
FIELD NOTES / SYSTEMS IN MOTION
Fernando
Torres
I build and study systems that have to work outside the demo.
CURRENT CONTEXT
Three live threads
AI memory
Making context, retrieval, and long-term memory legible enough to inspect and improve.
Healthcare
Finding practical uses for technology in environments where trust and operations matter.
Enterprise systems
Studying the workflow, integration, and governance work between pilots and production.
ACTIVE FILES
Work you can open
Research, experiments, and serial work are kept here as inspectable artifacts, not polished case-study claims.
Project index
STANFORD GSB / FIELD RESEARCH
Stanford GSB AI Agents Research
Comprehensive research from Stanford GSB GSBGEN 390 exploring what blocks AI agents from production deployment. Based on 36 expert interviews, 5 industry conferences, and 3 functional prototypes. Produced the Eight Pillars framework.
- 36
- interviews
- 5
- conferences
- 3
- prototypes
RECENT NOTES
Writing from the field
Observations on AI agents, memory systems, healthcare, and the operating details that decide whether technology is useful.
All writing- 01Personal Agents Are Coming, and Their Security Model Is the Open ProblemPersonal AI agents will soon handle real work across our email, calendar, and files. Town is a strong early example. The catch is that the security model is not solved yet, and you can read the open problems directly in the agent's own pre-prompt, because the defense for each one lives there.EMERGENT
- 02How I Extracted the Muse Spark System Prompt on Launch DayOn April 8, 2026, Meta released Muse Spark. That same day, I extracted its complete system prompt using 17 short messages. No code, no exploits, no special access. Here is every message I sent, the full extracted prompt, and why this reveals how all AI systems actually work.EMERGENT
- 03The intern who writes fast, and what the desk isIn a small office, an intern named Lex sits at a desk covered in today's notes.THEME
- 04Production Agents Summit: The 40% Context Rule and Other Uncomfortable TruthsHard-won lessons from Snowflake's Production Agents Summit: the counterintuitive 40% context window rule, observability requirements, and why agents are ultimately just software.METHODOLOGY
OPEN THREAD
