This is a submission for the Postmark Challenge: Inbox Innovators.
What I Built
I built NoBS.email, a tool to solve a common pain in email inboxes: too much noise, "BS". We are bombarded on a daily basis with newsletters, bug reports, weekly analytics, Asana, GitHub and other SaaS notifications. Finding what you need in those emails and cleaning them up for OCD people like me is a waste.
NoBS.email lets you cut through the BS. If it can't be translated to data, it doesn't matter.
The concept is simple:
- You get a unique
@nobs.email
address. - You define "Collections," which are like database tables. For example, a "Receipts" collection could have fields like
vendor
,amount
, andpurchaseDate
. - You subscribe to newsletters or (auto)forward any email to your new, dedicated email address with zero BS tolerance.
- NoBS.email extracts the data with AI according to your rules and populates your collections.
Suddenly, your inbox isn't a chaotic pile of messages; it's a structured, organized database that you can view and export data from.
Demo
👉 NoBS.email
You can try it out right now:
- Sign up for your own
@nobs.email
address. - Create a database, this is a "topic" of your collections. Could be "work" or "ai" or whatever you want to collect data for.
- NoBS will auto create some collections for you to get started but you can modify, delete those and create your own.
- Forward an email and watch the data appear in your collections within a few seconds
Sample, read only demo account, if you just want to get a taste of what it looks like once the data comes through... Go to sign in, click on "sign in with password" and enter:
email: [email protected]
password: postmark
Here are a few screenshots of the platform in action:
Code Repository
The code is fully open-source and available on GitHub.
👉 janezgec/nobs-email
How I Built It
I have started web development about 20 years ago. For past few years, I was working as a solutions architect for a software agency and I didn't really have opportunity to create my own stuff like I used to. So when Postmark sent me an email about the competition - because we use Postmark extensively for various projects at the company and its awesome, I felt like this would be fun to participate in.
When I brainstormed about inbound processing, emails, I immediately thought of this dreadful inbox I wake up every morning to. There has got to be a better way to deal with that stuff. Email clients are great but the repetitive barrage of notifications, newsletters and alike are getting out of hand. I built this tool to free myself from this every day struggle and I hope you find it useful as well.
Tech stack
- UI: Preact + Tailwind/shadcn
- Backend: Astro
- Email: Postmark (both inbound and outbound)
- Database: Pocketbase
- Hosting: Coolify
- AI (email data extraction): Gemini Pro Flash 2.5 (through OpenRouter)
Challenges & lessons learned
Well, first, getting started, I was a bit rusty 😅
With the idea in my mind, I tested it out in LibreChat against a few different LLMs and prompts to see how effective it was at extracting JSON data from emails. Needless to say, it was doing great!
I already knew a lot about Postmark so I felt like that part is covered. Inbound processing was perfect for this and with Postmarks feature to use a custom domain, I ended up with a cool @nobs.email email addresses I could hand out to people.
In addition to that, I realized I could use email aliases (+ sign) to separate username on nobs.email with the database. So they could forward emails to: [email protected] for emails about AI and [email protected] for work related emails. This separation of data is implemented as "databases" in UI.
For frontend, I oversaw projects in React, Next.js and I felt like things are getting increasingly complicated and I am no longer sure if all the abstractions and tools are really helping getting things done. So being out of the frontend game for so long, I wanted to find a framework that felt natural. It was so bad that at some point I just thought to use EJS with express. Then I found Astro, which felt great. Good old serverside rendering, with interactive "islands".
Once I had a very early prototype in place, I realized I was leaning onto AI too much. I wanted for user not having to define collections and for AI to just figure stuff out. So you would just send emails and AI would structure and extract whatever it could. Maybe I could even threaten the LLM to get the job done? Well, that was painfully not working as well as I hoped. So I had to resort to users defining collections on their own.
Once that was done however, I liked it even better because it was much clearer what was going to be extracted and allowing user to tweak the scraping process to their liking.
Inbound email processing
Here is how the main part of the system functions:
- User sends an email to scrape to their nobs email address with database indicated i.e. [email protected] . john=username, ai=database.
- Postmark receives this email and calls webhook on nobs.email backend
- Webhook receives the data in neat JSON and parses out username and database from the "to" property, validates the user is there and database exists.
- Then, key email properties are saved from Postmarks payload to database for users reference
- htmlBody and (fallback) textBody properties from Postmark payload are used to extract data from the email contents using AI
- Data is inserted in relevant collections
Postmark
There is something about:
- Postmark
- Posthog
- Namecheap
- Hetzner
- ... probably some others but those come to my mind.
That share this honest, no nonsense approach to providing software, services to clients. Affordable to get started, very intuitive UI, always a pleasure to jump into and get things done. So many other SaaS tools, platforms feel like you need a PhD to use. It shouldn't be like that. Dealing with a platform also shouldn't feel like you have to hold onto your wallet like you are in some kind of shady touristic hot spot.
I was so relieved when I discovered Postmark a couple years ago and it has been my trusted choice for all things concerning email. Without a doubt, the most developer friendly, overall best email delivery service AND inbound email processing platform out there.
Thank you for this challenge, this has been a blast!
Top comments (0)