Weber Wong


Brooklyn

Projects:

  1. FLORA LIVE
  2. STOCHASTIC MAN
  3. FLORA
  4. TINTED MIRROR
  5. this is not a flower
  6. Creative Coding
  7. Misc.

Writing:

  1. Workflow-oriented

Beliefs


Workflow-oriented

August 2025
At FLORA, we’re trying to make something that hasn’t been done before: an intuitive workflow-oriented creative tool.

This is because we want FLORA to be intuitive enough for most creative professionals to use, but powerful enough to run scalable generative workflows. We think generative workflows are the future due to the creative agency it provides to the user to scale their ideas, but that future will not happen unless creatives know how to use a workflow tool. 

An intuitive workflow-oriented creative tool hasn’t been made before because of the very simple fact that it is very hard to make a tool both extremely powerful & generally intuitive. Having a low barrier to entry and a high ceiling of mastery has been a battle for toolmakers since we started making tools.

We’ve been grappling with this fascinating design problem at FLORA for the past year. Here’s what we’ve learned, and how we approach building the world’s first intuitive workflow-oriented creative tool.

The status quo of creative interfaces

 Typically creative tools fall into two UX paradigms: Sequential & Workflow.


Sequential: Layer-based tools like Figma, Illustrator, Photoshop, or Time-based tools like Premier, Aftereffects

Sequential tools let you create by altering & reordering layers (image editing) or time (video editing), but do not show the transformation steps explicitly.

These tools prioritize showing the current state of your work over each step you’ve done to get there. For instance, in Adobe Illustrator you see the current state of your creative work, but you have to hit Cmd Z to see what your creative work looks like at previous steps, then Cmd Shift Z to toggle back to the present.

The benefit of these tools is they provide more control at each step and are more intuitive for users to understand. The downside is you can’t save these steps into repeatable, scalable actions. As a result, your creative agency is limited to what you are willing to manually do over and over again.



Workflow: Node-based tools like TouchDesigner, Blender, Grasshopper, Max/MSP

Workflow tools show each transformation step explicitly as a node, and let you connect nodes together to chain together an entire creative workflow.

These tools prioritize showing the process over fine-tuned controls over each step. Once you set up a workflow, you can try different inputs and get many different outputs without manually doing each step over and over again.

While there are clear benefits, there are two main downsides: less control at each step, and higher barrier to entry.

Workflow tools typically have less control at each step because by prioritizing showing all the steps, the toolmakers naturally deprioritize focus on each step. 

The bigger issue is the high barrier to entry. Node-based tools have a higher barrier to entry because:
  • You need to think in data or technical terms (node-based tools transformations typically rely on the numerical tweaking of parameters) 
  • You need to think in a very premeditated manner, because you have to set up the full flow before running it to see the result. While there’s certainly a delayed gratification feeling here, it starts feeling more like (visual) programming rather than entering creative flow state.


This is the bipolar creative tooling world we live in. You have to choose between:

Quality: bespoke Sequential tools that force you to do every step by hand
Quantity: bulk Workflow tools that feel like programming, with limited control at each step

What do we do about it at FLORA?


Defining “workflow-oriented”

 We must distinguish between existing “workflow tools” as defined above, and FLORA as a “workflow-oriented” tool. 

FLORA is not a workflow tool, it is a creative tool. It just happens to be capable of building scalable workflows.

You should be able to build workflows in FLORA, but that’s just a part of the experience. You should also be able to 1) create the workflow intuitively, and 2) have high control over each step like you have in more Sequential tools.

It needs to be intuitive, because it doesn’t matter how powerful a workflow is if the user never builds it.
It needs to have control, because it doesn’t matter how many outputs you generate if all of them suck.
The creative experience of doing bespoke work in FLORA should be so good it doesn’t require the delayed gratification of creating a repeatable workflow to justify it. But that delayed gratification will be there anyway, because FLORA is workflow-oriented. As you do each step intuitively with high control like you’re using a Sequential tool, you naturally end up with a scalable workflow like you would have in a Workflow tool.

Current workflow tools have it wrong. They have put the cart (the workflow) before the horse (an intuitive workflow creation experience). We believe if we put the horse back in front of the cart, we can build a uniquely delightly creative experience.


Workflow-oriented principles

 Based on the above thinking, here are our principles for building FLORA as a workflow-oriented tool.

I. Make it intuitive

Instant Gratification
  • A big barrier to adoption for workflow tools is needing to think in steps, where you often need to build the whole workflow before you get a result - this relies a lot on delayed gratification to motivate usage.
  • To counteract this, we must provide instant gratification and immediate value at each transformation step so they continue building the workflow.
  • The goal is to have each step be so rewarding that they would use FLORA even without the delayed gratification of using the final workflow.

Simple primitives
  • We should not have infinite types of esoteric building blocks to create a workflow. Thinking in steps is hard enough, so we need to reduce the options
  • Our nodes should not map to technical primitives; they should map to how creatives think: text, image & video, rather than VAE Decode, CLIP, and other esoteric machine learning terms
  • One thing we can do better here is simplifying model selection - how are non AI experts supposed to know what model to use?

No technical / complex transformations
  • Our transformations are typically just a prompt where you are asking for what you want, rather than the complex parameter transformation of most node-based tools
  • As we add deeper control features, we must maintain this simplicity


II. Make it controllable

As mentioned, most workflow tools typically have less control at each step because by prioritizing showing all the steps, the toolmakers naturally deprioritize focus on each step.

This is less of a structural design trade-off, and more just due to lack of attention to detail given limited time & effort.

We will have this attention to detail, and will pour significant time & effort to ensure we have control at each step. Much of these control paradigms already exist in many Sequential tools, which we can reference and adjust to ensure it’s workflow-oriented.


III. Make it scalable

Once you intuitively and controllably do many steps in FLORA, you should naturally end up with a workflow: a chain of steps that you can reuse again. Making it scalable is about ensuring you can apply those same steps to a different input.

Once you get to “workflow stage”, we also have to ensure the experience of using workflows is both intuitive yet controllable.

Intuitive
  • Is it easy to run the workflow?
  • Is it easy to understand what each step of the workflow is?  
  • Is it easy to try different inputs?
  • Is it easy to see the resulting outputs?
  • Can I easily tell which inputs resulted in which outputs?
  • Is it easy to abstract away all steps to make it easier for someone else to use?

Controllable
  • Can I control running workflows (quantity, locking certain steps, changing different inputs)?
  • Can I have multiple inputs & outputs?
  • Can I alternate between different options parametrically within a single run?
  • Can I queue up multiple runs with different inputs at once?


Next steps

 We’re trying to build a new type of creative power tool that breaks out of the pre-defined norms of creative tooling. FLORA is not just a Sequential tool, and it’s not just a Workflow tool. It’s probably a bit of both, and something else entirely. This requires us to think first principles about what FLORA should be, which I’ve done at a high-level above.

These high-level principles provide some frameworks and terminology, but the actualization of these ideas will come from the day to day work of designing individual interactions, engineering delightful solutions, and listening empathetically to our users.

Luckily, I think we all love doing those things at FLORA. I look forward to inventing a new paradigm for creative interfaces together.