Hi, I'm Fred.
I've spent the last few years building AI tools that solve complex enterprise problems. I started TinyMacro to build sustainable software with emergent patterns I found from experience.
What I am trying to fix
Most companies do not need another place to type prompts. They have research that takes too long and jobs held together by one person knowing every exception. I care less about whether a system looks intelligent than whether it removes that drag.
Working inside a company shows me where the written process and the real one disagree. Sometimes a model helps. Sometimes the right answer is better search or a small piece of ordinary software. I want to remove the bottleneck, not force every problem into an AI product.
That is why consulting and TinyMacro fit together for me. Consulting gives me the raw material. TinyMacro is where a solution that keeps recurring can become software.
What I think will matter
AI becomes useful when it finishes work a company already values. A research brief arrives when someone needs it, or a record stays current without manual cleanup. The company should not have to care how many prompts or subagents were involved.
A company should be able to adopt a better model without rebuilding the job around it. The durable part is the harness: the state, tool access, recovery, and evidence surrounding the model. When a better provider appears, switching should not mean reconstructing the work.
An answer can look right and still come from the wrong source or an unsafe action. That is why I care about evals. I would rather test a few claims tied to the job than maintain a large scorecard nobody uses.
Company knowledge matters only when a system can find the right material and show why it used it. I start with navigable documents and direct reads. Vector RAG has to beat that simpler baseline. For larger bodies of material, I am testing whether RLMs can inspect and divide the context instead of compressing everything into one prompt.
I expect the best vertical software to come from work inside a specific industry. The exceptions and unwritten rules are where much of the value sits. Consulting lets me learn them before I generalize.
Experience and education
I built AI infrastructure for Wheel the World, an accessible travel company, and Larta Institute, a federal research accelerator working with NOAA and USDA programs. I have also built content systems for ShoplyAI.
I am finishing an economics degree at UCLA in December 2026, with a minor in Environmental Systems and Society.
Consulting · LinkedIn · GitHub · Resume
Have a specific problem?
For a project or a conversation about TinyMacro, email me or book a call.