---
title: "When code gets cheap, software gets expensive | SamuylovAI"
description: "AI writes code faster than ever. Architecture, verification and operation still cost what they cost. The argument from Alpine Tech Forum in Davos, during WEF week."
url: https://www.samuylov.ai/stories/when-code-gets-cheap-software-gets-expensive
language: en
---

Article · January 19, 2026

# When code gets cheap, software gets expensive

AI has made code cheap. It has not made software cheap: someone still has to design it, check it, secure it and run it.

That was the thread of _GenAI Is Eating Your Software_, the session that opened [Alpine Tech Forum](https://www.alpinetechforum.com/) in Davos on 19 January 2026, the first day of the World Economic Forum. Denis Samuylov started it with a short keynote, _The Current State of GenAI in the Enterprise_, given together with Eric Anderegg, and introduced the discussion that followed.

![Opening keynote slide questioning whether generative AI is failing.](https://www.samuylov.ai/_img/stories/when-code-gets-cheap-software-gets-expensive/keynote.b3af7d9b.w1600.webp)

Alpine Tech Forum, Davos, 19 January 2026.

## Where AI pays off

The keynote ranked five categories of enterprise use cases, from lowest to highest value:

- Marketing and sales content
- Knowledge work and internal productivity
- Customer service and operations
- Back-office and core processes
- Software engineering and IT workflows

![Eric Anderegg presenting a slide titled Marketing & Sales Content, marked 1/5.](https://www.samuylov.ai/assets/1526b315/stories/when-code-gets-cheap-software-gets-expensive/five-categories.webp)

The five categories, ranked from lowest to highest value.

In the projects we see, software engineering and core processes return the most: work with clear inputs, a clear owner and results you can check. It is the same pattern as [the boring use cases that work](https://www.samuylov.ai/stories/the-boring-ai-use-cases-are-the-ones-that-work).

## Cheap code, expensive software

The panel that followed, with Hugo Venturini ([SkipLabs](https://skiplabs.io/)), Alexander Taboriskiy ([Mentiora AI](https://mentiora.ai/)), Dr. Petar Tsankov ([LatticeFlow AI](https://latticeflow.ai/)) and Frederik Bonde ([paterhn.ai](https://paterhn.ai/)), took that ranking to its consequence. Code is cheap now. Software isn’t.

More generated code means more code to review. Static analysis, formal verification and model checking become the invisible backbone of an AI-assisted pipeline. Governance decides who is responsible for what reaches production.

## What it means for your project

Budget for verification as seriously as for generation. If AI doubles what your team can produce, check that review, testing and operations can keep up, and decide who owns the result before anything consequential ships.

Our [ticket-change copilot](https://www.samuylov.ai/stories/ticket-change-copilot-for-a-large-travel-group) is built this way: every new version is checked automatically against the cases the client’s experts listed before it ships.

Planning AI-assisted development? We help you size the review, testing and operating work alongside the build. See [advisory & implementation](https://www.samuylov.ai/consulting), or [Start a conversation](https://www.samuylov.ai/contact).

[Join the discussion on LinkedIn](https://www.linkedin.com/posts/denis-samuylov%5Fwef26-davos-share-7418987597416288256-QGsI/)

[Download as PDF](https://www.samuylov.ai/stories/when-code-gets-cheap-software-gets-expensive/pdf)

## Have a project in mind?

Tell us what you are building or improving. We'll help you choose a practical next step.

[Book an intro call](https://www.samuylov.ai/contact?book=1&topic=other)

## More stories

[All stories](https://www.samuylov.ai/stories)

[Document classification for a global procurement services provider](https://www.samuylov.ai/stories/document-classification-for-a-global-procurement-provider) – Case study · Procurement Services · February 2026

[The 95 percent is a measurement problem](https://www.samuylov.ai/stories/the-95-percent-is-a-measurement-problem) – Article · November 2025
