Why I built a chrome extension to save ASAP to Airtable

Martin Malinda
7 min readAug 12, 2024

Powersave 0.7.0 is published!” message appeared in Slack for me few days ago. I set up an automation that periodically checks the state of my extension in the chrome web store, compares it to state in Airtable and if it detects a new release, syncs the Airtable state properly and notifies me on Slack.

If you develop regular web apps, you can deploy new versions whenever you want. With browser extensions it’s different — each update needs to be approved by the store admins. It makes me slightly nervous that I can’t just instantly publish new code, for example to fix a bug that I just find, but on the other hand, there’s something natural about releases and versions — it creates a meaningful ritual and a milestone.

So I’m happy that I’m already at version 0.7.0 🙃.

The passion for meaningful data collection

I spent last few months putting this extension together. I grabbed a bunch of code from a previous project and just focused on utility. I wanted to create the most practical tool.

The premise was simple and unoriginal: I’m a fan of Airtable and I want to store information there in the most efficient way. Airtable is exactly the type of software that I like to work with and that I’m happy that exists. It’s flexible, it puts high priority on user experience, it’s a type of software that is worth learning. It doesn’t prescribe one specific way of doing things, instead it provides a general methodology that empowers you.

I think the idea of a single source of truth is fundamental. Store information once, and then look it up in different places, display it differently via a view or interface.

Having a good dataset in Airtable is a big win.

I think Airtable does a good job to display and categorize your data. All the filtering, sorting, views, interfaces. You get a lot of benefits for free if you store your data there instead of a spreadsheet.

But what if you just want to quickly add something? An advanced UI might not be optimal for this usecase.

If you visit the main UI you wait for bases to appear, you search for the right table, you search for the add record button… that’s a lot of time spent waiting ad switching between different contexts.

So I thought there’s space for an alternative little client that is optimized for just saving.

After all, we often are in these two different modes — data collection & data analysis — and often they don’t overlap.
As you’re browsing, working, researching, chances are you want to primarily save — you save websites you find, you add tasks, images, prospects, products, articles… you just want to save asap.

And then every once in a while, you want to analyze and at this time Airtable truly shines. You set up a perfect view or interface to see your dataset in the right way and you filter for the columns you need.

But there’s a challenge — the better the UI is for analysis, the more clunky it might become for saving. You just want to quickly add something and then you have to find the add button amongst a bunch of UI that is not relevant in that moment.

A form interface can definitely help here. That’s what they are for, you can craft one to optimize for the quickest most efficient input.

But even interfaces have their limits. I think there’s a potential to push it further.

That’s why I thought of Powersave. I just wanted a tool that is always available and allows me to push data to Airtable ASAP — to the right base and table with minimal effort.

The Airtable Web Clipper is great but I thought there should be also an extension that does not require prior configuration to work. An extension that you can install and use right away and where you have AI assistance available.

AI in the browser

LLM can be one way to make good data autofills work right away, without any config. So that’s what I focused a lot on with Powersave.

In theory you don’t need to have AI integrated directly to your browser. You could just send a URL to Airtable and a server would visit it in the background and exctract and save the data for you in its own pace. You could have an automation trigger like “Website URL is filled and fields X,Y,Z are empty” and then use Airtable AI or Make.com to do the fetching, analysis and updating.

But there’s a big benefit of making the AI work on the browser level:

  • As you have visited the website you already have the information available, it’s already loaded. Why waste time instructing a background process to load it again?
  • Browser extension knows in advance what information you’re looking for based on which fields you want to autofill. It can work with LLMs more efficiently by preprocessing the data and minimizing the amount of input for LLM to analyze. Browsers are naturally good at parsing and processing the information that is just present, they work with the DOM (Document Object Model).
  • Browser extension works with up to data browser state, just like you see it. There’s no need for bypassing of captchas and other anti-scrape protections. You can be sure that the data you see is going to be the data to be analyzed.
  • You can truly benefit from streaming. All LLM services can naturally stream the output. So in case of Powersave you will see the streamed output being autofilled into the form inputs ASAP, you don’t have to wait for the entire response to finish. You can see the output as soon as possible.
  • You can validate and adjust the data before you save. I think this is very important as you work with AI. Even though the models are getting better, they still hallucinate. If you can quickly review the autofilled data as you’re still on the website, it’s a big win. At this point as you’re visiting the website, you still have all the context available and you have the highest potential to catch any inaccuracies.

I have implemented a lot of this philosophy into Powersave but there’s still a ton of potential for improvement.

The “second brain” race

AI is getting better and better at data analysis. AI driven business intelligence will be of higher quality and more accessible to more businesses. But to gather insights you need to have the data well collected and stored in the first place.

Big corporations already know that and have their data science departments specialising in storing and crunching all the big data. But what about small and medium sized businesses?

They might not have the time or space to even think about it. They might not even tap into the benefits of AI business intelligence because the data are not available. Either they are not collected at all or they are distributed in different platforms, spreadsheets where they exist in total isolation.

Big companies can create special scripts to collect and categorize the data. They know what data they need and they invest time to craft solutions to gather them and store them. But a small business does not have the capacity to develop such solutions. That’s where generic solutions could help.

A good browser extension can make sure that once you see relevant data you can make sure it’s stored in a good format right away. If you see something of value, you save it, instantly. You don’t have to create a task or mental note to do it later, you just save right away to the right place. This should be technically possible.

What can Powersave do now

Powersave is my continuous effort to make this kind of instant saving possible.

At this moment it can:

  • Do AI analysis of the page based on which base and table you want to save to.
  • It can create one record or even multiple records for repeating information on the page (think tables, lists, grids…).
  • It can autofill single and multiple select fields, date fields, checkboxes… not just simple text inputs. The AI integration is tailored for Airtable.
  • It can also update existing records based on what URL you’re just visiting. This is useful for any kind of content management or content auditing. No need to switch tabs, just update data right away as you’re browsing.
  • It can do instant non-AI autofills via a simple algorithm that checks page meta data. It works well for SEO optimized websites and for well recognized data such as events, books, movies or products in e-commerce stores.

If you’re an Airtable user I hope you find it useful! With enough support I can push the solution further and deliver more features such as custom autofill configuration that combines AI analysis with the more classical extraction via CSS selectors. I think there’s still a lot more that can be done 👀.

To learn more about powersave:

Powersave.pro
📚 Documentation
🚀 Features & Releases

I’m building in public and I’m sharing the process on LinkedIn.

--

--