The working documents and marketing drafts from the current FarmAgent / FieldAgent / Chief-of-Staff effort — all openable in any browser, no account needed. These are living drafts; treat them as in-progress.
Two product families (Field & Labor), the scheduling-as-trojan-horse thesis, and where FarmAgent & FieldAgent sit on the path to release.
What this session hardened — closed security criticals, the customer-id cleanup — and the honest punch-list that remains.
How we build: the spec-driven lifecycle, the discovery phase, and the real story of feature 038 — human vs. autonomous.
The shared context system every agent and teammate reads first — the map, the conventions, the one rule.
Exactly what the contador receives — the payroll CSV fields — and the boundary: hours only, no wages/CFDI/IMSS. We feed the contador, we don't replace them.
The accountant's real questions — "does it replace me?", "what data, what format, does it stamp the CFDI?" — answered straight.
The referral pitch: ~49% of these purchases start with the contador. Rev-share terms are a marked placeholder pending sign-off — no numbers published yet.
[FINISH] = the real demo link + a real case study, which the engine won't invent.Agave is a 6–7 year manufacturing process — you meter every stage of your distillery except the field. Measure it end-to-end (drone + phone); harvest optimization is the cherry.
The unmeasured stage → a measured process → Process/Quality/Labor management → the cherry on top: harvest optimization → CTA.
The process‑management case for the grower/owner — measure the whole 7‑year asset, not just the harvest.
Teremana's exact question, inverted: improvement only reaches what you measure, and the harvest optimization you want is computed from the early years — skip them and you forfeit the cherry.