I gave a talk at FOSDEM this morning: https://archive.org/details/akkartik-2022-01-16-fosdem (20 minutes)

Main project page: https://github.com/akkartik/teliva

Hey Folks,

Nicolae, one of the cofounders of Clay (clay.com) here (along with @Kareem Amin). I’ve been a long time Future of Coding member (and largely lurker though I used to head to the in-person events and know some of y’all 🙂

We just launched Clay on Product Hunt (https://www.producthunt.com/posts/clay-8) & have also opened up the product to a wider audience. Appreciate any support y’all can give us!

The FoC community and exploring ideas in this space is very near and dear to our hearts, and at some point we’ll need to publish some more in-depth posts of all the directions we’ve explored and how the product’s evolved. We started as a more developer focused product (similar to tools like Parse & Supabase), but have over time skewed more to end-users and internal tools and automations. Today, the product is a spreadsheet-like database that has a few interesting features.

Tables that are aware of their data-type, and can automatically fill in missing information - e.g. People & Company Tables

• We observed that one of the main use cases of Spreadsheets is people building lists of things. In the case of search engines, similarly people tend to search for known “types” of content. Schema.org and their lists of Schemas: https://schema.org/docs/full.html is a pretty good list of the types of entities people end up actually looking for. We’ve explored what it could be like if you gave users tables that were more opinionated entity tables that could tap into public data sets, and improve over time to autocomplete - specifically for Co’s & People which happen to be common use cases for growth & recruiting. These entities exist in lots of different tools companies use, and also lots of different enrichment APIs, and so by mapping these structured entities we can make it easier to have one-click integrations where things are pre-mapped and can just be turned on.

Integrations that you can use as “magic-columns” in your tables (think Zapier & Airtable combined)

• One of the areas we’ve gone deeper is giving people an easy way to set up integrations as columns in their table - e.g. you have a list of companies and want to see which of them are already in your CRM - that should be as easy as just adding a CRM column to your table and then being able to easily create them in the corresponding tools. The known entity types should make this easy, as well as creating and enriching those entities back into your tools and databases.

Pull in data from anywhere into tables in a few clicks using sources

• Sources let you quickly pull in lists of data from your tools or places like Github, Twitter, etc. - so that you have some interesting data to get going in a few clicks. You could for example easily build an entire table of every website matching some Google Query, and then go and scrape those sites to check if they mention specific keywords - all with a few simple sources & integrations!

Mapping, Scraping & working with structured data from the web - a shared repository of website data mappings

• One of the ideas that we’re excited about is that Clay makes it really easy to map structured data on any webpage via our Chrome Extension - for example, going to a Zapier App description page and opening up the Chrome Extension shows a structured understanding of the data on the page. Ee let people describe how to find the page based on the URL structure and the shape of the page), and then define selectors to pull data from the page in a point and click way (you can also define them yourself if you know how selectors work). The idea is that once a type of page is mapped, everyone should then be able to use that definition rather than having to redefine it themselves. This way you can quickly re-create the database of most sites and then continue to add additional data - e.g. I did some quick data analysis on what types of apps does Zapier focus on and how highly are they rated on G2, Capterra and other sites, all in a few clicks).
• We think this approach has broader applications, but it’s already been pretty cool to see it work in action, and we’re looking forward to making it as easy as pasting a link to get the structured data you want from a site.

Happy to answer any questions y’all have! We’re also hiring so please get in touch if you’re interested!

Here’s a video that shows a lot of the core concepts

https://www.youtube.com/watch?v=V44XjAXvLlY&t=55s

Hey folks 👋 ,

I’m kicking off a new Papers We Love chapter focused on Computer Science Education in particular. If you haven’t encountered Papers We Love, it’s kinda a book club for academic CS papers. This chapter will be focused on papers about learning and teaching computer science.

Sharing here since I know there’s a lot of interest in the process of learning programming and how our language design and choices influence it. If you have a paper about CS education that you love, or would like to present, we’d love to hear your suggestions! You can submit them via our GitHub Discussions board. Would love to welcome y’all to present!

We are still ramping up, but are targeting the last Thursday of each month for live presentations, with paper discussions continuing async via the discussion board in between.

More adventures in sandboxing[1][2]

I've been working on a Zettelkasten app: https://archive.org/details/akkartik-teliva-2022-02-10 (video; 4 minutes)

One interesting insight here was that putting raw file operations directly in main makes the app easy to audit. You can inspect just that one function and give just it permissions to perform file operations on your computer, while denying them to the rest of the app including callees of main.

Anyways, I'd love to hear thoughts on this. I don't have experience with note-taking apps, and my UX skills suck. Let me know if you have suggestions, feature ideas, or if you'd like a private copy tailored for yourself. Situated software and all that.

Main project page: https://github.com/akkartik/teliva

[1] 💬 #two-minute-week@2021-12-26T06:52:58.098Z

[2] https://futureofcoding.slack.com/archives/CCL5VVBAN/p1644198434431509

For those following along on the development of Clerk, we're preparing to ship a feature where you can livecode HTML controls in the notebook view that have two-way binding to Clojure state atoms on the JVM side. This allows for moldable user interface development in the context of one's usual editor environment. There are a couple of examples in the images attached to this tweet: 🙂

https://twitter.com/jackrusher/status/1492110714588250118?s=20&t=kHaW82d1qYPo8pEEEJzQJw