3M's journey with AWS Quick

3M is a global science and manufacturing company with around 63,000 people, more than 135,000 patents, and over 110 manufacturing sites worldwide. It works across three businesses: safety and industrial, transportation and electronics, and consumer. As the company likes to put it, wherever you are in the world, you are about 10 feet from a 3M product. That scale makes for a complex data picture.

3M runs its data and AI work top down. It picks a strategic objective first, then works out how data and AI can drive the business outcome. One such program is 3M eXcellence, which targets sales and marketing across the customer lifecycle, from acquisition to growth to retention. Retention stood out as a clear opportunity. On its Q1 2025 earnings call, 3M said it had begun deploying a predictive analytics tool to flag at-risk accounts and address the main drivers of customer attrition.

Over the year, the team learned that an insight alone doesn't change an outcome. The work had to reach behavioral interventions, the moments where sales teams change how they work with customers. The most frequent of those moments is the one-on-one coaching conversation between a sales manager and a rep.

The scale was significant: 1,700 sales reps and more than 100,000 one-on-one conversations a year. When 3M mapped its sales process, the industry benchmark held true. Sales professionals spend more than 30% of their time gathering data and customer information from different sources before they can use it.

3M framed the program as four stages, each with its own kind of AI. Insight comes from AI/ML models, such as the one that flags at-risk accounts. Coaching uses generative AI to turn those insights into material a sales leader can use. Planning uses generative BI, so a rep can ask questions of the data in plain language. Action uses agentic AI to carry out the next step.

To decide where AI fits, 3M's product managers start with the whole business process rather than a single use case. They lay out the value stream for the business area, then sort every idea into three lanes of AI value: workflows to accelerate, insights to generate, and knowledge to aggregate. Then they rank the ideas by impact. In the sales process, the time reps spent gathering data was the opportunity that stood out, so 3M focused there first.

3M had built chatbots before, but each one lived in its own application, so sales teams switched between tools to get an answer. Working with AWS as an early partner on AWS Quick, 3M set out to replace that with one experience.

Two Quick capabilities made that possible. Action connectors let Quick bring data and information from 3M's systems together into one answer. The Quick browser extension puts that experience next to whatever the rep is working on, and what it surfaces changes with that context. A rep looking at an account in the CRM system and a rep reading product data on a website each get insights for that page, without opening another tool.

"What made the solution so valuable for us was the browser plugin, because now we had one unified experience where we could bring actions together." — Nithin Ramachandran, VP of Data and Analytics, 3M

3M designed the experience around two personas, the sales rep and the sales leader, and built five assistants in Quick for the work each one does. Material-science solutions are complex, so a rep's preparation for a customer meeting is intense, and a leader coaches many reps at once.

• Sales Plan Effectiveness: Helps a rep build an account plan and gauge how effective it is.
• Customer Meeting Prep: Researches the customer and the opportunity before a meeting.
• Sales Analysis Assistant: Analyzes large amounts of data across systems, including during a customer conversation.
• Product Assistant: Compares products so a rep can advocate the right solution.
• Sales Coach: Brings the work of many reps together so a sales leader can coach each one individually in their one-on-ones.

The assistants draw on the same connected systems and live in the one browser extension. A rep's planning, research, and analysis feed straight into the leader's coaching conversation. 3M built the assistants with Quick chat agents, Spaces, connectors, and Apps, and reps can also use them in Quick Desktop.

3M treats Quick as one layer of a larger AI stack, as Figure 1 shows. At the base sit a data lakehouse and a knowledge graph. Above them, an AI studio provides shared resources: developer tools, a model catalog, prompt management and guardrails, AI data stores, security, observability and FinOps, and AI infrastructure. Platform teams turn those resources into product blueprints for analytics, machine learning, generative AI with retrieval-augmented generation (RAG), and agentic AI.

Teams build on those blueprints with pro-code, low-code, or no-code tools. Amazon Bedrock supports the pro-code path, while Quick gives business users a way to put those blueprints to work with no technical setup, and it delivers agentic experiences such as the browser extension. 3M's earlier generative AI work ran on Amazon Bedrock, and adding Quick has been a big accelerator. Platform teams create a blueprint once, and teams across the company use it through Quick instead of each building its own application.

"Quick has become an integral part of our AI stack as we start looking at business outcomes, agentic products, and agentic experiences that we want to drive."— Nithin Ramachandran, VP of Data and Analytics, 3M

Where 3M is today

3M has kept expanding its use of Quick since re:Invent 2025. In April 2026, Amazon reported that Quick saves 3M's sales reps more than five hours a week by streamlining how they gather information for customer meetings. 3M also used Quick Desktop during its preview, before it became generally available in September 2026.

The coaching program is showing results as well. On its Q4 2025 earnings call, 3M said it had tripled the number of structured reviews between sales managers and sales reps.