Butterfly Network connects handheld ultrasound with AI tools and POCUS workflows

Butterfly Network combines handheld ultrasound hardware with imaging software, AI-assisted measurements and tools for running a point-of-care ultrasound program. Its iQ3 probe is one part of the offer. Compass AI addresses the organizational work around scanning, including documentation and quality review. The practical decision is whether a buyer needs a portable imaging device, a more consistent institutional workflow, or both, with the relevant software access included.

01 / ProductThe probe and the program software solve different problems

The iQ3 product page describes a handheld ultrasound device that connects to compatible mobile devices and supports multiple clinical applications through one probe. It includes imaging modes and assisted tools such as automatic bladder volume estimation and B-line counting. The AI-related feature set is therefore attached to specific imaging tasks, rather than a general chatbot that interprets arbitrary medical information.

Compass AI sits around the scan. It supports documentation, quality assurance, education, fleet management and links to clinical systems. Butterfly describes a voice-driven Documentation Agent and an AI-assisted QA Agent. The platform also supports non-Butterfly ultrasound devices, which makes it relevant to institutions that already have a mixed fleet. Device compatibility and workflow scope still need verification for the proposed configuration.

These layers should remain distinct when comparing products. Acquiring an iQ3 does not automatically establish a health system's governance process, and a Compass deployment does not imply replacing every existing scanner. The company's offer spans capture and operations. Buyers can assess each layer against the problem they actually have, while checking where shared accounts, memberships and integrations connect the two.

02 / AudienceIndividual clinicians and institutional leads buy differently

An individual clinician may primarily need portable imaging and a defined set of software features. An institutional POCUS lead may instead need consistent documentation, review and visibility across departments. In the latter case, the device is only one source of data. Training status, patient identity, examination records and the receiving archive can matter more than a single specification in a probe comparison.

Aidoc is an adjacent example of clinical imaging AI integrated with enterprise workflows. Viz.ai addresses clinical detection and coordination through a different set of products. Butterfly adds the ultrasound acquisition layer and POCUS program software. These comparisons locate its offer without treating the products as substitutes. A device purchase and a clinical information workflow require different evaluation criteria.

The product is not a substitute for the qualified professionals and institutional procedures required to perform and interpret ultrasound. It is also a poor fit for a buyer expecting unrestricted access to all software simply because a probe has been purchased. The pricing page explicitly separates hardware from memberships and distinguishes individual plans from organizational arrangements. Those distinctions need to shape the evaluation from the beginning.

03 / WorkflowA proposed pilot follows one examination into the record

For a proposed evaluation, choose one existing ultrasound workflow and involve the staff responsible for scanning, documentation and quality review. Use an institution-approved demonstration or retrospective process. Define the specific device, mobile operating system and membership being evaluated. This avoids a demonstration in which a feature is shown on a configuration that differs from the one the organization plans to buy.

Start with the device and the actual mobile hardware. The technical specifications list iOS and Android support, but operating-system support is not a guarantee that every phone or tablet works. Confirm the exact supported combination and review practical matters such as connection, charging, cleaning and availability between users. The pilot should expose the work needed to keep the device ready during ordinary shifts.

Next, examine the path from acquisition to a saved examination. Confirm that the user can associate the correct patient, retain the required images and complete the intended documentation. If evaluating an assisted measurement, review the relevant feature instructions and how its output is displayed alongside the source image. This is an operational demonstration, not a claim that the measurement is clinically validated for every proposed use.

For Compass AI, extend the trace through documentation and the QA queue. Ask who reviews a draft, what happens when the reviewer disagrees and whether the completed record reaches the expected clinical system. A voice-generated entry and a reviewed examination record are different stages. Measure whether the proposed configuration makes that distinction clear to clinicians and administrators rather than assuming automation eliminates the review step.

Include an interrupted or incomplete examination in the demonstration. A scan might be captured without finished documentation, require further review or fail to reach the archive. Inspect how staff find and resolve that state. The purpose is to see whether the program has an observable completion path. Sequenced has not performed this pilot or independently tested Butterfly's imaging, AI accuracy or claimed operational savings.

04 / PricingHardware and membership are separate purchases

The pricing page displayed dollar-denominated individual probe prices of $3,899 for iQ3 and $2,699 for iQ+ when consulted. It separately listed Core Technology at $299 per year and Advanced Technology at $420 per year, each including one user seat. The Advanced plan is the listed individual route that includes Auto Bladder Volume and Auto B-Lines; the lower membership should not be assumed to include those tools.

The same page offered an Advanced Technology one-time payment of $1,500, with software access guaranteed for five years. That wording is narrower than lifetime access and should be preserved. Probe cost is additional. The online page also places limits on the number of probes purchasable through individual routes, so a departmental deployment needs the organizational discussion rather than repeated assumptions about an individual checkout.

Organizational pricing is inquiry-based, with different packages and integration capabilities. Obtain a proposal for the devices, users, sites and Compass features required. Keep hardware, software access, mobile devices and implementation effort separate when comparing total cost. Displayed dollar figures describe the page inspected, not a globally available tariff; confirm local currency, taxes, delivery and feature availability for the purchasing jurisdiction.

For an individual comparison, calculate the cost over the intended ownership period without presuming that a future renewal price remains fixed. For an institutional comparison, include the effort needed to connect the existing archive and clinical systems. A subscription that appears inexpensive at the device level can still leave substantial workflow work outside the quoted scope. Conversely, existing compatible hardware may make an organizational software route more relevant than a fleet replacement.

05 / DistinctionsAI appears in both imaging and administrative work

Butterfly's broad distinction is that it works at more than one stage of ultrasound delivery. Imaging assistance addresses a task during the examination, while Compass addresses documentation and program oversight. These are different forms of automation with different evaluation criteria. A useful measurement output does not prove that documentation is complete, and better documentation does not establish that an image was adequate for its clinical purpose.

The iQ3 launch announcement also describes ScanLab as a separate AI-powered educational application. Education should be evaluated as education. A training environment can help users practice without implying that its behavior authorizes an unsupervised clinical workflow. The institution should identify which tools support teaching, which support examination tasks and which manage the resulting records.

A mixed-fleet platform can be valuable when several departments have adopted ultrasound independently. The editorial inference is that a common review process may make existing equipment more useful by reducing fragmented records. That benefit depends on the actual connected devices and interfaces. Vendor claims about revenue capture or faster documentation are starting points for local measurement, not guaranteed results for every institution.

06 / QuestionsFeature access is only one part of suitability

The first unresolved question is the precise relationship between a demonstrated AI feature and the contracted product. Check its availability on the chosen probe, software version, mobile platform and jurisdiction. A feature name in a marketing presentation is insufficient if the local purchase route or device combination differs. Obtain the current instructions relevant to the exact feature rather than generalizing from another probe generation.

The second question is how the institution distinguishes educational, incomplete and clinically finalized examinations. A program can collect more images without improving the completeness of its records. Demonstrate the permissions and status transitions that staff will use, including review and correction. The aim is to understand how AI assistance fits the organization's responsibilities, not to assume that automatically generated content is ready for clinical use.

The third question is whether the organization's claimed efficiency benefit is observable. Select a small number of concrete workflow measures, such as completion of required documentation or time spent resolving unmatched records. Keep those measures separate from clinical outcomes and from vendor financial projections. A pilot that shows a more reliable record path can justify further evaluation without claiming unmeasured improvements in diagnosis or patient care.

07 / DecisionBuy the layer that addresses the actual bottleneck

Butterfly Network is relevant to AI coverage because its imaging and workflow tools connect machine learning to a physical clinical task. For a clinician, the starting point may be a supported probe and appropriate membership. For a health system, the starting point may be the examination's path into documentation and review. The better choice depends on which dependency currently limits the program.

Evaluate a complete, locally supported configuration and retain the distinction between assistance, review and final professional responsibility. If that configuration makes the intended workflow demonstrably more reliable, broader adoption can be assessed on its merits. If the institution cannot yet identify the receiving record or QA owner, adding hardware alone is unlikely to resolve the underlying operational problem.