Product Roadmap · Now / Next / Later · 6‑month horizon

Enspescio — June to December 2026

Two part‑time builders (Andy + Slava). The roadmap is deliberately thin in Now so the near‑term commercial milestone isn't crowded out, and weighted toward three threads: engine reliability, launch & onboarding, and the data foundation underneath both.

⚑ Anchor Second paying tenant live by June 30, 2026. Everything in Now either unblocks this onboarding or is required to keep the first tenant healthy while it happens. (~6 days out — now looks unlikely on time; as of June 24 no reconciliation progress reported and the June 23 app‑vs‑Amazon validation went unmentioned. See capacity risk.)
Filter lane:

Assumptions

  • Confirmed — one paying tenant exists today. The Amazon S&S account is the active test tenant; the June 30 goal is paying tenant #2.
  • "Launchable" = trustworthy numbers for one onboarded tenant, not a broad public launch. Self‑serve can wait.
  • ~1.0 FTE total across both of you. Now holds ≤3 real efforts at once on purpose.
  • The diligence/M&A line is real strategy but not a Now/Next priority given capacity — parked in Later.

Risks & dependencies

  • June 30 is tight — now likely to slip. ~6 days, part‑time. As of June 22 the active‑count work was still open (Slava finishing data validation, app‑vs‑Amazon validation set for June 23, unified page not yet implemented or demoed). June 24 brought no reconciliation progress — the June 23 joint validation went unmentioned, Andy said he hadn't had time to decide next steps for the account, and the day was given to backlog housekeeping (rescheduling pending jobs, LeadHub sync, rebuilding a dropped process chain). With the central blocker apparently untouched, the milestone looks unlikely to land on time. Protect tenant‑2 onboarding and let cycle‑axis cleanup slide to Next.
  • Active‑count reconciliation — partly a display bug, partly still open. June 22 split it in two: the internal 339/347/401/421 inconsistency was a broken/empty 'source' column (now hidden — resolved), while the reconstructed ~339 active vs Amazon's reported ~515 gap is still unvalidated. A single global KM window (~1.5 yr) plus only ~2 yr of history may overstate active; per‑cohort windows are the proposed fix, but unproven and add work right before the milestone.
  • Inferred‑frequency trap. Per June 14, when frequency is estimated from intervals the cycle‑axis can collapse to a delivery count. The trailing‑window cadence mitigates the mode‑mislabeling case, but a sequence like 30/30/180 still collapses to a median of 30 — misclassifying a genuine monthly→6‑month shift (flagged June 19). Axis decision still open.
  • Identity gaps. ~30% of orders historically lacked email; a June 18 re‑run reportedly pulled them all — unverified, so still a risk until confirmed.

Explicitly not now (parking lot)

Incremental‑ROAS (Bayesian MMM) productization · public self‑serve signup · breadth of integrations beyond Amazon + Recharge · packaged M&A scorecard · advanced alerting/automation. All valuable, none worth diluting the June 30 push or the part‑time capacity.