Get Weaving: How visual-programming AI tools are redrawing the creative process

👋 Greetings earthlings,

A short rest from me today to bring you a really insightful think piece by Spark AI lead creative AI coach Matthew Maxwell on working with the creative AI tool Weavy,. 

Matthew is an artist and designer teaching machines to be creative, and humans to be a little more so. After a quarter of a century directing design for global clients, he’s now nearly a PhD in Creative AI. His work explores how imagination shifts when shared with algorithms, and what that means for the art (and artifice) of being human.

That’s enough from me - over to Matthew now to share what he thinks about how we work with Weavy - the core tool we use with our clients to build creative AI workflows in our AI Accelerator for Agencies.

Enjoy!

Emma
Co-Founder and CEO of Spark AI

P.S. Weavy was acquired by Figma a couple of weeks ago - if that’s not an endorsement I’m not sure what is!

Opening the Black Box

One of the problems with AI is you can’t really see how it’s done. Words, pictures, videos just… emerge. They seem to pop up out of the dark, fully formed, through processes so complex, and so opaque, it’s difficult to understand or explain how decisions or creative results come about.

It’s too mysterious to be reliable. (Which, of course, is part of its charm and power.)

But if you’re a visual person (and we all are, to a greater or lesser extent) it’s unsettling. Typically, we need to form mental maps to understand how and why changes happen. What connects cause to effect? Which way to the waterhole? We use spatio-temporal information to orientate ourselves mentally as well as physically. Otherwise we’d just blunder through life in the dark.

But AI works in the dark; transformers and algorithms and neural networks beavering away like spectral oompa-loompas who vanish when the lights are turned on. And just like Willie Wonka’s factory, we can only see what comes out of the gate, not how it is made inside the ‘black box’ of AI’s manufacturing.

But imagine if you could crack open the box and make AI’s inscrutable workings visible. If you could see the pipelines and junctions, switchboards and stockrooms. Would that help us better understand how AI thinks? If we had a map of AI’s thought processes, would that help us work together more creatively?

An intelligence orchestrator

Visual or node-based programming has been around for a while: Unreal Engine, Blender, VVVV and others have seen the value of a diagrammatic UI. Visual tools for visual people. So, in a way, it’s surprising Generative AI models have taken so long to adopt it.

I’m glad they have. The techie-bar has finally lowered just enough for me to get over it. Despite working in interactive design for over twenty years, I’m a coding doofus. I can’t tell my } from my ). But making the programming visual has made the difference. So here’s a speedy overview of what you see when AI turns the lights on. It’s a crowded field, but I’m most interested in Weavy.ai.

In Weavy, you drag a node onto the infinite canvas. It hums into place, a square representing a model: Runway Gen-4, perhaps, or Flux. Another node follows and links to another. Soon you’re constructing a cognitive road map: signals moving down thin wires, information intercepted, rerouted, transformed. Each connection adds intent; each edit has consequence.

Less a content generator, more an intelligence orchestrator. You chain together different AI engines and Weavy handles the coordination in the background, building pipelines that map out the creative process.

The stuff we do intuitively, Weavy draws as a picture. Pictures, famously, speak a thousand words. But Weavy reminds us how a thousand words can also paint a picture. It’s reciprocal.

So for designers and creatives it’s a potentially groundbreaking way of interacting with AI systems: the power and subtlety of natural language is exposed – the connection between words, and the pictures they paint, is laid out like a dissected frog.

This is radical. Because many, if not most, of us have lost the ability to read closely. We’ve forgotten how expert we are at using language.  Node-based pipelines reveal the relationship between language and image. Change “gleaming” to “glowing” in a prompt and the image changes too. Suddenly, we can all be programmers – if we’re prepared to slow down a bit and use the most powerful programming language of all: our mother tongues.

But did I mention?

A few Caveats though – in case this should read too much like a sales pitch. Like any such platform, Weavy is in ‘perpetual beta’. As a user, you’re helping to shape its development. It’s an evolving organism. So anything I say now may well be irrelevant in a month.  Nevertheless, here are a few watch-outs worth thinking about.

1. Watch the Clock 

Weavy’s credit-based system keeps costs flexible, but it also forces users to think like accountants. Every experiment burns credits. Every mistake has a price. The creative process, messy by nature, becomes shadowed by the quiet pressure of resource management. It’s hard to be spontaneous when you can see the meter ticking.

2. Reliance on the Cloud 

Because everything happens in-browser. Weavy’s performance depends entirely on the network. Lose connection, lose progress. Heavy pipelines can slow to a crawl if your internet falters. It’s the kind of vulnerability that makes professionals uneasy. Particularly those working with large video or 3D files.

3. Limited Transparency of Third-Party Models 

Weavy integrates multiple external AI models, but not all of them are equally transparent about training data, rights management, or licensing terms. For agencies handling commercial work, this raises questions: who actually owns the outputs? And can every model be used safely for client projects?

4. Still Finding Its Ecosystem 

As a new entrant, Weavy hasn’t yet built the plugin ecosystem or community that older creative tools enjoy. There’s no decades-old forum culture, no vast library of tutorials. Early adopters must rely on official documentation and instinct — fine for pioneers, frustrating for production teams on a schedule.

6. Not Yet a Replacement for the Classics 

Despite its power, Weavy won’t make Photoshop, After Effects, or Blender obsolete. It complements them. Complex finishing still demands those established workhorses. Anyone hoping for a single, total replacement will be disappointed.

Get Weaving

But on balance, the opportunities outweigh the threats. AI-Agent aggregators like Weavy offer a whole raft of advantages to design businesses. Not just in the studio – in the strategy, leadership and operational areas too.

There’s a seismic shift taking place – are you ready?

I lead our AI for Creatives programme which is part of Spark’s AI Accelerator programme for Agencies, teaching creative teams to build practical AI workflows. If you’d like me to send you more information on what we cover then let me know by pressing one of the options below →

Have a great week!

Matthew

Lead creative AI coach, Spark AI

This article was written by Matthew Maxwell – an artist and designer teaching machines to be creative, and humans to be a little more so. After a quarter of a century directing design for global clients, he’s now nearly a PhD in Creative AI. His work explores how imagination shifts when shared with algorithms, and what that means for the art (and artifice) of being human.

About Spark AI

Spark AI empowers creative and brand leaders turn AI curiosity into confidence through structured training and business transformation.

We have worked with 60+ agencies running AI Fundamentals workshops and AI Accelerator programmes based on our # 1 bestselling book Shift - AI for Agencies.

Trusted by Oxford University Saïd Business School and backed by Innovate UK.

👉 Find out more 

Not a subscriber yet?