The Nodes Are Coming…

The interface is changing. Instead of hunting through panels and menus, more work is moving onto a canvas of boxes and wires. The pieces are familiar. A node for an input. A node for a transform. Thin lines showing how data flows. This used to be the territory of VFX suites and the bravest Blender users. Not anymore.

Figma just validated this thesis with their $150-200M acquisition of Weavy, now rebranded as Figma Weave. The message is even clearer than predicted. AI features will not sit in a single prompt bar off to the side. They will live in graphs you can edit, share, and reuse. Adobe responded at MAX 2025 with a preview of Project Graph, their own node-based system for workflow automation. Two giants confirming the same direction. The trend is now reality.

Because AI work branches. You try a prompt, compare models, mask a region, add a color grade, then save a few variations for stakeholders. A node graph keeps that branching under control and keeps the steps visible. More importantly, it preserves the recipe. If you swap the input next week, you can re-run the same steps and expect the same look. Predictability stops being a luxury.

This shift matches the multi-model reality of current practice. With Figma Weave, you can route through Seedance, Sora, or Veo for cinematic video. Switch to Flux or Ideogram for photorealism. Apply Nano-Banana for precision control. All in the same workflow. You might pass an image through an in-house style model, then to a face-refiner, then apply a brand LUT. On another job, you replace only the middle step. A graph lets you do that without rebuilding everything. Provenance gets better too. A saved graph is an audit trail for how an image or clip was produced.

The power shows in real workflows. Feed identical prompts to Dall-E, Midjourney, and Stable Diffusion simultaneously. Compare results side-by-side. Chain the winner through additional editing steps. What used to require hours of manual coordination now happens in one canvas. As one industry observer noted, Figma is positioning itself as "the conductor's podium where multiple AI tools perform together."

Figma Weave is now publicly available, though it will initially operate as a standalone product before deeper integration into the Figma platform. The team from Tel Aviv, founders Lior, Itay, and two Jonathans will continue developing the technology while aligning with Figma's collaborative framework. This means real-time collaboration on AI workflows, where multiple designers can tweak a single node graph simultaneously. Version history for AI outputs. Shared AI assets in team libraries.

Adobe's approach differs. While they previewed Project Graph for workflow automation, their main MAX 2025 push centered on conversational AI assistants and partner model integrations across Creative Cloud. They're adding nodes, but not making them the centerpiece. Time will tell which strategy wins.

The competitive landscape is heating up. Perplexity acquired Visual Electric. Krea raised $83M. New players like FLORA are launching node-based AI canvases that prove this isn't just a big-company play. The interface pattern is spreading.

Save the graph with the project, not just the export. Name nodes and versions with intent, the way you name layers and components. Expose a small set of parameters and lock the rest, so teammates know where they are allowed to touch. Cache heavy steps so reviews feel fast. Treat prompts like components that can be versioned, not throwaway text from last Thursday.

Studios benefit from structure. Add short "graph reviews" to stand-up where the team walks through the pipeline and trims accidental complexity. Standardize inputs and outputs so graphs can plug into each other without friction. Agree on sizes, color spaces, and file types. sRGB vs Display-P3 should not be a surprise at the end of a sprint. Keep an eye on compute costs too. Parallel nodes feel magical until the render bill arrives.

Classrooms can fold this in without throwing out the old curriculum. Teach graph literacy next to layers and timelines. Ask students to submit a runnable graph with their work and a short note on three parameters they chose to expose. Swap the input during critique and see if the look holds. That single move encourages system thinking, not just one-off image polishing.

Interoperability will matter more than ever. If Figma and Adobe graphs cannot talk, expect demand for neutral export and import formats. This becomes especially complex as Figma now needs partnerships with OpenAI, Anthropic, Stability AI, and others relationships that could get complicated as these companies develop their own creative tools.

Model routing has become table stakes. Users need quick A/B testing across engines without leaving their workspace. A marketplace of small, audited subgraphs is already forming shadow harmonizers, safe-area croppers, brand tone filters. Teams want to share these utilities like they share design components.

Governance rides along naturally. Graphs make attribution and usage logs easier to capture. Every node is a record. Every connection is documented. Compliance teams will appreciate the transparency.

The bottom line remains simple. The canvas shows what you made. The graph shows how you made it. With Figma Weave live and Adobe's Project Graph in preview, process has become a first-class deliverable. The acquisition signals aren't subtle , this is Figma's largest deal to date, and Adobe's response was swift.

For creatives, this means learning a new literacy. Node-based thinking used to be niche, familiar only to 3D artists and VFX specialists. Now it's going mainstream. The learning curve is real, but the payoff is control. Real control. Not just prompting and hoping, but building repeatable, shareable, scalable creative systems.

For teams, it means new collaboration patterns. Designers can now hand off not just assets but entire generative workflows. "Here's how we made it" becomes as important as "here's what we made." Dev Mode already bridged design and code. Now Weave bridges ideation and production.

For the industry, it signals a fundamental shift. We're moving from AI as a feature to AI as a medium. From single-model tools to orchestrated pipelines. From black-box generation to transparent, editable processes. The interface change reflects a deeper change in how we think about creative AI. Start saving recipes, not only results. Build your node literacy now. The future isn't coming, it's already shipping in production. Time to wire up and get building.