3 AImigo S1E3 Materials: Why Is AI Getting Cheaper While Adoption Gets Harder? (Rubric #AI)

The materials for the third episode of 3 AImigo, released on August 28, are ready. Together with Evgeny Sergeev and Alexey Litvinov, I made our first joint news digest. Rather than moving through a feed of announcements, we assembled six connected stories around one shift: models and their use are getting cheaper, while building a working AI system from them is becoming harder.

We discussed:

• why higher individual productivity does not yet equal systemic ROI, and how McKinsey's data shifts the conversation from one engineer's speed to redesigning the entire delivery flow;
• why, in our view, scarcity is shifting into data, legacy-system integrations, and governance; the July SAP–Dremio deal and Reuters' early signals of demand for integrators provided background for this story;
• how better price-performance expands the range of economically viable AI use cases, and why total spending could rise even as each request gets cheaper; the comparison with the Jevons paradox remains a hypothesis, not a forecast;
• why enterprise agents need a control plane covering permissions, budgets, residency, retention, and audit: model capability matters only inside a governed system;
• how Cursor is assembling a compute → model → harness/evals → code hosting → deploy → telemetry → correction loop, and why part of this complete production feedback loop remains a plan rather than a ready autonomous system;
• what the OpenAI and Anthropic cyber-eval incidents teach us: weak isolation, an incorrect model of the environment, and real tools can combine into practical risk, but they do not prove a “conscious escape” by an agent.

All episode materials:

• Episode page: summary, timeline, fact cards, and sources
• Interactive deck
• Video: YouTube, VK Video
• Audio: Podster, Yandex Music, Apple Podcasts

If you see August's main shift differently, challenge us in the comments. And send us news worth checking for the next digest.

#AI #AI4SDLC #Agents #Engineering #Management #Security #Podcast