App Update · October 27, 2026
Vervi 1.01 lets the agent build a full playbook from a protocol document or a short prompt, and upgrades the learning layer behind patient question topics and AI drafting.
Update 1.01 is our first numbered release. It brings two major changes: the Vervi agent can now build complete playbooks for you, and the learning layer that powers patient question topics and AI drafting has been upgraded. It also includes a set of reliability and quality fixes from the past few weeks.
Playbooks deliver each instruction at the right point in a patient's timeline. Until now, building one meant writing every instruction first and then assembling the timeline by hand, even when a single protocol document already described every step.
The new Create Using Agent tile on the Playbooks page removes that work. It offers two paths.
Choose Suggest Playbooks and the agent reviews your published instructions, then proposes playbooks that combine them into timelines, such as a pre-op preparation guide, day-of-surgery instructions, and recovery instructions for the same procedure. Each suggestion shows its timeline at a glance. Pick one and it is saved as a playbook you can edit.
Choose Build New and either describe the procedure in a sentence or two, or attach a PDF or Word document such as an existing protocol or patient packet. The agent then:
Builds run in the background, so you can keep working and come back when the playbook is ready.
The learning layer is the part of Vervi that gets better the more care teams use it. It learns from how instructions are written and edited, and from the questions patients ask, so Vervi understands what a piece of content or a question is actually about, not just the words in it. In 1.01 it gets three upgrades.
Inquiry Analytics groups patient questions into topics. Previously, closely related questions could end up in separate topics, for example "Pain Level" and "Pain Management". Classification now checks a new question against the existing topics for that instruction and merges it when the clinical subject is the same, opening a new topic only when it is genuinely different. Near-duplicate topics are also reconciled over time. The result is fewer, clearer topics and more accurate counts.
The builder, the editing agent, and AI enhancements now draw on patterns learned across the platform and within your specialty, for example how plastic surgery practices tend to use visuals or how orthopedic practices structure activity progressions. These patterns are learned from de-identified content and edits. They sit underneath your practice settings and personal rules, which continue to take precedence.
Instruction content is now tagged by the role each section plays (such as preparation, wound care, or warning signs), the intent behind edits, and tone. This is what allows the agent to recognize that an existing instruction already covers a playbook phase, and it gives future features a better understanding of how good instructions are put together. Content is scrubbed of protected health information before it is analyzed.
Agent-created playbooks are available now from Playbooks → Create Using Agent in the provider workspace. Questions or feedback on this release are welcome through the contact form.
From the Playbooks page, choose Create Using Agent. The agent can suggest playbooks assembled from your published instructions, or build a new playbook from a procedure description or an attached PDF or Word protocol. Each phase becomes its own instruction on a timed timeline, and the playbook opens for your review before anything is sent to a patient.
Yes. When one of your published instructions already covers a phase, the agent links it instead of writing a duplicate. It only writes new instructions for the phases your library does not cover.
The learning layer is the part of Vervi that improves as care teams use it, learning from how instructions are written and edited and from the questions patients ask. In 1.01, patient questions are now merged into an existing topic when they are about the same clinical subject, so Inquiry Analytics shows fewer, cleaner topics. AI drafting and enhancements also draw on de-identified patterns learned across the platform and within your specialty, while your own rules continue to take precedence.