Philippe Höij

Philippe Höij

@hoij.net

Our purpose is to strengthen human agency 💪 Get superpowers from structured information with dfrnt.com. Use explainable rules-based AI and easy decentralised version controlled collaboration to handle model-based graph-structured information.

485 followers63 following271 posts

Pinned

Using git-for-data time travelling with TerminusDB saved the day. Our static website builder gave up building our RSS feed today... Switched GraphQL endpoint to a specific commit, branching, merging, asserting correctness of the metadata. It was too smooth! Wrote a point on it 🚀 🤩 #buildinpublic

hoijnet.medium.comSaving the day with git-for-data revision control and GraphQL for content managementI sat down this morning to update our website and one of the plugins we use now had a field that was mandatory and inconsistently applied…

Just published up my honest review. I compare the reMarkable 2 and Paper Pro experiences. They are both eInk paper devices, the Paper Pro with colour. Both are great, but I chose not to upgrade in the end. The felt tip experience of the RM2 is irreplaceable. Hoping for a future felt tip upgrade

medium.comHonest review: The reMarkable 2 is better than the Paper ProI don’t regret my decision to return my Paper Pro. This is my breakdown of why I returned my Paper Pro within the 100 days.

Finally, it's out there. I have procrastinated this blogpost for too long. It shows how to implement taxonomies and master data management using the software I built. This is super hard to build correctly and in a trustworthy way using traditional databases and knowledge graphs. #buildinpublic

medium.comBuild advanced collaborative taxonomies using hypergraphsIn the enterprise you’re often faced with digital business information that is hard to process using traditional approaches. Far too many…

I'm so proud! We're hosting a 2 day workshop with 25 people (!) from one of our enterprise customers at our own offices in Lund, Sweden. So much to do and great results from the first day yesterday! So happy for the progress!

Please help? I'm looking for a specific quote on innovation, it might have been cited in Harari's books. Innovation breakthrough often happens when technology gets a user interface, making the innovation available to broader markets. The browser, app store, ChatGPT for NLP/AI. #buildinpublic

Who's my mysterious 400th follower in the #buildinpublic community? Amazing community growth over the past week here! Love newfound friends and builders out there 🚀

Just turned 400 ⭐️ Turned out it was @mzaremski.bsky.social that picked magic spot 400 🚀 Great question, $ or darkmode? And a great follow-up on making it effortlessly. But still, focus on the valuable customer problem to be solved! So good to see you in #buildinpublic Marcin!

Marcin Zaremski@mzaremski.bsky.social

This is the answer 💪 Or just use libraries like Radix that implements darkmode out of the box!

Graph-structured information + rules-based AI = awesome On Monday we show the power of unification with a user interface. Learn how to leverage old-school logic-based AI, reborn into the cloud age by combining knowledge graphs with a datalog. Join our webinar, Monday at 16.00CET 🚀 #buildinpublic

DFRNT.com@dfrnt.com

@dfrnt.com Joins @bsky.app: 💡We turn messy CSV files into structured knowledge graph objects! On Dec 9th, we’ll show how WOQL + Twinfox Platform empowers you to: 1. Combine CSVs into structure-enforced graphs 2. Structure your Information 👉Join us! us06web.zoom.us/meeting/regi...

Follow my @dfrnt.com to get the latest news about our cutting edge Twinfox content graph platform for business information. We provide structure-enforced knowledge graphs with a rules-based AI datalog logic engine and git-like version controlled collaboration. #buildinpublic

Roast my micropitch. With dfrnt.com, our purpose is to strengthen human agency 💪 We put deep tech in the human loop. Our platform enables easy collaboration and results from model-based structured information with explainable rules-based AI. Making the hard easy. #buildinpublic

We updated to put deep tech into the human loop into "we provide digital exoskeletons for an information-centric future." Which is much more aligned to what we do and what we want to accomplish!

The kinds of AI to use matters. How AI is used matters more. All stochastic AI have biases, such as ML and LLMs. We don't want competitive AI, that compete with us. You want COMPLEMENTARY AI, that makes us better. With rules-based AI, you have no bias. You use rules. Combine for good outcomes

Wow! 10x the followers over the weekend. The #buildinpublic community moved over in earnest. Seems things are settling down a bit since the weekend. I had 35 followers last week, now I have 333 followers 🚀 Honored, looking forward to contributing. We're launching our next webinar soon!

Failed hard yesterday. Was so nervous I failed our demo. The software did exactly what it should. But I didn't follow the right sequence. Got super nervous and stressed from not rehearsing enough. The webinar was about git-like content graph collaboration with push and pull. #buildinpublic

Once I pressed "fetch" and then "push", it did exactly what was expected. It pushed the content graph layers from the local engine to the cloud instance. But... that was too late. After the webinar. We have more webinars ahead, I promise to rehearse more. #failinpublic

More than 200 followers over the weekend. #buildinpublic on Bluesky is exploding 💥 And, our purpose to strengthen human agency 💪 seems to relate. I believe the only relevant approach is a human-centric one. Agree?

So tell us, what is the burning purpose for what you are building? Start with why! (Simon Sinek) How do you accomplish that? #buildinpublic

We doubled our paid customer base for our pro subscription. From September to November at dfrnt.com. With updated pricing and new features. Since then I have onboarded a team who are learning fast. It's super exciting times for my #buildinpublic project. RDF KGs, git-like collab, and datalog 🚀

What's the piece of #buildinpublic infrastructure you find most challenging to have to deal with? That thing, if made easy, would make all the difference. To build faster, cheaper, better?

It's Sunday. Reflect on the week ahead. What key steps to take? I will be thinking about what to ask help for from my ecosystem. What will move your things the most? Will you do it? #buildinpublic

I have a presentation to make on Monday. I'm torn. Help me choose. Just answer 1. or 2. Bonus points for why! 1. rules-based AI 2. logic-based AI #buildinpublic

Philippe Höij reposted

the following strategy helped me selling my b2b product #buildinpublic my product is full of sharp edges and missing quite some features to fit I was struggling to sell it I switched to sell paid pilots 🧵

Philippe Höij reposted

Bluesky, with a well curated follow list, has already replaced Twitter for me as the place to immediately see what's happening of relevance to me as a media producer in the independent media space

For you that code in typescript, how valuable would it be to have a modelling tool for building deep nested types and easily copy them? Idea is to be able to work with object definitions visually and be able to use a graph datalog engine to process objects as a graph with datalog. #buildinpublic

Philippe Höij reposted

Indiehackers that just joined BlueSky! Are you looking to follow other indiehackers building in public? I created a Build in Public starter pack for you! 🥳 If you want to be included: 👉 Comment what you are building 🔄 Repost to help others #BuildInPublic go.bsky.app/H7Ey2GR

Philippe Höij reposted

Replying to @hoij.net

so true! i once tried to sell my b2b product to 3 sales dudes they grilled me 😂 “Only from the thirtieth minute did you start asking meaningful questions.“ they told me to start asking questions way earlier during demo calls so that I can align the demo better with prospects needs

Sharing accurate data is costly and painful today. Humans are used as database engines in highly regulated industries. Students in the space industry. Financial accountants in finance. And others But, we're missing 400000 financial accountants world wide. Mainly because of three key factors 🧵👇

View full thread

Sharing accurate data is costly and painful today. Humans are used as database engines in highly regulated industries. Students in the space industry. Financial accountants in finance. And others But, we're missing 400000 financial accountants world wide. Mainly because of three key factors 🧵👇

View full thread

2. It is very hard to build and use rules engines; it's still not easy half a century later after prolog was invented in the 70'ies. Datalog seems to be gaining traction in the knowledge graph space which brings hope for a better future. But easy will take a while.

3. It is very hard to collaborate on high quality structured information, coders have “Github”, but the rest of us don’t... The thing is that information consists of facts, each fact has meaning, context and structure. That needs to be preserved when communicated. Github-like collaboration is key

Sharing accurate data is costly and painful today. Humans are used as database engines in highly regulated industries. Students in the space industry. Financial accountants in finance. And others But, we're missing 400000 financial accountants world wide. Mainly because of three key factors 🧵👇

1. It is very hard to assign computer-understandable meaning to information, it’s “just” data, without context or meaning. With the semantic revolution ahead, we're headed towards a model-based information-centric future.

Just released 1.6.1 with added support for CSV uploads to our Logical Studio in our datalog engine for knowledge graphs. We further improved our full immutable version history support with git-like content graph sync with a localhost open source TerminusDB instance. #buildinpublic

Giving up on Excel's VLOOKUP sometimes? Imagine you had the full logic-based AI power of a datalog engine to also process your CSV/Excel files. We go further. Augment the CSV information with model-based RDF graphs (not shown today). A common chore. Getting CSV files to play nice. #buildinpublic

Giving up on Excel's VLOOKUP sometimes? Imagine you had the full logic-based AI power of a datalog engine to also process your CSV/Excel files. We go further. Augment the CSV information with model-based RDF graphs (not shown today). A common chore. Getting CSV files to play nice. #buildinpublic

View full thread

See the explanation by pausing the video and reading =) Let me know your thoughts! What should the next video be about? RDF Content Graphs? Git-like collaboration for portable content repositories? #buildinpublic

Giving up on Excel's VLOOKUP sometimes? Imagine you had the full logic-based AI power of a datalog engine to also process your CSV/Excel files. We go further. Augment the CSV information with model-based RDF graphs (not shown today). A common chore. Getting CSV files to play nice. #buildinpublic

In the video, this is what we do: 1. Upload two CSV files unified by one common column 2. The column name Name is the same in both files 3. One file has the EmployeeID, one has the Title 4. We want to connect EmployeeId, Name and Title See the video to see the DFRNT Logical Studio in action.

We're just scratching the surface of what's possible with a full version controlled portable RDF-based knowledge graph that you can process CSV files with too. It's 100% compatible with the WOQL used by terminusdb.com DFRNT is a user interface and hosting environment endorsed by them.

terminusdb.comTerminusDB - Made for developers who see the value of dataGit-for-data - branch & merge, access & pull requests, all in the data layer. TerminusDB helps devs make data the star of their builds

I build dfrnt.com, a tool that enables you to skip the ORM. Build your graph information model visually, process your data using declarative logic with datalog logic and collaborate using version control and locally synced content graphs. End the struggles with Excel and the database nightmares

Is it foolish to #buildinpublic with enterprise-scale products? Who's joining the enterprise play in due time? What are the challenges you envision?

To solve today's messy data issues, we need to transcend to working with facts and information. Data is, by definition, without context and meaning. Information can be exported to data. But you need to add context and meaning for data to become meaningful. Start with a model. #buildinpublic