Denver keeps Mountain time. It is known for aerospace, telecommunications, energy and software, and for the federal research laboratories nearby. What follows is, for each of those industries, the thing it tends to need from AI, the service that answers it and the published work that carries over, or the plain statement that none does.
Mountain time against Monterrey
From mid March to early November Denver and Monterrey are on the same hour. From early November to mid March Denver is one hour behind Monterrey, because Colorado changes its clocks and Monterrey does not. Either way the two working days nearly coincide, and nothing waits overnight because of the clock. I keep calls short and put the rest in writing, next to the code.
Aerospace programs
An aerospace program produces requirements, test reports and change notices in volume, and every one of them is controlled. Engineers want to question that record without leaving it. No case study on this site comes from aerospace. What transfers is retrieval over a controlled corpus, the work I describe under RAG and data platforms, and access to the systems behind it that is narrow and logged.
Telecommunications and network companies
A network company has its truth in many systems: billing, provisioning, tickets, the network itself. Before an agent can help a customer or a technician, it needs one place to read from. The published work here is a data warehouse fed from operational systems, and an analytics platform built on the same idea. With that base, internal tools with AI have something solid under them.
Energy
Energy companies run on field reports, permits, contracts and maintenance histories, many of them written far from a desk. No case study on this site comes from energy. What transfers is a pair: answers drawn from the company's own documents, and small internal tools that put those answers in front of the person who needs them.
Federal laboratories and the companies around them
A research laboratory cares about method: the claim, the test that could have refuted it, the record of what was run. No case study on this site comes from government. What I can show is how I work on my own questions. My research on agents is pre-registered, and a guide on automated LLM evaluation sits among the engineering pages.
Software companies
Software teams tend to bring one of two requests. The first is an agent that needs tests before its next release, which is LLM evaluation. The second is a product that agents should be able to use, which is an MCP server. The open-source Airtable MCP server is the public example of the second. The example of the first is the set of evaluation gates in a multi-agent system for a higher-education group with 700,000+ students.
From first message to first delivery
Reach me by email, on LinkedIn or on a scheduled call, with the role or the problem. I read the process, then write back with a first deliverable, the access it needs and what stays out of scope.
Where this stops
At the door of your building. I am in Monterrey, and a role that must be filled on site is not one I can fill.