Roupen Odabashian, Oncologist at Abbotsford Regional Hospital and Cancer Centre, Founder at MeDucation AI, Podcast Host at OncoDaily, shared on LinkedIn:
“Healthtech funding this month: Trase raised $107M and xCures $46M, both to scale AI agents.
Meanwhile the takeaway from ASCO 2026 was blunt – the next era of oncology AI “won’t be won by the best model, the highest benchmark, or the flashiest press release.”
Those two facts are in tension, and the tension is the whole game. Capital is flowing to model capability. Adoption is decided by everything else: does it fit the workflow, does it earn clinician trust, does it survive contact with a real clinic on a real Tuesday.
I’ve felt this building in medical education. The model was never the hard part. The hard part was the boring scaffolding around it, making the output trustworthy enough that a busy learner or clinician actually changes what they do. That’s where the moat quietly lives, and it’s unglamorous enough that most demos skip it.
The winners in clinical AI probably won’t have the best benchmark scores. They’ll have the best last mile. Are investors funding the model, or the mile?”
