Evoke

We'll be launching Evoke on Product Hunt tomorrow! Would love to see everyone subscribe and see their support on launch day.

This has been brewing for 2 weeks now after our official launch outside of product hunt and making some adjustments for production readiness, so definitely very excited.

Please subscribe to get notified here: https://www.producthunt.com/products/evoke-2

After 4 months of hard work, we've finally launched our stable diffusion API!

For now, everything is very MVP, and customers may run into some errors. But rest assured, we're making updates almost every day.

In our current unoptimized state, our image generations already take on average 3 seconds and cost $1 per 200 images, so we're excited at what updates will bring!

Sorry for no updates here in the past couple of weeks, I've been busy with school and also working on Evoke with my co-founders.

If anyone wants to try it out, the website is: https://evoke-app.com/

We also have an active AI community on discord

Thank you everyone for supporting our journey!

If you want to see more build in public, feel free to follow me on twitter: https://twitter.com/TheRealEtch

The past week has been quite an interesting one. Beginning with the bad news, we have not had our website up due to finding more issues. We planned to have the website to be a tool for collecting emails before launch.

However, my co-founder said that the release of the actual product would be delayed even more than it already is if he took any time to get the pre-launch website ready.

For the record, the pre-launch website is already finished. However, he would just need to set up Mailjet. This is the part that takes the most time, so unfortunately, the website has still not been set up. If this goes on, we might just push out the pre-launch website without the email form.

On a more positive note, our discord has ballooned in size. Growing from around 100 members to around 500. This was quite a surprise to us, and I believe it's due to how we promoted it on Reddit.

Most users were upset about ChatGPT censoring or filtering their prompts, and we leveraged that frustration by informing them about an upcoming API we're developing in the future for BLOOM: an AI without filters or restrictions (though with worse performance than GPT-3).

I'm hoping that the email collections we missed out are more than made up by our discord server. I don't have too much experience with email marketing, but it seems people check and engage with discord more than email and set up more notifications for it.

Hopefully, it all pans out fine. We're behind schedule on releasing our stable diffusion API, but if any of you are curious and developing an AI art app yourself, feel free to join our discord: https://discord.gg/DsJXxcMmkC

I've also got a twitter where you can follow the project: https://twitter.com/TheRealEtch

So far, the past week or so, things have been going smoothly. My co-founder working on the technical side started hitting walls on getting Stable Diffusion set up on AWS.

What I worried about being easy as 123 being done by the average person and pricing us out of the competition was suddenly revealed to actually have a very high barrier to entry.

So of course I was frustrated things were taking so long, but I was secretly relieved as well.

As for the website, it had been bad foresight on my part not to have one early on. My co-founder had convinced me that having a demo site would make us look unprofessional, and that we should release all of it at once.

Luckily, we soon both realized that the power of email collection and building up hype was much more important than some vague descriptor of "unprofessionalism", and I put together a pre-launch landing page that will go up soon.

As for marketing, I'm continuing to do what I've done before; leaving advice and valuable comments on indiehackers, product hunt, and reddit, with a little self promo cherry on top at the end.

Our discord has grown from 0 to 100 members in a little over a month from this strategy, and my other social media seem to be growing at a steady pace as well.

Overall, at the pace AI is moving vs the speed we're moving, I'll admit I'm quite nervous!

Especially since I'll be returning to uni after Christmas break when my co-op term ends, and might have less time to work on the project.

But we prevail onwards.

Many people I've seen list out their plan and what they're going to do for their startup. They'll state "I'll have x done by z and so on".

However, it'll be weeks after their due date and their project is still not finished.

They'll have valid excuses too. Payment method wouldn't work, co-founder/employee was sick, took forever to solve this bug, etc, etc

This is known as the planning fallacy, which is when one assumes things will take shorter than they actually do, and I'll tell my story of how I realized I was falling victim to it and overcame it.

In the past couple of weeks, mostly the month of November, we had some issues with registering an AWS account, which was required for how we operate.

For some reason, the card kept declining.

Simple problem, right? Just ask Amazon and they'll have everything ready in a day or two.

Dead wrong. The whole process took us 3 weeks to solve, which was 3 weeks worth of time we could have spent developing our product. (In the end, it was because Amazon only accepts Visa, American Express, or Mastercard, and our card was neither).

I had fallen victim to the planning fallacy.

The simple and easy response would just be to plan earlier right? Well, yes, that helps you avoid being late for a deadline, but it doesn't help you complete your task any faster...

How I overcame it

The time had unfortunately already been lost, so there was nothing to be done about that. However, I did learn something, and it was the importance of finding ways to be productive even if an external problem is holding you back .

For the Amazon issue, I actually oversimplified how it played out. In reality, it was only my business card that threw up the error, but my personal card was fine.

So I gave myself the excuse, "I don't want to mix my personal expenses with my business expenses, so let's just wait for Amazon".

Sure, that's a totally valid concern to have, but for me, it was just an excuse to prevent taking action.

I wanted to have a valid excuse to not take action, and I got one, which delayed our work massively.

For others, it may take a different form. How many times have you thought to yourself or heard others say: "I'm waiting for x to process/download/respond, and I can't get anything done without x, so I'll just wait until it's done to take action".

There's always something you can do, especially as an entrepreneur. Website down? Take the time to put extra care and personalization into customer support. Can't get a crucial part of your product working no matter how much you try and awaiting an answer on Stack Overflow? Promote your product. Computer literally broken? Promote your product from your phone.

If there's one thing I learned, if all other options with regards to actually working on your project are cut off due to unplanned issues, then there is always promoting your product and building your social media.

It might not be top priority right now, it might even be low priority compared to getting the product to work, but it only matters that it's something.

When you finally get the issue resolved, you'll find the other areas you worked on in your downtime are much further along than if you just waited.

Always be doing something.

What I did next

The next time something like this came up, it was when we applied for a Visa business card with the bank. Rather than waiting another precious few weeks for it to come, we just used my personal card for AWS.

Furthermore, I was promoting my waitlist for the product on social media every day, and especially whenever I was waiting on something to process or verify.

It was really that simple. All I had to do was to embrace an action taking mindset.

When does it not apply?

However, despite what I've said, there ARE times when you should genuinely just take a break, where it's fine to do "nothing".

For example, if your children are ill, you're spending time with family, or if you've been foregoing your own personal health and putting off going to the doctor for some concerning symptoms.

Your health and your family still matter.

"Always be doing something" is a heuristic, not a hard rule.

But it's a damn good heuristic that will help you push through unplanned obstacles.

You can find more of by #buildinginpublic and insights on Twitter: https://twitter.com/TheRealEtch

You can also join my discord to get alerted for the launch of the project I'm working on: https://discord.gg/DsJXxcMmkC

Hello everyone, a different post from the updates. I wanted to share my strategy and experience for getting customer interviews through the month of October to December.

To give some preface, Evoke is an AI API that allows developers to run AI like Stable Diffusion and text generators like Bloom (GPT-3 alternative) for their projects and apps.

I actually started the process later than most; I was building the product first as an idea already before doing interviews, my mistake, I know.

But nonetheless, I was going to get some validation anyways since we weren't too far in and we didn't even have a website nor demo.

I had a basic strategy, create a form and message customers to schedule a no-strings attached non-promotional chat just for research.

I found Jotform.com to be especially helpful, since their free pricing option (100 submissions and 5 form templates) was more than enough for my interview needs.

I started out on LinkedIn. I'm building an app that mostly appeals to developers, so I mass searched for developers on LinkedIn, and when my monthly searches ran out, I used https://recruitin.net/ to search on Google for profiles.

Here's what I used for messaging users:

*Hi [name],

I’m doing research on difficulties programmers are facing to better understand them.

Would you be up for a 30 min chat?

You don’t need to prep anything, there’s no pitch, hidden agenda or trick questions

Got a few slots here: https://form.jotform.com/223026301072035.

Thanks!*

Do keep in mind that LinkedIn messages have a restrictive character limit, so you have to be succinct.

As for the form design, you can see the design by clicking on the link and copying it in Jotform.

I set availability to the current week or one week ahead. Not more not less. This allowed me to generate a level of urgency.

Results and what I learned
From this strategy. I got a total of 5 interviews in a week, with two not showing up. However, the ones that did show up were very poor in terms of learning the wants and needs of my customer base.

They were general programmers that had nothing to do with anything remotely similar to my product. They didn't even work in AI.

What I did next
I decided to try Reddit next and narrow down my focus. I joined some Stable Diffusion, GPT-3, and OpenAI subreddits and searched up "app", "software", or "program" into the search bar.

Lo and behold, there were dozens of people building apps and software with these AI!

And they were complaining about issues with them as well; a perfect ground for market research.

I usually commented on a few of their posts providing genuine advice for their apps. Give value = get value back.

Then I DM'd these people with a similar message, but I mentioned more of what my startup was about and also interest in current projects they were building due to having a higher character limit.

I got a much higher response rate, and even some potential customers excited to join my Discord, which I was building to prepare for the launch of the product.

However, when it actually came to showing up to the interviews, these people were far worse. I got 3 interviews total in a week.

However, the information I got from these was FAR more useful. They outlined specific issues they had with competitors and what solutions they were looking for.

So what did I learn?

• Quality is much more important than quantity for customer interviews
• High quality interviews are found from selective targeting - subreddits are a great way to narrow this down
• Give value first by building rapport with the potential customer through other avenues before asking for anything from them, even if it's just their time
• Reddit will throttle your DMing privileges if you do it too much as a new account to prevent spam. Post a few comments first and DM no more than 5 people a day. You'll know you've been throttled if your DM takes forever to load when you click on a new chat.

Overall, this was an interesting learning experience, and I hope it was helpful both in terms of advice and resources provided to any who read this.

The resource that by far helped me the most for getting customer interviews was this blog post: https://wynter.com/post/customers-for-interviews

If you're interested and want to see updates for Evoke, you can join my Discord: https://discord.gg/DsJXxcMmkC

I also post about Evoke a lot on Twitter, so follow me there as well: https://twitter.com/TheRealEtch

You can also just follow my product page here.

Thanks for reading and feel free to discuss in the comments!

Hello everyone! From here on, I'll post something new every week about Evoke.

To give a short summary, I started this project with a friend in September. He had some frustrations as a programmer at work with AI.

Despite giving him a high end work laptop, the AI he was running (InstantNGP) still took so much processing power that it took hours to generate results.

That was when we had the idea of making the process easier for programmers to allow them to simply access popular AI from the cloud with an API.

There were already hundreds of API like ours, but we found they all had exorbitant pricing, really bad documentation, or didn't host the full AI model.

From this, we decided to launch Evoke.

As of now, our website is not up yet, but it will be once we release. Follow us on twitter for updates!