Ventures
FarGrit invests at the intersection of physical product and applied technology. The focus is applied AI in regulated industries: medtech and connected devices, drug discovery, clinical development, healthcare delivery, nutrition, and adjacent domains where rigour and verification matter.
The teams I work with are typically pre-seed to Series A — founders with deep technical or clinical expertise who need to turn that into a fundable, deployable, defensible business. Sometimes I co-found. Sometimes I invest. Sometimes I sit alongside the team for a defined window — a raise, a launch, a regulatory inflection.
The support draws on founding design-led product companies — Veryday, acquired by McKinsey in 2016 — and six years there as an Associate Partner leading customer experience design for Life Sciences across EMEA, the US, and Asia. It falls into three areas.
Business development. Due diligence on partnering opportunities, team capability, and technology readiness. Deal-making with Big Pharma, biotech, contractors, and investors — Master Services Agreements, performance-based contracts, joint ventures. A working network across Life Sciences from R&D through commercial.
Product, service, and operations. Established best practices from global R&D and commercial development — asset development, target product profiles, trial acceleration, drug delivery and combination device design, HCP and patient engagement. Technology assessment and value-creation estimates. Sharpening value propositions, messaging, and channel choice. Design and brand asset development for joint ventures.
Team and culture. Recruiting senior commercial, design, and clinical talent. Setting the operating cadence and digital tools that make small teams effective.
What ties this together is a view shaped by FOS Medical and the verification work behind it: in regulated domains, founder depth and disciplined verification matter more than raw model intelligence.
Active Ventures
FOS Medical is a Sydney preclinical venture working on glioblastoma, multiple myeloma, and lung adenocarcinoma. The team includes Prof Dale Bailey — h-index 64, past ANZSNM President, one of the architects of total body PET — and Dr James Drummond, RNSH neuroradiologist and Brain Imaging Lab Director. The FOS Engine runs three frontier models in parallel against expert-curated theranostic data, with adversarial uncertainty induction before consensus voting. First candidate to human trials in 18 months — roughly 5× the industry pace. The RINO Project on systematic GBM target selection was presented at SNO’s 7th Quadrennial World Federation of Neuro-Oncology Societies meeting in late 2025.
The thesis is simple. In regulated clinical workflows — prior authorisation, medical coding, claims adjudication, specialty documentation — frontier model intelligence sets the floor. Expert-verified data and a real verification architecture set the ceiling. The forward-deployed engineering model that Anthropic and OpenAI are scaling into healthcare assumes that layer already exists. Mostly, it doesn’t. CureBench builds around that gap: clinical task libraries, eval suites grounded in real workflows, and verification scaffolding that gives FDE teams something defensible to deploy against. Concept stage; structure and partners forming through 2026.
Radiology departments lose roughly 25 minutes per patient to manual prior retrieval — a coordinator locating, retrieving, and loading imaging history into PACS before a radiologist can work. Falcr automates that coordination layer without replacing the underlying software or the professional. The architecture has five independently auditable layers: observation, training, execution, governance, and human escalation. Per-customer LoRA adapters fine-tune on the hospital’s own behavioural traces — the moat is the workflow data the system accumulates, customer by customer. Go-to-market is a 60-day pilot with department heads, producing a documented hours-returned metric that drives the CIO conversation and enterprise expansion.