Punters Finance operating model,
mapped task-by-task.
Every Job-To-Be-Done across four ecosystem sides — decomposed into atomic tasks, each rated on the Stanford Human Agency Scale (H1 fully-AI → H5 human-essential), each mapped through SIPOC. This is the source-of-truth for what Max (the digital human), PF.OS (the orchestration layer), and real humans each own.
Aggregate Human Agency Scale distribution
How the 108 tasks split across the five HAS levels. This is the shape of the blended workforce.
Explore by ecosystem side
Customer
8 JTBDs · 54 tasks· 5 H5End clients arriving at PF — often after a bank has said no, often via warm referral.
Internal PF Staff
9 JTBDs · 32 tasks· 7 H5PF team performing triage, diagnostic reframing, matching, escalation, and lifetime relationship management.
Broker Network
5 JTBDs · 13 tasks· 3 H5External specialist brokers in PF's trusted network who execute the loan work and hold the client relationship post-match.
Financiers / Lenders
3 JTBDs · 9 tasks· 2 H5Lenders on the panel — both mainstream and non-bank / private funders (~60+ in current network).
H5 hotspots — the irreducible human core
17 moments where PF's design says human must own this. Automating them would be a strategic mistake.
- A3Escalate immediately to PF HumanOnboard, give consent, share situation · executed by PF Human
- A5The advocacy call — human-to-human, "here's what we're going to do for you"Get matched to the right broker · executed by PF Human
- A5Broker's first call — no re-explaining, immediate expertiseGet matched to the right broker · executed by Broker
- A6Notify client via warm PF Human moment, not a system emailGet funded — the loan journey itself · executed by PF Human + Broker
- A8Ask for referral at peak-emotion moment (not later)Refer others in · executed by PF Human
How to read this taxonomy
Every task has one HAS rating. When you see H1 → H1 → H5 → H1 waveforms across a JTBD, that's not a mistake — it's the design principle: automate the mechanical, protect the emotional.