FLORA ships creative agent FAUNA on Vercel AI stack
FLORA deployed FAUNA, a long-running creative agent that handles parallel image generation and model orchestration, using Vercel's AI SDK, Workflow SDK, and Fluid compute. This unified stack eliminated infrastructure complexity and delivered 2x faster production cycles compared to managing separate tools like LangChain and Temporal.
FLORA on Vercel
2x faster to production with their generation system
Zero infrastructure debates after migration
50+ image models orchestrated
and its provides out-of-the-box primitives that power FAUNA's behavior (models, tool calling, and the interface layer for agent logic). AI SDKagent framework
adds durable orchestration to AI SDK's agent framework, so each step in the loop persists, retries on failure, and can run for any length of time without losing state.Workflow SDK’s
DurableAgentruns jobs cost-efficiently as the agent fans out into many concurrent tasks, which is critical for parallel, long-running image generations. Fluid compute
A seasonal fashion launch is a story, not a single frame.
Crafting that story is a process of exploration: It’s the same piece, worn by different models. The same pose, with different lighting and angles. The same set, with a background that shifts from glossy to gritty.
FLORA was built to make that kind of visual iteration available to anyone through a digital canvas. Their new creative agent FAUNA acts like a design partner, turning ideas into a map of creative directions.
Flora started as a node-based creative workflow canvas. For advanced users, that canvas is powerful: you can create steps and branches, adding detailed prompts at each stage to hone every detail of the images.
But the canvas also came with a tradeoff: it asked every creative to think like a workflow designer. You had to configure each state of an exploration with prompts and model choices, interrupting the creative process.
That’s why they built FAUNA.
The agent removes the setup burden, but keeps the full power of AI. Instead of starting with a blank canvas and a pile of example images, designers can start with their ideas. Users tell FAUNA what they want to create, like a campaign visuals, moodboards, or a lookbook direction, and the agent will pull references, choose models, and automatically generate variations to explore and refine.
Under the hood, FAUNA is a long-running, highly parallel agent. Alec Jo, Head of Applied AI at FLORA, leads the effort.
As Alec and his team were building it, they faced the same pain as their early canvas users, spending too much time configuring AI tech instead of iterating on the agent itself. That's when they started exploring Vercel.
If you use AI mainly for text-based tasks, it's easy to underestimate the additional demands of visual workflows for professional creatives.
Image and video generations can take minutes, and a single creative session can fan out into many concurrent jobs: different angles, backgrounds, styling directions, model providers, and “what if” branches.
Those jobs need to run simultaneously, will finish at different times, and still need to roll up into a coherent exploration path the user can navigate.
Early on, Alec's team used LangChain building blocks for agent primitives and considered Temporal for orchestration. It worked, but it came with a cost: two separate systems to maintain, two sets of abstractions to keep in sync, and too much AI plumbing to keep the product moving.
FLORA migrated to the Vercel AI Stack to simplify building and deploying their agent:
The point wasn’t just swapping tools for simplicity. It was moving to a single, integrated AI stack where the primitives, orchestration, and compute are designed to work as a system, not individual components.
FLORA's ultimate goal is to support all design work that isn’t UI/UX, and FAUNA is their first step in orchestrating visual ideation end-to-end.
But building FAUNA had an unexpected impact on Alec's team: it helped understand their users on a deeper level. In creative work, the process is the product. And the longer you can maintain a flow state, the better the outcome.
Just like new visual ideas require new cycles of iteration, new image models and capabilities launch every week, and now FLORA can take advantage of them without plumbing or infrastructure getting in the way.
From FLORA to FAUNA: a creative agent
Infrastructure makes the agent
The process is the product
Source: original entry ↗
More from Vercel
Follow Vercel to get its new changes in your feed and email digest.
OpenAI Decisions API now available on AI Gateway
OpenAI's Decisions API is now accessible through Vercel's AI Gateway with an OpenAI-compatible endpoint, enabling decision models to answer typed questions and return probabilities, choices, and scores for routing, triage, and guardrails use cases. Support is available across the OpenAI SDK, AI SDK, HTTP API, and CLI with the latest versions.
Timestamp attributes now supported in Vercel Flags
Vercel Flags now supports timestamp attributes for entities, allowing you to create time-based targeting rules. Use this feature to run limited-time campaigns, show content between specific dates, or target users based on registration date.
Glyph Cluster now available in stealth on AI Gateway
Glyph Cluster, a reasoning model for coding and long-context analysis, is now available as a stealth model on Vercel's AI Gateway for Pro and Enterprise plan teams with purchased AI Gateway credits at no cost during the stealth period. The model supports function calling, streams responses, and can be accessed via AI SDK, OpenAI-compatible APIs, and coding agents.