Insights
Notes from the work.
Practical writing on building enterprise AI that reaches production.
GPT-Rosalind starts billing today. A specialist model is a new production tier, not just a better answer
From October 5, OpenAI charges for GPT-Rosalind, its life sciences model, under trusted access. The price is close to general frontier models. The real decisions are eligibility, routing, validation.
Read
Every AI vendor now sells delivery. Decide what your company keeps after the handover
Engineers trained by model providers, partner software paid from commitments, AI capacity shipped in weeks: delivery is becoming a product. La Madre's view on what to rent and what to own.
Read
"Human in the loop" is not a control if the human can't challenge the AI. IBM's CHRO study shows the gap
IBM found 71% of executives prize the ability to supervise, validate or override AI output, against 38% of employees. Review works only with skill, time and authority.
Read
Private AI infrastructure is being sold by workload, not by GPU count. Lenovo AI Express shows the shift
Lenovo and NVIDIA now package hybrid AI capacity in small, medium and large tiers defined by users, tokens per second and model size, shipping in 15 to 25 business days. Agents change the sizing math.
Read
Private AI is getting smaller. When does an enterprise agent belong on a box under the desk?
NVIDIA's 64GB DGX Spark arrives from six OEMs on October 23, sized for local agents with models up to 100 billion parameters. A framework for deciding what should run locally and what should not.
Read
Agent autonomy should be earned, not configured. A ladder for granting it one action at a time
Autonomy is usually set once, per agent, in a settings screen. La Madre's view: grant it per action class, on evidence, and take it back when the model or the risk changes.
Read
In regulated AI, the ontology may matter more than the chatbot. Databricks' Genie One for funds shows why
Databricks describes agents that reconcile positions and prepare NAV overnight, grounded in a governed business ontology with a trace behind every number. The design lesson outlasts the product.
Read
AI-built apps are easy. Safe access to live business data is the hard part, and AWS just moved it
Amazon Quick Apps can now query governed data live, as the viewer, with row and column security enforced on every query. That changes the failure mode of AI-built internal apps.
Read
Procurement, not benchmarks, may shape your AI stack. OpenAI's marketplace turns commitments into currency
Eligible OpenAI enterprise customers can now spend part of their commitment on 32 partner products, Decagon among them. Committed spend is becoming a factor in build vs. buy.
Read
When an AI agent becomes your SRE, permissions become reliability engineering
AWS DevOps Agent investigates incidents across AWS, other clouds and on premises, and the new Well-Architected Agent reviews architecture in preview. Both show where operational agents draw the line.
Read
Stop picking one best model. OpenAI's GPT-6 guide makes the case for routing by task
OpenAI's new guide to the GPT-6 family treats model, reasoning effort and speed as choices per task. For enterprise teams, the unit to optimize becomes cost per successful task.
Read
Model retirement is the new API deprecation, except it changes behavior, not syntax
Databricks retired Gemini 2.5 endpoints on October 2 and retires Claude Sonnet 4 pay-per-token on October 9. Third-party model lifecycles are now a production dependency to manage.
Read
Have an AI use case stuck between prototype and production?
Tell us what you’re trying to ship. We’ll reply with honest next steps.
Discuss a use case