Using the Pareto frontier to compare model quality and task costs rather than token prices alone. Open weights, serving economics, model size, and illustrative margin calculations explain why openness alone doesn't produce hundredfold savings.
Open-weight control and flexibility through a trusted provider's API. Nemotron measurements on an A100 show the economics of concurrency and utilization; in my view, owning a server makes no sense for the vast majority of use cases.
The road from prompt engineering through context, agents and spec-driven development to closed loops - and why an executable definition of a good result is a company's most valuable asset.
Practical techniques for reducing token consumption in agentic coding without losing result quality.
A reflection on how agents change the unit of work, move interaction from specification toward intent, and create a new economy of human and token capital.
A practical demo of how Foundry Toolbox Tool Search reduces input tokens through progressive disclosure, cold search, and warm auto-pin.
The search index loads when you start typing.
Copilot, skills, context, feedback, and practical work with agents in software development.
How user interfaces, pricing, consumption, and business models change in AI-first products.
Practical personal workflows for using AI to accelerate learning, reading, and work with information.
Retrieval, vector search, knowledge graphs, and strategies for getting the right data into AI answers.
Concrete patterns for using AI over text, data, agents, and operational telemetry.
This page is static; as the archive grows, older items are progressively revealed in batches.
