How Seed Health Scaled AI Responsibly

When Seed Health, the microbiome science company, prepared to scale its use of AI across the organization, it asked a question that reflects its broader commitment to impact: How can we adopt AI more thoughtfully and sustainably?

At Seed, sustainability is not just a department; it is a core value that guides how the company thinks, builds, and cares for the systems that sustain life. Those systems include humans, microbes, and the planet, whose health is interconnected. The company tracks and works to reduce its greenhouse gas emissions, designs packaging for reuse and recycling, and chooses responsibly produced materials.

As part of that commitment, Seed incorporated AI services into its corporate carbon inventory—but understanding AI’s footprint was only one part of the equation. As AI use across the company grew, Seed also wanted employees to feel empowered to use these tools with care and understand their potential impact.

Throughout 2025, teams experimented with AI on their own and shared tips in Seed’s knowledge base and at all-hands meetings (in Seed terms: All Microbes). After listening across the organization, leaders found that early adopters were using AI daily—and curiosity was spreading faster than internal support could keep up.

In April 2026, Seed formed a new AI-focused function to identify and build AI solutions with measurable business impact. To help employees use AI effectively and responsibly, Seed partnered with Clarinet to launch Project Accelerate, an internal initiative designed to build their skills and confidence.

“We have a strong culture as a team around speed and efficiency, balanced with responsibility for our brand and the planet. We knew that the best way to drive that same ethos in the context of our AI use was to educate our team on where AI can be best used, as well as how to develop processes and tools efficiently—and Clarinet was instrumental in working that thinking into our training,” said Lilla Cosgrove, Chief Product and Technology Officer.

Seed designed Project Accelerate around three paths, each building on the last:

1. The Superorganism Path gave all employees a shared baseline of confident, responsible AI fluency.
2. The Leadership Path helped leaders decide where AI could help the business most.
3. The Ambassador Path trained builders across the Seed ecosystem to innovate and deploy real AI solutions for specific business problems.

‍

All Seed employees took Clarinet’s AI Fluency Assessment, which measures how well people use AI, not how often. Rather than relying on self-ratings, the 20-minute assessment has people write prompts and make decisions in realistic work scenarios, covering technical skill, responsible use, and mindset, and flagging risky behavior. Each participant received a personal report for their individual development, and the AI Innovation Team received an org-wide report segmented by department and role to identify fluency gaps and pockets of strength.

Seed then offered Clarinet’s AI Fluency 1: Becoming an AI Superuser, a hands-on live workshop, to every employee. Participants practiced prompt precision, learned where AI tools break down, and each built an AI assistant (such as a Claude Skill) for their own role. At Seed’s request, Clarinet also built efficient token use into every session, which made thoughtful, purposeful AI use part of the training, alongside prompt quality and responsible use.

"Live training sessions can be hit or miss (especially when they involve a third party), but this was one of the best and most valuable that I've ever attended," one participant wrote.

The Impact:

• 94% found the training useful and intend to apply what they learned
• 84% of Seed employees completed the Fluency Assessment

‍

Next, leaders joined an AI Opportunity Spotting session built around the priorities Seed’s leadership had just set for the year. Leaders used Clarinet’s Five Frictions Framework to find where AI could speed up projects they owned or automate routine work to free up time. Each leader was then asked to run the same exercise with their team, so Ambassadors could start their apprenticeship with ideas already in hand.

“A key reason that we chose Clarinet out of a number of AI education vendors was because of their focus on immediate practical impact. The goal of Project Accelerate wasn’t to just broadly educate on tools or workflows; it was to implement a hands-on enablement approach, so our participants emerged from the program having built something that both delivered immediate business value and drove personal confidence,” said Lilla.

The Impact:

• VP-and-above leaders can now spot and prioritize high-impact AI opportunities

The deepest path was for Seed’s AI Ambassadors, who joined a four-week Agent-Building Apprenticeship. Across three workshops, Ambassadors moved from an idea, to a prototype system, to a deployable agent built with Claude Code or Claude Cowork.

Between sessions, Clarinet’s facilitator held two office hours and worked alongside Ambassadors in Slack. That let Clarinet spot blockers early, such as tool access and data permissions, and flag them to Seed’s team to resolve.

Ambassadors from across 11 functions built tools for real, recurring team tasks, 14 builds in total. Builds included:

1. Finance: A Slack command that answers vendor payment status questions
2. Employee Experience: One daily-refreshing dashboard that brings together people data from about 15 sources
3. MX Operations: Automatic twice-daily logging of customer loyalty interactions, replacing manual entry
4. Brand + Creative: Scaffolding that turns a creative brief into ready-to-design Figma artboards
5. Data Analytics: A Claude-powered layer that routes questions to the data team’s specialized skills
6. Growth: Plain-English input converted into structured setups for email testing
7. AI Innovation: A weekly pacing view that replaced a daily report

Ambassadors have now brought their Claude builds to their own teams and will soon be presenting them at company-wide showcases. The goal is a snowball effect, where one Ambassador’s build becomes a whole team’s habit. Seed is carrying that momentum into the new year through AI Labs workshops, where Ambassadors walk colleagues through how they built their tools, and shared repositories that make it easy for anyone to find, reuse, and build on what others have created.

“Clarinet’s expert guidance throughout this entire program has really been transformative for our team, and their knowledge truly moved the needle for us in ways I didn't think possible,” said Mallory Pendleton, Director of Education + Experience. “I watched my colleagues show up and build in ways I can safely say I've never seen before.”

The Impact:

• 10+ Claude-powered AI workflows or agents are in active use, replacing manual work like twice-daily data logging and daily reporting, and saving at least 10-20 hours a week
• 79% of Ambassadors intend to keep building and 86% report meaningful AI confidence growth.
• Both Monthly active Claude usership and Cowork usage increased by 200% after the completion of Project Accelerate.
• The program earned an average likelihood-to-recommend score of 9 out of 10.

For Seed, AI is no longer a side experiment. Scaling their AI strategy means continuing to build the judgment and skills to use it thoughtfully as the company grows. That work will continue beyond the initial training. 

This fall, Seed will begin tracking where AI is making a measurable difference in employees’ work, while Clarinet will reassess fluency across the company. Together, those insights will help Seed shape its 2027 AI enablement strategy and keep its approach to AI grounded in sustainable and responsible use.

Reflecting on the partnership, Mallory shares, “Working with Clarinet on Project Accelerate changed what people at Seed could do with AI. We went from experimenting in pockets to Ambassadors building tools that their teams now rely on, and some of those builds were things we couldn’t have imagined at the start.”