Meet Ava. The AI assistant for benefits teams.

Ava gives your benefits team expert benefits work in minutes, not weeks, in two ways: in Ava Workbench you do the work, and with Ava Agents (beta) you delegate it. In Workbench, your team asks a question or starts a task and Ava returns finished work: cost and trend analyses across medical, pharmacy, dental and vision claims, vendor and point solution reviews, leadership reports, and employee communications. With Ava Agents, your team hands off ongoing work, like employee trend reports and vendor reviews on a schedule, and Ava follows through in the background and flags what needs your attention.

Ava is built for the questions benefits leaders actually need to answer: Where are our biggest cost drivers? Which vendors are delivering what their contracts promise? Are employees using the programs we pay for? What should we change at renewal? Zscaler answered its GLP-1 cost question instantly, and reports and briefs that took weeks came together in an afternoon.

Ava is not a static dashboard. It knows your program through the Benefits Ontology, and it knows the work: every agent comes with instructions, tools and memory built for benefits, so you don't have to direct every step. Ask a high-level question and Ava plans the analysis, coordinates specialists, and checks its work; then you follow up to dig deeper, the way you would with a skilled analyst on your team. Every answer shows its sources, so your team can review the work and keep the judgment.

Ava does the analysis itself, from all of your benefits data, in minutes, instead of storing data for someone else to query or delivering a point-in-time report weeks later. Data warehouses and reporting platforms organize structured data like claims, enrollment and costs, but they take technical skill to query and run on a fixed cadence. Consultant reports bring valuable strategic interpretation, typically on a periodic schedule.

There is a deeper gap, too. Some of the most consequential information in your program lives in documents: SPDs, PBM contracts, vendor agreements, stop-loss policies and TPA administrative agreements. Vault brings those documents together with your claims and other data, and patent-pending parsing reads complex plan documents that general AI can't. The Benefits Ontology then organizes everything into a trusted model of your program, so every answer reflects the full picture.

That is how Ava can check vendor contracts and performance guarantees against actual claims, find conflicting or outdated plan documents, or surface a PBM spread buried in a contract addendum or a claims pattern that contradicts the plan document.

Ava does not replace your data infrastructure or your advisory relationships. It makes both more valuable: your team and your broker or consultant work from the same live data, and your consultants spend their time advising instead of pulling data.

Ava works from every type of benefits data, structured and unstructured, brought together in Vault through prebuilt integrations. Structured sources include medical, pharmacy, dental and vision claims; point solution and vendor data (wellness programs, EAP, fertility, MSK, mental health); HRIS and employee demographic data; and financial and cost data for your benefits program.

Unstructured documents matter just as much: SPDs, plan guides, PBM contracts, vendor agreements, stop-loss policies and TPA administrative agreements. Patent-pending parsing reads complex plan documents that general AI can't, so nothing gets left behind, and the Benefits Ontology organizes it all into a trusted model of how your program actually works.

Ava can also ingest data from existing benefits data warehouses like Artemis or SpringBuk, so you don't need to rebuild pipelines you already have. You control what data you connect and when, and Ava uses only the data it needs. Most employers start with Carly's employee question data (available immediately) and core claims data, then add sources over time. The more you connect, the more complete Ava's picture of your program becomes.

Both: you can ask Ava anything in Ava Workbench, and Ava Agents (beta) takes on ongoing work in the background and flags what needs your attention. In Workbench, you ask a question or start a task and Ava returns finished work. With Ava Agents, you hand off work once, like a monthly employee trend report, a recurring vendor review, or a standing update on what's driving medical spend, and Ava follows through on a schedule, refreshing the work as new data arrives.

For example, Ava might flag that utilization of a point solution has dropped, that a particular population is driving outsized pharmacy costs, or that employees keep asking Carly about a benefit that isn't well documented. You find these issues while there's still time to act, not at renewal or in a quarterly review.

Ava Agents works within each user's permissions, never sends email or takes other external actions without approval, and admins can stop or revoke it at any time. Ava does the work; your team keeps the judgment and makes the decisions.

Ava turns what employees ask Carly into a clear view of where your program is working and where it isn't, because Carly and Ava share one platform. Every question shows where employees are confused, what they're looking for, and which benefits they can't find, while there's still time to fix it.

For example, if many employees ask Carly how to access mental health support, that signals both demand and friction in your current program. If employees keep asking about a specific plan provision, that points to a communication gap your team can address. Samsung found why dependents were skipping its mental health point solution in a ten-minute chat, used Ava to build targeted communications, and utilization rose 66%.

The loop runs both ways: what employees ask shapes what your team improves, and what your team learns shapes what Carly tells employees. HR and benefits teams see aggregate patterns and trends, never individual employee conversations.

Yes. Ava provides the unified data visibility employers need to measure and optimize benefits ROI across multiple dimensions. That includes tracking cost trends over time, evaluating vendor and point solution performance against engagement and outcomes data as well as contractual guarantees, identifying underutilized programs, and understanding where benefits investments are (and are not) delivering value.

Because Ava unifies claims, financial, vendor, and employee engagement data in a single platform, you can connect dots that are difficult to see when data lives in separate systems. For example, Ava can help you understand not just what a point solution costs, but how many employees are actually using it, what impact it is having on related claims, and whether it is worth renewing. These are the kinds of questions that traditionally require weeks of manual analysis. Ava makes them answerable in minutes.