Rivelle partners with Ansel to put Ansel Enterprise in the hands of 3,412 employees. Every function now works alongside an assistant it built itself, and the use cases compound week over week.
Rivelle has worked at the intersection of protein science, manufacturing, and clinical care for eleven years. Its mission is to move a therapy from a validated target to a dosed patient in a fraction of the time the industry considers normal.
The company has worked with Ansel since the spring of 2023. Ansel Enterprise is now changing how each function operates, from regulatory writing to plant scheduling.
Rivelle plans to bring nine new products to market over the next five years, from a respiratory vaccine to individualized oncology treatments. To get there it has taken a deliberately people-first, technology-forward approach: test early, measure honestly, and keep the humans accountable for every clinical decision.
“We believe very deeply at Rivelle that this technology reshapes how medicine gets made. We are looking at every business process—legal, research, manufacturing, commercial—and asking how we would redesign it with AI in the room.”
Putting AI in every pair of hands
Rivelle adopted generative AI the way it adopts any other technology: start with the person doing the work, then invest in the ground they stand on. Before a single license was bought, the team spent a quarter mapping where judgment actually lived in each workflow.
The objective was full adoption and working proficiency across everyone with a digital seat, inside of six months. “We believe in collective intelligence when a paradigm shifts,” said Raghunathan. “It has to be everyone together, everyone with a voice, nobody left behind.” A dedicated team of eleven ran a bespoke transformation program combining individual, collective, and structural change.
Individual programs included in-depth listening research and trainings held in person, online, and with dedicated AI learning companions. “Using AI to teach AI was the thing that made it stick,” Raghunathan points out. Collective programs included a prompt contest that surfaced the 94 strongest internal practitioners, who were then organized into a standing cohort of AI champions. Local office hours ran in every business line and geography, and an internal forum on AI now has 1,840 active weekly participants.
That work led to an early win: an internal assistant called mAsk, launched at the start of 2023 on Ansel’s API. It was adopted by roughly four out of five employees and became the foundation the Enterprise rollout stood on.
“Nearly every company wants to do generative AI and only a sliver of them get anywhere. They fail because they never built the machinery for moving a workforce onto new capabilities.”
Building momentum with Ansel Enterprise
When Ansel Enterprise launched, Rivelle had a decision to make: keep developing mAsk as an all-purpose internal tool, or give every employee the full product?
“We are a science company, so we tested it like one,” said Marta Sandoval, Head of AI Products and Platforms at Rivelle. Her team ran a six-week evaluation comparing mAsk against two alternatives. “The satisfaction gap was not subtle. Enterprise was the company favorite and the one we decided to double down on,” Sandoval said.
Once employees could build their own assistants without writing code, the limit became imagination. “We were never here to fill a bucket, we were here to light a fire,” Sandoval said. “The fire spread. Hundreds of use cases, most of them ones we would never have specified from the center.”
Within two months of the rollout:
- Rivelle had 612 assistants running across the company
- 43% of weekly active users had built at least one themselves
- Each user averages 118 Enterprise conversations per week
Augmenting clinical trial design with assistants
One of the solutions Rivelle continues to develop and validate is an assistant the team calls Dose Atlas. It reviews clinical data, integrates and visualizes large datasets, and works as a data-analysis partner to the study team—augmenting clinical judgment rather than standing in for it.
“Dose Atlas gives us the supportive rationale for why a particular dose was picked over the alternatives. It lets us build custom visualizations, and it lets the study team keep interrogating the data from angles we would not have had time to run manually,” said Ife Adeyemi, Director of Clinical Development at Rivelle.
The assistant applies standard dose-selection criteria, provides a rationale, references its sources, and generates the charts that illustrate the finding. That leaves a detailed review led by humans, with AI input, prioritizing safety and sharpening the profile before a candidate moves into late-stage trials.
“It raises the amount of work we can get through as a team. We can evaluate genuinely large amounts of data efficiently and accurately, while keeping security and privacy intact,” Adeyemi added.
Tightening compliance and telling the company’s story
Rivelle’s legal team built Contract Companion, an assistant that reads a supplier agreement and returns a plain-language summary of every substantive provision. What used to be a two-day read is now a first pass measured in minutes, with a lawyer signing off on the judgment calls.
The investor relations group went further and built an assistant that keeps the quarterly earnings narrative consistent, so that content prepared for analysts uses the same defined terminology as the language filed with regulators.
“Sometimes we are in our own world, and in there the brand thinks beyond that,” explained Kate Cronin, Chief Brand Officer of Rivelle. “What would my mother want to know about Rivelle, versus a regulator, versus a nurse? An assistant is remarkably effective at holding all three audiences at once.”
| Number & title | Section | Detailed analysis of the substantive provisions |
|---|---|---|
| Recitals | Background information | Establishes the context for the agreement, noting that the supplier manufactures and distributes certain products and the distributor wishes to obtain exclusive rights in a defined territory. |
| 1. Appointment | Distributor designation | Supplier exclusively appoints distributor within the named territory for the products listed in Exhibit A. |
| 2. Term and termination | Agreement duration and termination conditions | Initial term of 36 months with automatic renewal, terminable for cause on 30 days’ written notice. |
Rivelle workspace chats are not used to train Ansel models. Assistants can make mistakes.
How a team of 3,400 performs like one ten times its size
With nine launches planned over the next few years, Rivelle treats AI as a core part of staying small enough to move while setting new benchmarks on speed.
“If we had to do this the old biopharmaceutical way, we would need something like thirty thousand people,” said Marchetti. “We think we can serve more patients with a few thousand people and the right technology underneath them.”
Rivelle was well positioned for it, having spent the last decade building a real data platform rather than a slide about one. The culture rewards curiosity, which shows up in hiring: the company attracts people who are comfortable adopting something new every quarter.
“I am grateful for the Ansel team and the time they spend with ours, because what we are actually building together is time back for patients,” Marchetti said.
03:14