In this post, I’m going to tell you about my first product for #12months12startups challenge. If you need more context or you would like to know why have I started all this read the previous post “Just ship it”.
Giltch.com - is SaaS for removing silence or, you guessed it, glitches from video or audio files. The name of the service contains glitch which I found funny.
On January 15th a friend of mine wrote me a message: “Last week I developed a program which can remove glitches from the video stream”. I propose to build it as a SaaS under very strict conditions: - 1 week for the release (i.e. 22nd of January) - pay as you go, 10 minutes for free, after that 5$ for 10 minutes We chose to build a limited features product in order to fit the preconditions. In a matter of minutes, we agreed on a domain name, I registered giltch.com and we start to plan how it’s going to be done.
Even though there is only 2 of us we still need some sort of planning and task management. In order to have our spendings in control, I’ve registered virtual(or internet if you preferred) USD card and I use it for any billing operations. For the task management system, I choose Trello, which I haven’t use for a while and, boy, after bulkiness of Jira Trello is a breath of fresh air. After a domain name was transferred to CloudFlare, I registered at Zoho for the email accounting to have cool @giltch.com email. Also, I’ve registered a couple of email groups and connect one email to the Trello board using Zapier in order to see which user request is not processed.
We decided to build a website, with minimum functionality as we could get. A user should only have the ability to upload media, pay for the analys and download processed media. In one and a half hour, we have settled our vision. Face two face meeting deserve hours of chatting. At this point, we had 6 days until release.
Since my friend had a full time job our first coding session held on the weekend. It was a 2-day hackathon during which we developed most of the functionality and build a website. At the start, we wanted to go with AWS and Lambda, but unfortunately, it wasn’t clear for us how to manage file processing with s3. You can have a video stream with CloudFront though, but it had a delay before you can use a file and this was clearly not an option for us. We wanted to have a fast and robust program which ideally reduce a user’s waiting time to blink. So we moved to the VPS, I started to create infrastructure while my friend built a file-uploader. As an afterthought decision to try to do everything with AWS was a big time waste for us. On the other side, AWS Cognito saved us a lot of time since we had user management in exchange for 2 hours. A friend is convinced that next time it would take no more than 15 minutes. Hackathon launched at 1 p.m. on Saturday and finished at 6 a.m. on Monday with a 3-hours sleeping break and 2-hour breakfast break, so it was around 35-37 hours of a non-stop building. Two days later, my friend said to me that it was not only fun but also productive since we had everything set up, except for processing in Docker. I felt confident that we can finish on Tuesday evening. “Never have I’ve been so wrong”.
When it comes to finishing a product you just can’t afford to forget about the notorious Pareto principle: 80% of the result produced by 20% of the input(time). So to tell the long story short proper containerization of the processing took us a week. We were tired due to lack of sleep and discouraged because we couldn’t manage to ship in time. The analyzing was correct but the output file size was enormous - 60 MB of input turns to almost 2GB. The solution was obvious - have a good rest, sleep well and on the next day, we managed to have proper processing, without any reencoding with a file size under the source size. The development is finished on 27th of January at 4:30 a.m.
- NameCheap as a domain registrar
- VPS - Vultr for a ffmpeg to work
- CloudFlare for a free multi-subdomain SSL certificate. Cache, forwarding, DNS management as a bonus.
- Zoho, for @giltch.com emails
- Mongo, Express, Node.js as a server, everything in docker.
- AWS Cognito for user management
- Nginx as a reverse proxy
- Bootstrap 4 +
- Minio as file storage
- Sendgrid for transactional emails
- Paddle for billing
- martech: Google analytics, Hotjar, Freshchat
- Zapier to make me happier :)
As I mentioned, I have a separate card at my internet banking, so it’s really easy to calculate:
- $9.06 for a domain name
- $1.00 AWS authorization
- $5.00 - VPS for Vultr
As you can see we are really LEAN in terms of money spend. Time wise we spent around 120-140 man hours of a Senior developer, so it’s not that much either
0 because we haven’t launched it yet.
Feature developing is stoped as we want to get a sense of what people want and how they are using our product. We need feedback on the quality of the processing, price model and then we will proceed with the development. I’m a big fan of riskest assumption test, so the main question here: “What is the smallest amount of code we could write to test whether people use it or not”. So developing is finished, now comes the trickiest part, at least for me - marketing. I want to promote this project to as many innovators as I could possibly reach and a plan here is quite simple:
- to storm video editing forums and blogs, Reddit and HN.
- release on Betalist, ProductHunt and so on
- post to social networks, but I’m a big fan of #deletefacebook so twitter might be the only option here
- sleeping is very important, especially for the long rung, but hackathon was very fun and you should definitely try it
- develop something is only the first step, now it’s the time to show people our work and try to iterate using feedback
I’m very excited to see how it turns on and looking forward to your feedback. Writing in English is the new thing for me, so I appreciate if you contact me and tell me what you think.
If you like this story and want to know more subscribe to the mailing list below.