Study an AI-powered expertise launched at least viable product to assemble actual consumer suggestions and iterate quickly.
After we speak about agentic AI, it’s straightforward to default to summary conversations about fashions, prompts, and orchestration. However probably the most compelling tales I see are those the place AI unlocks one thing deeply human—creativity, instinct, and experience—at completely new velocity and scale.
That’s why I used to be excited to host Coloration Meets Code: Pantone’s Agentic AI Journey on Azure, a webinar that includes two Pantone leaders, Kristijan Risteski, options architect, and Rohani Jotshi, senior director of engineering. Throughout the session, Kris and Rohani shared how they’re making use of agentic AI to one of the crucial foundational parts of artistic work: shade—and the way an AI-ready database, Azure Cosmos DB, performs a central position in making that doable.
The problem: Scaling shade experience in a real-time, interactive world
Pantone is widely known as a worldwide authority on shade. For many years, their groups have mixed human experience, shade science, and pattern forecasting to assist designers and types outline, talk, and management shade throughout industries—from trend and product design to packaging and digital experiences.
However as Pantone defined within the webinar, translating that depth of experience into a contemporary, conversational AI expertise got here with actual challenges. Creating shade palettes is each time consuming and demanding to the design course of. Designers typically collect inspiration by navigating between instruments, shade pickers, and pattern reviews earlier than they ever land on a usable palette.
Pantone noticed a chance to rethink that workflow completely: What if designers may work together with many years of Pantone analysis, pattern knowledge, and shade psychology by a chat-based interface—and generate curated palettes immediately?
Introducing the Palette Generator: An agentic AI expertise
The result’s Pantone’s Palette Generator, an AI-powered expertise launched at least viable product to assemble actual consumer suggestions and iterate quickly. Reasonably than providing static suggestions, the Palette Generator makes use of multiagent structure to reply dynamically to consumer intent, conversational context, and historic interactions.
Within the webinar, the Pantone group described how they designed the system to incorporate specialised brokers—reminiscent of a “chief shade scientist” agent and a palette technology agent—every accountable for totally different features of reasoning, context retrieval, and response technology. These brokers work collectively to ship curated shade palettes that mirror Pantone’s proprietary knowledge and experience.
What stood out to me was not simply the sophistication of the AI, however the architectural self-discipline behind it. Agentic AI isn’t nearly fashions—it’s additionally about knowledge.
Why Azure Cosmos DB was foundational
On the coronary heart of Pantone’s Palette Generator is Azure Cosmos DB, serving because the system’s real-time knowledge layer. Azure Cosmos DB is used to retailer and handle chat historical past, immediate knowledge, message collections, and consumer interplay insights—all of that are important for responsive, quick, context-aware brokers.
As we did our analysis to search out the very best persistence storage, we explored totally different databases. What we discovered for Azure Cosmos DB was how straightforward it was to combine it into our techniques. We had been in a position to make our preliminary proof of idea with a couple of traces of code and retrieve all the information very, very quick, like in a couple of milliseconds.
Kristijan Risteski
Azure Cosmos DB was additionally chosen due to its scale, permitting Pantone to serve customers all around the world with quick knowledge retrieval.
This can be a important level. As functions shift from “doing” to “understanding,” databases should help excess of easy transactions. They should deal with huge volumes of operational knowledge, adapt as AI workflows evolve, and help superior situations like conversational reminiscence, analytics, and vector-based search.
Pantone’s structure demonstrates what it means to be “AI-ready.” Azure Cosmos DB supplies the scalability and suppleness wanted to trace consumer prompts and conversations throughout classes, together with analytics that assist Pantone perceive how prospects work together with the Palette Generator over time.
From textual content to vectors—and what comes subsequent
One other perception Pantone shared throughout the webinar was how their structure is evolving to enhance relevance, accuracy, and contextual understanding. Whereas the present system already helps wealthy conversational experiences, the group outlined subsequent steps that contain transferring from conventional textual content storage to vector-based workflows. This contains embedding consumer prompts and contextual knowledge, permitting for vector search, and enriching responses with deeper semantic understanding.
Azure Cosmos DB performs a job right here as nicely, supporting vectorized knowledge, integrating with agent orchestration, and embedding fashions deployed by Microsoft Foundry. This enables Pantone to iterate with out rearchitecting your entire system—a vital functionality when working in a fast-moving AI panorama.
Actual-world outcomes from agentic structure
Pantone didn’t simply speak about imaginative and prescient—they shared concrete outcomes from actual utilization of the Palette Generator. Based on the webinar knowledge, customers throughout greater than 140 international locations engaged with the software, producing hundreds of distinctive chats throughout the first month of launch and interacting in dozens of languages. The system noticed a number of queries per consumer session, indicating that designers had been actively experimenting, refining prompts, and exploring concepts conversationally.
Simply as importantly, Pantone emphasised how quickly they’ve been in a position to be taught and adapt. Immediate sensitivity, consumer conduct, and architectural tradeoffs round velocity, price, and reliability all knowledgeable ongoing refinements. Azure Cosmos DB’s flexibility made it doable to seize these insights and evolve the expertise with out slowing innovation.
Classes for anybody constructing agentic AI
Pantone’s journey reinforces a number of classes I see repeated throughout prospects constructing AI brokers on Azure:
- Agentic AI is inherently knowledge pushed. With out a real-time, scalable database layer, even probably the most superior fashions wrestle to ship constant, context-aware experiences.
- Suggestions loops matter. By capturing prompts, responses, and consumer interactions in Azure Cosmos DB, Pantone can repeatedly enhance each the AI and the product expertise itself.
- Flexibility is nonnegotiable. AI architectures evolve rapidly—from orchestration patterns to embedding methods—and databases should evolve with them.
What Pantone has constructed with the Palette Generator is greater than a function; it’s a blueprint for a way organizations can translate deep area experience into clever, agent-driven functions. By combining Microsoft Foundry, Azure AI companies, and an AI-optimized database like Azure Cosmos DB, Pantone is exhibiting how creativity and expertise can transfer ahead collectively.
As extra organizations embrace agentic AI, the query gained’t be whether or not they can deploy fashions—however whether or not their knowledge foundations are able to help real-time understanding, reminiscence, and scale. Pantone’s journey makes that reply clear: AI-ready functions begin with AI-ready knowledge.
