The pharma industry has tried to solve fragmented sales attribution for two decades. Pharmafy's R&D programme combines geo-fencing, language models and graph relationships to do something no off-the-shelf CRM does: infer the link between a sale and the doctor who actually drove it.
Geo-fenced visit logging
Pharmafy uses the iPad's location to detect when a rep enters a known practice, hospital, or pharmacy boundary. The visit is auto-suggested in the call list — the rep just confirms what was discussed. Dwell time becomes a quantitative measure of effort.
R&D programme · geofence detection on partial-coverage 5G/4G networks; offline buffering; privacy-respecting opt-in geometry.
Text-to-link extraction
Free-text visit notes, emails, even SMS get parsed by an LLM to extract: doctor name, drug name, requested action, mood, follow-up dates. The system writes them back as structured CRM links automatically — no manual tagging.
R&D programme · AU-medicine-specific entity extraction; ambiguity resolution against the local doctor graph; confidence-based human-in-the-loop confirmation.
Rep-effort & territory analytics
Heatmaps of visit density per postcode, dwell-time distributions, time-of-day patterns. Anomaly detection flags reps who are over-stretched or covering "ghost" territory. Managers see effort, not just outcomes.
R&D programme · privacy-preserving analytics on movement data; territory-load optimisation that respects rest hours and travel realism.
Sales-to-prescriber inference
The hardest problem in pharma sales: invoices are issued to wholesalers, not doctors. Pharmafy combines geo-fenced visits, contact relationships, postcode coverage and Xero invoice timelines to statistically attribute sales back to the prescribing HCP.
R&D programme · Bayesian attribution; uncertainty quantification; transparent provenance — every attribution shows its evidence chain.
Relationship graph & auto-linking
Pharmafy maintains a live graph of doctors ↔ practices ↔ reps ↔ products ↔ sales. New invoices, contacts, or visits are automatically linked using name + address fuzzy matching, with a confidence score.
R&D programme · graph-based entity resolution tuned for AU medical naming conventions and AHPRA references.
Live engagement signals
Survey responses, sample drop counts, follow-up timeliness — fed into a per-doctor "engagement score" that's correlated with prescription growth. Tells the team where to invest more rep time.
R&D programme · multi-signal scoring with feedback-loop refinement.
