One-Button Authoring

Updated 2026-07-13 · markdown version

One-button authoring is the class of AI tools that promise to produce a finished medical communications deliverable — a slide deck, a monograph, a response document — end-to-end from a prompt and a pile of sources. The button is both the interface and the ideology: it encodes the belief that the deliverable is the work, and therefore that generating the deliverable completes the work.

The category is real and well-funded. Sponsor-side money is flowing into automated authoring and review tooling (Eli Lilly participated in Revisto's $4M seed, Nov 2024; vendors across the space market "90% faster" figures — all vendor claims), and every pharma procurement conversation now includes an AI line. The appeal is honest: deliverables are templated, sources are finite, output language is regulated — on paper, ideal automation conditions.

The failure mode is equally honest: a one-button tool optimizes the cheapest part of the pipeline. Typing was never the cost. The cost is substantiation — proving every claim against its source at the page/column/paragraph level — and review, where unverified fluency goes to die (see the substantiation gap and the MLR wall). A button that produces more unverified sentences per hour raises the load on exactly the bottleneck it claims to remove.

Diagnostic questions for any tool in this class: Can it show, for every generated statement, the source passage that backs it? Does it maintain a claims ledger reviewers can audit? Does its speed metric count drafts produced or rounds survived? The honest ones will say "drafts." The category's contrast is the co-creation model.