I worked at [Vana](https://vana.com/) as designer and frontend engineer for a few years, helping them explore various product ideas in the crypto and AI space. In that time we built almost a dozen product prototypes as we navigated the idea maze of what secured, personalised data might look like in the age of AI.

One of those prototypes was [Portrait](https://portrait.vana.com/): a web app that provides text-to-image generations using a [GPT](https://en.wikipedia.org/wiki/GPT-3) large language modelbased on personal likeness. The original version of Portrait was released in September 2022 after 7 days of development. It took off and we’ve been iterating on it ever since.

```cv
gallery
{"layout":"hscroll","images":[{"src":"https://firebasestorage.googleapis.com/v0/b/maitake-project.appspot.com/o/pages%2FhOKbuyX8ieXPY6k0m7AP67qVBF83%2F4RMKhST9ch5MrHb9YDbJ%2F1c35039f-7a9a-43d9-b552-d8a184889297.jpg?alt=media&token=d078a27e-09dd-4433-bf07-6e4f8a677ea5","id":"1c35039f-7a9a-43d9-b552-d8a184889297","width":1440,"height":880,"filename":"video-poster-portrait.jpg","type":"image/jpeg","caption":"","border":false}]}
```

Late 2022, we also explored personal “daemon” chatbots based on the same personalisation data we had collected from Portrait. GPT-me was one such app from these experiments.

```cv
gallery
{"layout":"hscroll","images":[{"src":"https://firebasestorage.googleapis.com/v0/b/maitake-project.appspot.com/o/pages%2FhOKbuyX8ieXPY6k0m7AP67qVBF83%2F4RMKhST9ch5MrHb9YDbJ%2F5049cf28-4df8-4038-ae17-6ae20f47a862.mp4?alt=media&token=d62249e5-ade2-4d8a-80a4-093ebd99136a","id":"5049cf28-4df8-4038-ae17-6ae20f47a862","width":1728,"height":1080,"filename":"vana-gpt-me.mp4","type":"video/mp4","caption":"","border":false}]}
```

Working at Vana was fascinating for two primary reasons.

* The first is that we’re building products with a technology that is very new to consumer-level applications: AI. The two-fold conundrum, which Portrait has so far navigated, is figuring out how AI can be used within software products successfully, as well as how AI products may impact typical users unaware of the power AI has right now, let alone the near future.

* The second reason is that after spending my prior career in Australian and English cultures, joining the highly skilled and ambitious Vana engineering team was incredibly energising. Being San Francisco based, Vana’s company culture always included a willingness to challenge the status quo and take on hard problems with unusual speed.

Visit [Vana Portrait](https://portrait.vana.com/).

Vana

When building software products, we often hear two separate phrases with different meanings used interchangeably: iteration and prototyping. They’re not the same thing, but they’re related over time during a build process. The easiest way to see the difference is to map them on a [Hill Chart](https://www.feltpresence.com/hills.html).

```cv
gallery
{"layout":"hscroll","images":[{"src":"https://firebasestorage.googleapis.com/v0/b/maitake-project.appspot.com/o/pages%2FhOKbuyX8ieXPY6k0m7AP67qVBF83%2FuPVjhgHLIL1BrmHfiWOy%2Fd2fe23fd-03aa-49ec-a32b-54c9cb25d19b.png?alt=media&token=77127115-feff-495c-9eea-000008fc45da","id":"d2fe23fd-03aa-49ec-a32b-54c9cb25d19b","width":1442,"height":350,"filename":"hill.png","type":"image/png","caption":"","border":false}]}
```

**Prototyping** comes first, at the start of the hill. We’re trying to find the best method of solving for a particular intention within the project and use that as a tool to make better decisions. For example, do we show an onboarding flow as multiple choice radio buttons or do we rephrase the questions as yes/no and use a swipe left/right flow? If it’s the later, how responsive to touch will it be on a smartphone?

**Iteration** happens after the apex, when we’ve “walked up hill” to discover all the interrelated dependencies so that we’re able to decide exactly how we’re building the thing. Iteration is the descent along a singular path towards a release. No new requirements should be introduced. Instead, like any editing process, we’re probably shedding them as we nail down the existing details and make them work.\*

During a build process, we always hope not to discover anything that puts the project at risk. And that is the job of prototyping: discover technical or design possibilities and determine if they’re opportunities or trade-offs. If there is some essential discovery that puts the project at risk, we want it to occur during protoyping, if at all. But we would not want nor expect that during any iteration work. That would be a sign of not enough initial prototyping or even insufficient research prior to signing-off the build process.

It’s important to note that any project build phase should be aiming to release as soon as possible. Deferring features for later releases is the easiest way to stick to the timeline. This means we can also think of iteration via projects, not just within a project.

We can now generalise these methods together as specific tempo-based jobs: *Prototype to discover. Iterate to release.*

```cv
gallery
{"layout":"hscroll","images":[{"src":"https://firebasestorage.googleapis.com/v0/b/maitake-project.appspot.com/o/pages%2FhOKbuyX8ieXPY6k0m7AP67qVBF83%2FuPVjhgHLIL1BrmHfiWOy%2F4a9c7a93-76a2-406b-a8fa-66c82bcc7b93.png?alt=media&token=a0d00000-c57f-4b71-aecf-adc429ac4c11","id":"4a9c7a93-76a2-406b-a8fa-66c82bcc7b93","width":1442,"height":350,"filename":"hill-blobs.png","type":"image/png","caption":"","border":false}]}
```

## **Prototyping does not mean “design tools”**

Don’t make the mistake of thinking that because Figma has a “prototyping” mode that its your prototyping tool. You may want figure out how far you can get with the Google Sheet API before you add a database, and you can’t do that with any design tool.

And if it is a design hypothesis (of the visual, flow or interaction kind) that you want to test, then designing in code is as close as possible to [“the nitty-gritty material of the medium”](https://twitter.com/callumflack/status/1622380350562516992) you can possibly get. As [I’ve said before](https://www.callum.website/designers-should-code), what often looks perfect in Figma can look not at all perfect in the reality of the browser. Inspect your assumptions in the DOM.

## **Developing a spidey sense for discover-or-release**

While these two concepts are not the only two things that occur during a build phase, *discover-or-release* is a useful abstraction to distinguish between attitudes in decision making at different times during the process. Here’s a few ways to think of them:

* Prototyping is uphill. Iteration is downhill.

* Prototyping discovers requirements. Iteration binds requirements.

* Prototyping creates a flow of investment. Iteration creates durable stock.

* Prototyping pulls towards complexity. Iteration pushes towards simplicity.

* Prototyping diverges the possible. Iteration rounds the probable.

Which ever way you say it, the key question is: *are we trying to discover ways to solve a problem or are we trying to release a solution?*

This helps us—as a team—stay focused on doing the right thing at the right time to prevent unnecessary loss of momentum.

```cv
gallery
{"layout":"hscroll","images":[{"src":"https://firebasestorage.googleapis.com/v0/b/maitake-project.appspot.com/o/pages%2FhOKbuyX8ieXPY6k0m7AP67qVBF83%2FuPVjhgHLIL1BrmHfiWOy%2F17700561-5910-4947-9824-dbc5b4109ced.jpg?alt=media&token=0aea005d-b5d2-40b7-961b-3a2613e082c3","id":"17700561-5910-4947-9824-dbc5b4109ced","width":3840,"height":2160,"filename":"post-prototype-iteration-1920-1080.jpg","type":"image/jpeg","caption":"Never trust supposed simplicity before the process begins","border":false}]}
```

## **Simplicity on the far side of complexity**

> Simplicity is interrelationships that make it feel more like one thing. Complexity is interrelationships that make it feel like many different things.\
> **[Ryan Singer](https://twitter.com/rjs/status/1572573848046014465?s=20&t=qJF31ambTfg5uYMubOwuNw)**

Alfred North Whitehead said “the only simplicity to be trusted is the simplicity on the far side of complexity.” Simplicity is always the desired outcome for a software product because it represents a user experience that wins hearts (“delight”) and minds (“don’t make me think”). To get there—*to make the thing whole for a person using it*—the build process must traverse upwards through discovery to the fog of decisions before we can slash-and-burn, edit-and-cut, spit-and-polish downwards through the thicket of details.

The difference between iteration and prototyping

Web Engineer & Designer

Creativity starts with love and theft

Making something new doesn't happen by magic. An irresistible itch has to be scratched, love has to be found. An accidental essay about the missing triggers in the creative process, inspired by my years of collecting soul 45s.

The matter of taste

When business hinges on the user experience and how products make people feel, success or failure is often a matter of taste. But only when one understands taste as intuitive knowhow, not subjective preference.

Do the grid fins still fold in?

During a tour of SpaceX rocket-building facility, Elon Musk delivers an excellent distillation of his 5-step build process — question assumptions, eliminate excess, focus on essentials, speed up development, and finally, automate. A masterclass in problem priorisation that is widely applicable.

The instantaneous language of beauty

We rely on the pre-verbal language of aesthetics to sense our environment before we can think. We are naturally attuned to beauty because we innately know it represents a higher possibility of worthwhile information.

Iteration and prototyping are not the same thing but they're related over time during a build process. To determine which method you're using, ask yourself are we trying to test a solution or are we trying to release a solution?

Second cohort participant. Fantastic to be part of the initial "scene" around digital note-taking.

I worked at [Vana](https://vana.com/) as designer and frontend engineer for a few years, helping them explore various product ideas in the crypto and AI space. In that time we built almost a dozen product prototypes as we navigated the idea maze of what secured, personalised data might look like in the age of AI.

One of those prototypes was [Portrait](https://portrait.vana.com/): a web app that provides text-to-image generations using a [GPT](https://en.wikipedia.org/wiki/GPT-3) large language modelbased on personal likeness. The original version of Portrait was released in September 2022 after 7 days of development. It took off and we’ve been iterating on it ever since.

Late 2022, we also explored personal “daemon” chatbots based on the same personalisation data we had collected from Portrait. GPT-me was one such app from these experiments.

Working at Vana was fascinating for two primary reasons.

* The first is that we’re building products with a technology that is very new to consumer-level applications: AI. The two-fold conundrum, which Portrait has so far navigated, is figuring out how AI can be used within software products successfully, as well as how AI products may impact typical users unaware of the power AI has right now, let alone the near future.

* The second reason is that after spending my prior career in Australian and English cultures, joining the highly skilled and ambitious Vana engineering team was incredibly energising. Being San Francisco based, Vana’s company culture always included a willingness to challenge the status quo and take on hard problems with unusual speed.

Visit [Vana Portrait](https://portrait.vana.com/).

GPT-me

Quickly respond to questions buried in text by choosing what’s important and removing the rest

Replier

Web design & dev consultancy for brands, agencies and community across Australia and the US. Highlights include:

* [Kalaurie ecommerce website](https://kalaurie.com.au/). Design and development of several website versions (Shopify Storefront API then Shopify Hydrogen) over 6 years, focussed on expressing the Kalaurie brand experience without compromising on the customer journey. Read the [case study](https://callum.website/kalaurie).

* [The Library of Economic Possibility website](https://www.economicpossibility.org/). Design and development of an economics research tool for both professionals and enthusiasts. Intended to energise community conversations and, as users drill down, to work as a bi-directionally linked knowledge system. Read the [case study](https://callum.website/the-library-of-economic-possibility).

* Studio Round website. Website rebuild, CMS and performance strategy, and website management over a three year period 2019-2022. Read the [case study](https://callum.website/studio-round).

* [My website](https://callum.website/). Please check out further case studies on my site. It is also open-sourced [here](https://github.com/callumflack/callum-website).

Frontend Engineer & Designer

* Led design and frontend of "Portrait", a flagship image generation AI product

* Part of a small team that enhanced and maintained Portrait toward explosive growth: 0 to 455k users in 9 months

* Hired 4 team mates across the Engineering and Design teams

* Introduced new features and iterated existing features across Portrait in response to research and analytics

* Introduced the “Shape Up” method to the product team

* Introduced TanStack Query and Tailwind to the frontend stack

* Championed move to Linear for task management, establishing a single source of truth for engineering work

* Read the [case study](https://callum.website/vana)

Graphic designer

* Led the development of [Cleared](https://getcleared.com/) from initial commit through to its [sale](https://www.crunchbase.com/acquisition/lifemd-acquires-cleared--a1f894e0).

* Effectively communicated and translated founder-led ideas into technical requirements for the development team and to external providers

* Built and maintained a marketing website to funnel customers to a personalised allergy care system based on initial brand guidelines and website designs

* Design, built and maintained a web app to determine a customer’s allergy eligibility and provide monthly pharmaceutical care using a questionnaire and algorithm, integrated with a custom Stripe checkout

* Successfully hired and onboarded six team members, including an Engineering manager

* Configured door-to-door drop shipping with external providers via the customer database

* Ensured the web application and customer databases complied with HIPPA security standards

* Integrated customer re-engagement systems such as drip email campaigns

* Read the [case study](https://callum.website/cleared)

Design and development of web apps for healthcare services & products

Callum Flack

2025

2023