Skip to content
back to the archive page
#AI

3 AImigo S1E11 Materials: The AI-Native State

I've put together the materials from the eleventh episode of 3 AImigo, “The AI-Native State: What It Is and Where to Draw the Line,” released on October 23, 2026. Evgeny Sergeev, Alexey Litvinov, and I discussed whether the ideas behind an AI-native company can carry over to government. It's convenient when AI helps fill out a form. Things get more complicated when it decides whether to reject your application, especially if you have to appeal to another bot.

What we discussed

🔸 Whose outcome are we improving?
Digitizing each department separately still leaves citizens with a fragmented experience. We started with Evgeny's experience with electronic prescriptions and ended up asking where the analogy between a state and a company breaks down.

🔸 What do we need beyond an assistant subscription?
Preparing a document and carrying out a procedure independently are different tasks. I looked at the environment an agent needs: data, instructions, authority, verification of the result, and accountability for the consequences.

🔸 How do you reach a person who can correct an error?
Reversible decisions and a workable route to appeal became central themes. Evgeny described how an agent helped someone he knows get a bank to reconsider a rejection. The person followed the negotiations and corrected the assistant along the way. It's an individual story and tells us little about whether agents can already beat bureaucracy in general.

🔸 Who benefits from a complicated form?
Alexey used Hyrum's Law as an analogy: roles and dependencies grow around a cumbersome procedure. If only one person knows how to fill out a form correctly, simplifying it changes their position too. A good interface alone can't solve that.

🔸 How quickly might things change?
We debated staff initiative, management support, government data, and dependence on AI vendors. I expect digitization to continue gradually. Alexey thinks we should take the possibility of a major leap seriously. We didn't arrive at a shared forecast.

Episode materials
📌 Episode page with timestamps and long read on the AI-native state
🎬 Video: YouTube and VK Video
🎧 Audio: Podster and Apple Podcasts
📖 Written summary

Which government procedure would you entrust to an agent, and where would you insist on being able to speak to a person?

#AI #Agents #Management #Architecture #Podcast

Open video on YouTube