供资
Avante 融资 30M A系列扩大AI Agents全球受益团队
在 Google 上的首选来源中添加 Unite.AI
一个福利小组可以负责雇主支出的一大部分，同时仍然从电子表格、声称报告、合同和电子邮件线索中收集续约决定。 Avante拥有3000万美元的融资系列A，将这些信息与AI agents连接，后者负责调查程序，准备工作以供审查，并帮助雇员使用他们已经获得的福利。
Frist Cressey Ventures领导了这一回合，新投资者Sorenson Capital、赫斯特Ventures和Morgan Stanley Next Level Fund以及现有投资者FUSE也加入了这一回合。 融资带来 Avante的资金总额达4000万美元。
西雅图公司将利用资本进行全球扩张，并进一步开发其agent平台。 公告沿用了Ava Agents的β 发布，后者在背景中进行经常性的福利工作，并返回各小组检查和批准的产出。
阿凡特的商业命题结合了经常被单独处理的两个需求：帮助雇主了解其福利投资的绩效，帮助雇员浏览其雇主购买的方案。
效益分析超越问答工具
对雇主来说，对福利供应商进行评价可能需要将合同承诺与声称、使用和服务报告进行比较。 准备证据需要时间谈判才能开始缺少文件、定义不一致和最后期限增加了进一步的工作。
Avante 称公司agents公司将合同与实际的声称公司进行比较，审查计划文件中的错误和不一致之处，跟踪最后期限并编写领导报告。 目的是向福利小组提供完成的分析或评估的具体问题，而不是另一个断开的数据来源。
称 Rohan D'Souza是Avante的首席执行官兼联合创始人
最新的Ava Agents介绍使得工作流程更加具体。 通过邮箱连接，供应商报告可以触发与合同的比较。 如果缺少信息，Ava准备进行后续审批，监测答复并继续进行审查。 该小组收到一份来源的简要说明、调查结果和一份答复草稿。
Ava Agents与公司互动分析经验Ava Workbench共享一个工作空间。 用户可以检查证据或改变调查方向。 背景agents可用β；该状态应与客户已经使用的更广泛的平台区分开来。
结构化模式连接计划、人员和供应商承诺
Avante的平台架构描述了一种福利本体学：一种结构化的模式，将计划、方案、人口和资格规则连接到各国和各企业单位。 相关的Vault层带来组合计划文件，声称，资格数据和供应商信息，保留源文件和有效日期。
这一结构很重要，因为准确的效益答案不仅取决于找到一个相关段落。 该系统必须确定适用哪种计划、谁有资格以及信息是否属于正确的计划年。 同一文件可能对一名雇员有用，不适用于另一名雇员。
Avante 称其处理处理处理困难的表格，图像和嵌套规则，而审计 agents 旗缺少文件和相互冲突的信息。 新材料可以更新结构化模式，将问题浮出水面以供审查。
因此，该技术意在agent回答一个问题或评价一个程序之前明确关系。 例如，将使用与资格相比较，比仅看使用量更有意义：人数少可能反映合格人口的有限或雇员不知道的福利。
Ava支持团队，而Carly支持员工
Ava是从事分析和福利工作的雇主所在的agent。 Carly是帮助员工的助手，提供个性化的全天候指导，介绍个人的计划和选择。
Avante的AI技术文件描述了涵盖范围、计划比较、退休福利和休假程序等领域的专用子agents。 员工可以通过包括Slack,Microsoft Teams,ServiceNow在内的现有渠道获取服务，电子邮件和一个品牌门户。
称公司使用来自OpenAI,Anthropic和Google的模型，而不是训练一个基础模型本身。 其具体领域的工作围绕这些模式进行：文件处理、程序结构、检索、任务路由、评价和控制。
这种区分对于了解产品很重要。 普通模型可以解释保险术语；回答特定雇员可以获取什么，需要当前的计划信息和适当的许可。 Avante 称的答复引用了来源部分，并进行了实时评估，并附有标注的答复可供人类审查。 这些是产品控制，而不是证明每个答案都是无错误的。
为了造福领导人，这两种经验可以相互通报。 重复的雇员问题可能表明沟通不明确，或一个被忽略的有用程序。 团队随后可以调查差距，决定是改变推广范围还是方案本身。
客户增长是通过账户、更新和聘用计量的
阿凡特在发布18个月后报告超过70个客户账户，年复一年的预订量几乎增加了5倍。 每一个符合续约条件的客户都会续约预订是衡量合同业务的尺度，不同于确认的收入，续约索赔适用于符合条件的客户群体。
称公司现在的平台支持超过385,000名员工和家庭成员。 它还列出了与福利顾问的伙伴关系，包括Gallagher、OneDigital、Segal Benz、Howden和USI保险服务。
在Real Chemistry，根据公告，83%的美国劳动力与Avante有接触。 Olga Reupert)，该公司的SVP"Total Rewards,HR Global Operations and Eleasement"，称按需指导减轻了行政负担，让受益团队专注于战略重点。
Samsung半导体公司报告说，找到并使用现有心理健康支助的雇员和家属增加了66%。 这说明了利用率的提高，而不是声称66%的员工使用该福利。 该平台去年回答了8 500多个雇员问题。
这些是公司提供的客户结果。 它们说明了使福利更容易理解和管理的预期价值，而不同的雇主在认识、方案设计和数据质量方面会有不同的起点水平。
全球扩展新增另一个复杂层
A系列将支持从更广泛的国际角度看待各种福利方案，并在整个福利生命周期增加agents。 多国雇主必须比较因国家、计划年和雇员人口而不同的方案，使共享数据结构特别相关。
Avante的Ava Agents公告称组织数据仍然分离，agents在访问权限范围内运行，工作记录供审查。 起草供应商外联和雇员沟通文件供批准，让福利小组控制所发送的信息。
在人力资源业务中，从福利分析到预测员工队伍的时间安排，资金符合向AI的更广泛转变。 实际的机会是减少收集证据和协调日常工作所需的努力，同时与负责者一起作出相应的决定。
对Avante来说，下一个测试是将早期采用转变为全球福利团队可重复的操作模式。 3 000万美元系列 A为扩大这一努力提供资源，将方案分析、供应商问责制和雇员指导联系起来。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Funding
Avante Raises $30M Series A to Expand AI Agents for Global Benefits Teams
Add Unite.AI to your preferred sources on Google
A benefits team can be responsible for a major share of an employer’s spending while still assembling renewal decisions from spreadsheets, claims reports, contracts and email threads. Avante has raised a $30 million Series A to connect that information with AI agents that investigate programs, prepare work for review and help employees use the benefits already available to them.
Frist Cressey Ventures led the round, joined by new investors Sorenson Capital, Hearst Ventures and the Morgan Stanley Next Level Fund, alongside existing investor FUSE. The financing brings Avante’s total funding to $40 million.
The Seattle company will use the capital for global expansion and further development of its agent platform. The announcement follows the beta launch of Ava Agents, which performs recurring benefits work in the background and returns outputs for teams to inspect and approve.
Avante’s commercial proposition combines two needs that are often handled separately: helping employers understand the performance of their benefits investment, and helping employees navigate the programs their employer has purchased.
Benefits Analysis Moves Beyond a Question-and-Answer Tool
For an employer, evaluating a benefits vendor can require comparing contractual commitments with claims, utilization and service reports. Preparing that evidence takes time before a negotiation even begins. Missing documents, inconsistent definitions and deadlines add further work.
Avante says its agents compare contracts with actual claims, review plan documents for errors and inconsistencies, track deadlines and prepare leadership reports. The aim is to give the benefits team a completed analysis or a specific issue to assess, rather than another disconnected source of data.
“Benefits teams are responsible for one of the largest and most consequential investments their companies make, yet they’re still operating with outdated information, archaic spreadsheets, and technology built for another era,” said Rohan D’Souza, Avante’s CEO and co-founder, in the announcement.
The latest Ava Agents introduction makes the workflow more concrete. With a mailbox connected, a vendor report can trigger a comparison with the contract. If information is missing, Ava prepares a follow-up for approval, monitors the response and continues the review. The team receives a sourced brief, findings and a draft response.
Ava Agents shares a workspace with Ava Workbench, the company’s interactive analysis experience. Users can inspect evidence or change an investigation’s direction. The background agents are available in beta; that status should be distinguished from the broader platform already used by customers.
A Structured Model Connects Plans, People and Vendor Commitments
Avante’s platform architecture describes a Benefits Ontology: a structured model connecting plans, programs, populations and eligibility rules across countries and business units. A related Vault layer brings together plan documents, claims, eligibility data and vendor information, retaining source files and effective dates.
That structure matters because an accurate benefits answer depends on more than finding a relevant paragraph. The system must determine which plan applies, who is eligible and whether the information belongs to the correct plan year. The same document can be useful evidence for one employee and inapplicable to another.
Avante says its processing handles difficult tables, images and nested rules, while audit agents flag missing documents and conflicting information. New material can update the structured model, with issues surfaced for review.
The technology is therefore intended to make relationships explicit before an agent answers a question or evaluates a program. Comparing utilization with eligibility, for example, is more meaningful than looking at a usage count alone: a low number may reflect a narrow eligible population or a benefit that employees do not know exists.
Ava Supports the Team, While Carly Supports Employees
Ava is the employer-facing agent for analysis and benefits work. Carly is the employee-facing assistant, providing personalized, around-the-clock guidance about a person’s own plans and options.
Avante’s AI technology documentation describes specialized sub-agents for domains such as coverage, plan comparisons, retirement benefits and leave procedures. Employees can access the service through existing channels including Slack, Microsoft Teams, ServiceNow, email and a branded portal.
The company says it uses models from OpenAI, Anthropic and Google rather than training a foundation model itself. Its domain-specific work sits around those models: document processing, program structure, retrieval, task routing, evaluations and controls.
The distinction is important for understanding the product. A general model can explain an insurance term; answering what a particular employee can access requires current plan information and appropriate permissions. Avante says its responses cite source sections and are evaluated in real time, with flagged responses available for human review. Those are product controls, not proof that every answer is error-free.
For benefits leaders, the two experiences can inform one another. Repeated employee questions may reveal unclear communications or a useful program that is being overlooked. The team can then investigate the gap and decide whether to change outreach or the program itself.
Customer Growth Is Measured in Accounts, Renewals and Engagement
Avante reports more than 70 customer accounts 18 months after launch, with bookings up nearly fivefold year over year. It says every customer eligible for renewal has renewed. Bookings are a measure of contracted business, distinct from recognized revenue, and the renewal claim applies to the eligible customer cohort.
The company says its platform now supports more than 385,000 employees and family members. It also names partnerships with benefits advisors including Gallagher, OneDigital, Segal Benz, Howden and USI Insurance Services.
At Real Chemistry, 83% of the U.S. workforce has engaged with Avante, according to the announcement. Olga Reupert, the company’s SVP of Total Rewards, HR Global Operations and Enablement, said on-demand guidance has reduced administrative burden and allowed the benefits team to concentrate on strategic priorities.
Samsung Semiconductor reported that 66% more employees and dependents found and used existing mental health support. That describes an increase in utilization, rather than a claim that 66% of the entire workforce used the benefit. Personify Health said the platform answered more than 8,500 employee questions over the past year.
These are company-supplied customer results. They illustrate the intended value of making benefits easier to understand and administer, while different employers will have different starting levels of awareness, program design and data quality.
Global Expansion Adds Another Layer of Complexity
The Series A will support a broader international view of benefits programs and more agents across the benefits lifecycle. Multinational employers must compare programs that differ by country, plan year and employee population, making a shared data structure particularly relevant.
Avante’s Ava Agents announcement says organizational data remains separated, agents operate within access permissions and work is recorded for review. Vendor outreach and employee communications are drafted for approval, leaving the benefits team in control of what is sent.
The funding fits a broader move toward AI in HR operations, from benefits analysis to predictive workforce scheduling. The practical opportunity is to reduce the effort required to assemble evidence and coordinate routine work while keeping consequential decisions with the people responsible for them.
For Avante, the next test is turning early adoption into a repeatable operating model for global benefits teams. The $30 million Series A gives it resources to expand that effort, connecting program analysis, vendor accountability and employee guidance in the same platform.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.