switchboard.ai

Not sure where all that time went but here we are, 2023. Happy New Year 🎉

We spent the majority of last year on polishing and adding features requested by folks in our community.

We also ran a lifetime deal via AppSumo.

This was an interesting exercise in that it really highlighted the types of problems people are trying to solve with automation in the marketing world. It also highlighted the level of support required to put together a nice solution. Generally speaking, people need more handholding than I initially expected with an API tool, but the plus side is you really get to understand specific use-cases and sometimes there is some magic in the details that you didn't appreciate previously.

In terms of features, we've added AI background removal and smart-focus/auto-crop to image processing, as well as a Slab-Text strategy to text elements. During the summer of 2022 we also moved all the code to a new stack which allowed us to trim a lot of code from the previous incarnation of Switchboard. Now things run more smoothly and it's easier and quicker to develop new features.

On the community side (https://swbrd.io/c) we have a small number of super loyal users and it's become our primary place to solicit feedback and share the roadmap. We hosted the community on Circle - the jury's still out on whether this is the ideal place but it's working fine for now.

I've been spending some time on features for Switchboard Canvas alongside planning a marketing effort which I hope to kick off soon. Simple features I wanted to add included support for generating QR codes and support for applying LUTs to images.

In addition, I reworked some of the layout system to include support for anchoring elements to other elements. This is especially important for Text elements as the text can be dynamic and often you want to anchor an element to the text and not just its bounding box.

Finally, I revised the pricing for Canvas. The Creator plan (1,000 API calls per month) is now $19.

Users looking to scale marketing or produce content automatically at different resolutions can opt for the Agency or Enterprise plans which include 3 resolutions per request and 5 resolutions per request respectively.

Since the last update, we decided on another pivot. Since we started, we've done a lot of R&D on concepts related to video and machine learning and have decided to concentrate our efforts on something a lot simpler than our original vision.

As part of the journey to the new product, we salvaged one piece of the puzzle and turned it into its own product: Switchboard Canvas.

Switchboard Canvas is a dynamic image generation tool. You make a template in the responsive designer and then use either our API or a no-code tool such as Zapier to create images, replacing the elements in the template with real-time dynamic content.

Anyone who wants to streamline the creation of assets for social media will love Canvas. The templates are responsive and you can create multiple sizes in a single API call. For example, you can create a 1920x1080 thumbnail for YouTube, a 1080x1920 image for Instagram Stories, and a 1080x1080 image for an Instagram post in one go.

You can also translate text into 70+ languages on the fly, too.

We've already released our Zapier integration. Now your image generation can be connected to the entire Zapier ecosystem -some 3000 integrations - perfect for automating asset creation and loading/save to and from online storage such as Google Drive or Dropbox, and automated scheduling with Buffer or SkedSocial.

Switchboard is a set of tools to help creators be more efficient whether they're working with video, audio, images or text.

Switchboard and our small team have been on the most ridiculous journey to get to this point but Friday we finally deployed something to a small group of testers that have been patiently waiting for our product.

Not so Minimum

My biggest fault with this project was having a somewhat misguided view of what the minimum viable product should look like - and since joining Indie Hackers I've noticed countless opinions around the subject.

Our team has spent the best part of 24 months working on a number of proof of concept projects that we hoped would come together to form an enterprise SaaS product. During that process we did some great work but I was convinced minimum viability meant showcasing how these tools could hang together. And that just meant continuing to build.

Thanks to IH

In no small thanks to the folks at Indie Hackers, we turned a corner and started to think differently. What if we just package up one of the tools and tested the waters with our waiting beta testers? Why not just innovate from a simple starting point and see how the product is received?

Of course this sounds obvious in hindsight but with the immediate goal narrowing so dramatically things suddenly seemed more achievable... and more exciting.

So here we are. After a long time and a lot of work we have something out in the wild. We started with a fairly comprehensive transcription and captioning tool for video. We have a nice roadmap now because we've done so much work up front; developing and releasing the tools should be a quicker, more exciting process. Thanks to the folks that waited so long for us!

"I've read so many times that you should solve your own problem, and so last week I sat down to think about all the problems in my life."

This is very true and is how the idea for Switchboard was born.

I was spending a lot of time producing content for another business but a large part of that time was spent on repeating tasks. I was editing video, branding content, cropping and resizing for social networks and posting to platforms that would ultimately distribute content for us.

Eventually I thought there must be an easier way to complete these 'last-mile' tasks so I started to envision a system where content would go through a series of steps that would apply certain modifications and outputting completed assets somewhere convenient.