It is the lack of evidence. People who were completely unknown just a few months ago now confidently claim that this model or that harness is the best. They rarely show how they measured it, and almost never back it up with any evaluation. A nice demo on a greenfield to-do app is not proof that a tool will help you in a ten-year-old enterprise codebase.

At RIGANTI, we use AI every day. We build custom software for our customers, and we also build Mr. RIGANTI, an AI coding agent for Azure DevOps. So we have a pretty good idea of where AI genuinely helps, where it saves hours, and where it confidently produces nonsense that someone has to review and throw away.

We wanted to look under the hype and the marketing curtain, and talk honestly about the real benefits AI brings to us, developers. Not in theory, but on real code, with real mistakes, and with real costs. That is how Hypeless AI was born.

What Hypeless AI is

Hypeless AI is a series of online live-streamed events by Update Conference. The motto says it all: Real code. Real experience. No hype.

The series is hosted by Tomáš Herceg, our CEO. Every month, he invites a guest who actually works with AI in software development: a Microsoft employee, a Microsoft MVP, a conference speaker, or someone who takes part in building AI models or agent harnesses.

Each episode is a live stream of about four hours on the Update Now platform, and it follows a simple structure:

  • Two demo blocks where the guest shows their actual workflow, tools and results, not polished slides.
  • A live Q&A where Tomáš reads your questions aloud and the guest answers them.
  • A recording available after the event, so you can watch it even if you miss the live show.

The guests are free to show what didn't work, too. Honestly, those parts are often the most useful ones.

The episodes so far

#1: Live-Coding a Feature with Tomáš Herceg (June 25, 2026)

For the first episode, Tomáš put himself in the guest chair. He showed how he builds features for Mr. RIGANTI with AI: how to prepare a project for the AI age with instructions, skills and tools, which tasks he gives to the GitHub Copilot agent locally in Visual Studio or VS Code, and which ones he delegates to run in the background on GitHub or Azure DevOps.

The main part was live coding a real feature. The agent spent about 30 minutes and roughly $23 in tokens on a change touching around 40 files. But as Tomáš stressed, it's not vibe coding: the key planning decisions came from his knowledge of the project, and the result still needed proper review and manual end-to-end testing. Episode page · Free preview

#2: Stop Watching AI Demos with Gerald Versluis (August 19, 2026)

Gerald Versluis is a software engineer at Microsoft working on .NET MAUI. He walked us through how he uses AI in his daily work: triaging issues, reviewing pull requests, writing and refactoring code, exploring unfamiliar APIs, and debugging.

Just as important, he explained how he decides when AI helps and when it doesn't. In the second part, he showed MAUI DevFlow, where an agent works more autonomously to build an app. The episode was aimed at developers who feel they can't keep up with AI and want practical patterns they can start using the next day. Episode page · Free preview

#3: Build. Break. Defend. Repeat: Automating AppSec for .NET with AI (September 24, 2026)

Niels Tanis, a Microsoft MVP and software security engineer at Tidalis with a background in pentesting and static analysis, focused on security. In the Build & Break part, he covered the risks of AI-generated code, the attack surface of LLM features in your apps, and LLM-driven pentesting.

In Defend & Repeat, he built a staged review harness on a real .NET application: threat modelling from code, vulnerability scanning, triage in a fresh context, and patching. In the demonstrated run, 15 candidate findings went through triage, and only six of them survived. That is exactly the kind of number you won't see in a marketing video. He also showed how an overly broad tool can turn an agent into a security problem, and why controls must be enforced outside the prompt. Episode page · Free preview

Coming next — #4: From Copilot to Collaboration (November 26, 2026)

Our next guest is Johnny Hooyberghs, a .NET architecture and backend consultant at Involved and a Microsoft MVP since 2020. He will start with a debugging story using GitHub Copilot and Claude Code, then show how to build an AI agent into an application with Microsoft Foundry, Semantic Kernel and the Microsoft Agent Framework. As always, the episode ends with a live Q&A.

What we've learned so far

Three episodes in, one pattern keeps repeating. AI is genuinely useful, but the value comes from the people and the process around it, not from the latest model name. The guests who get the most out of AI are the ones who know their codebase, break work into well-defined tasks, measure the results, and review everything that comes out.

None of them claimed that AI replaces developers. All of them showed that a developer who knows how to work with AI, and knows when not to, is a lot more productive.

Join us

If you are tired of hype and want to see how experienced developers actually use AI, join us for the next episode on November 26, 2026. Tickets are €49 for an individual or €139 for a team of up to four people, and you get access to the recording even if you can't watch live.

You can also browse the past episodes and watch the free previews, or subscribe to the newsletter at hypeless-ai.net so you don't miss the next guest.

And if you'd like to try AI coding agents in your own Azure DevOps projects, take a look at Mr. RIGANTI. We use it every day, and yes, we talk about it honestly on Hypeless AI, too.

Hypeless AI

Need experts for your
next software project?

Request a free consulting