Update on Idea Bakery & the new work cycle

Shavin Peiries May 14, 2021 3 min read

May 14, 2021

When I first started building idea bakery, I released a version that was just good enough to get some feedback.

I sent it to a trusted few I thought would've found it useful & Tweeted if anyone was interested in early access. This is the feedback I received for the first launch:

Since then, I've figured out what was working and what isn't.

I shared the idea with my co-founder Mayun, and he thought it was interesting because Idea Bakery helps with lateral thinking. An area of interest that Mayun's passionate about.

So Idea Bakery is coming under the umbrella of Very Bad Wizards products. We are actively spending resources to help get it to a point where it's useful for independent creators.

Mayun shaped 2 pitches that would help us solve some of the core challenges with Idea Bakery.

The premise of Idea Bakery was to develop ideas that you're interested in talking about on Twitter using your experiences as lenses to talk about them.

For, e.g. Talking about Product Development [an Interest] through the lens of Failure [an Experience]. Then during the writing session, you develop ideas that you can think of for the given words.

The problem with it is the language for what an interest & experience is. It was plainly put, confusing. Sometimes an interest could also be an experience, e.g. Designing.

The starting point of an ideation session is gathering the right ingredients (just like baking) to get ideas. But without the correct language to guide the user to use the system in a meaningful way, it's just garbage in, garbage out.

So we decided to take out interests and experiences and call them ingredients. By definition, ingredients are keywords derived from experiences or interests when acting or reflecting. Although interests and experiences aren't entirely going away. We might use it for onboarding to make it easier to frame the right ingredients to get started baking.

The other point of resistance is whether users could develop ideas when they have two words staring blankly at them. The immediate answer to this was to use prompts to help nudge them along.

I wanted to use GPT-3 for this. So I tried emailing Greg Brockman, Open AI's CTO & re-applying to get beta access to Open AI's API. Unfortunately, it hasn't worked because I don't access to it yet. But thanks to Mayun's long time strategy mentor Mark Pollard, there was an easier way to accomplish what we intended with prompts.

When we distil ideas down to ingredients, we can use prompts like these (X & Y are the ingredients):

• X as Paradox of Y
• Contrast X to Y
• Compare X to Y
• Instruct X through Y
• Observe X through Y
• X vs Y
• Before X, After Y
• Now X, Next Y
• Formula for X and Y
• Case Study of X with Y

The purpose of the prompts isn't to give them the answer. Instead, it's about creating an initial spark where an idea can grow from.

With that, we're off to build it. Since we work in cycles, this one is 3 weeks long, and we've broken each week into what has to be accomplished.

On the Build track:

Week 1

• Improve the login flow
• Sign in with Twitter over username and password
• Hookup Hotwire to speed things up
• Start adjusting the structure to include ingredients instead of experiences and interests
• Onboarding flow fix (rules of the game)
• Highlighting to add new ingredients

Week 2

• Prompts (Mayun's going to shape this better in Week 1)

Between Week 2 and Week 3

• Keyword analysis using either an open-source platform or Monkey Learn to point out new ingredients within the ideas
• Update on landing page
• Buy me a coffee integration

On the Shaping track:

• Gamification

What I'm really hoping for by the end of this cycle is to launch a version that won't leave a wrong impression. I know that it's a possibility and could happen, but what's more important is to have a stable version that gets the job done so that I can talk about it more confidently.

Special thanks to Mike for donating $1000 and, more importantly, believing in us.