Evidence at the point of care.

You know the number. Eighteen minutes with each patient in the United States; about two in India. The notes and the inbox wait for the evening. The next patient brings a screenshot from the family group chat and asks you to explain it. At 2 a.m. you need a dose, so you ask the chatbot everyone uses. It sounds certain. It shows no source you can check. It owes your patient nothing. You carry what it gets wrong.

Asha for clinicians (1:34). Benchmark transcripts in the film are rendered in a chat layout, abridged and labelled; they are not app screenshots. Every Asha screen is real, captured 16 September 2026.

Built for exactly this

• Dosing, with the label. Ask for a renal dose. Asha answers from the prescribing information and shows you the label section itself, one tap away.
• Interactions, cited. The label’s own words first, then Asha’s analysis with the citations. No automated severity grading.
• Guidelines, with the sources. Ask what the guideline says. Every claim is numbered to its source, so you can check it yourself.
• A fiduciary, to you and to your patient. No ads. Your data stays yours and is never sold. When it is serious, she says so.

The citation gap, measured

The habit that matters most at the point of care is the one general assistants skip: showing the source. On HealthBench Professional, 524 shared items, September 2026, graded by GPT-5.4, a bare frontier model answered with no citation at all on 74% of items. Asha, running on the same underlying model inside her architecture, cited nothing on 0.6%.

Citations are the floor. On the benchmark’s official length-adjusted metric (525 items, 14 September 2026, graded by Claude Opus 4.8, two trials pooled), Asha’s best configuration scored 65.1 and the leading bare frontier model 67.0 in the same harness with the same grader. The full board, including the runs where Asha did not lead, is on the public register at dnai.systems/benchmarks. Neurosymbolic, and tested in public: every run with its harness, its grader and its date.

Research preview, by invitation

Asha for clinicians opens as a research preview on 1 October 2026. Full access goes to the first hundred invited clinicians who give feedback. Verification is self-serve: in the United States by NPI against the NPPES registry, in India by NMC registration number. Verified professionals use the clinician mode free. Read more at Asha for clinicians.

Built by physicians. A fiduciary healthcare AI that always has your back. Evidence at the point of care.