Arto Bendiken
@artob.bsky.social
CTO https://Haltia.AI. Autodidact, coder, cypherpunk, entrepreneur, hacker, maker. ex-NEAR, ex-Aurora, ex-Dydra. Also at https://x.com/bendiken.
My 1-minute demo at #SEMANTiCS2025 of the ASIMOV personal intelligence layer unlocking and linking the data silos in your life, here showcasing some initial utility for frequent conference attendance
September 5, 2025 at 2:16 PM
My 1-minute demo at #SEMANTiCS2025 of the ASIMOV personal intelligence layer unlocking and linking the data silos in your life, here showcasing some initial utility for frequent conference attendance
Presenting ASIMOV at #SEMANTiCS2025: a personal intelligence layer linking all the data in your life, built on an open-source platform for trustworthy neurosymbolic AI
September 5, 2025 at 9:42 AM
Presenting ASIMOV at #SEMANTiCS2025: a personal intelligence layer linking all the data in your life, built on an open-source platform for trustworthy neurosymbolic AI
What's our term for that satisfying feeling of arguing the AI to submission, prevailing on points of nuance that it lacks?
It seems like German should have one. Less Besserwisser and more machine whisperer. Would that be Maschinenflüsterer?
It seems like German should have one. Less Besserwisser and more machine whisperer. Would that be Maschinenflüsterer?
June 27, 2025 at 11:14 AM
What's our term for that satisfying feeling of arguing the AI to submission, prevailing on points of nuance that it lacks?
It seems like German should have one. Less Besserwisser and more machine whisperer. Would that be Maschinenflüsterer?
It seems like German should have one. Less Besserwisser and more machine whisperer. Would that be Maschinenflüsterer?
June 26, 2025 at 9:28 AM
Watching Bruce Lee stunts with my preschoolers as part of their morning homeschool today, we did a double take on that clip of Lee playing ping pong with nunchucks
June 18, 2025 at 8:36 AM
Watching Bruce Lee stunts with my preschoolers as part of their morning homeschool today, we did a double take on that clip of Lee playing ping pong with nunchucks
I feel some chagrin correcting my preschoolers daily now on their beautiful, logical proto-English.
Just this week: broked (broke), catched (caught), eated (ate), seed (saw), sitted (sat), sneaked (snuck), taked (took), teached (taught), winned (won); and many more, already forgotten.
Just this week: broked (broke), catched (caught), eated (ate), seed (saw), sitted (sat), sneaked (snuck), taked (took), teached (taught), winned (won); and many more, already forgotten.
June 15, 2025 at 7:47 PM
I feel some chagrin correcting my preschoolers daily now on their beautiful, logical proto-English.
Just this week: broked (broke), catched (caught), eated (ate), seed (saw), sitted (sat), sneaked (snuck), taked (took), teached (taught), winned (won); and many more, already forgotten.
Just this week: broked (broke), catched (caught), eated (ate), seed (saw), sitted (sat), sneaked (snuck), taked (took), teached (taught), winned (won); and many more, already forgotten.
After several rounds around the house, farmer's carry with two heavy kettlebells feels like good prep for that important Finnish summer tradition: akankanto (wife-carrying)
Though my wife does weigh less than the program minimum, so I might need to add a rucksack still.
Though my wife does weigh less than the program minimum, so I might need to add a rucksack still.
June 14, 2025 at 4:00 PM
After several rounds around the house, farmer's carry with two heavy kettlebells feels like good prep for that important Finnish summer tradition: akankanto (wife-carrying)
Though my wife does weigh less than the program minimum, so I might need to add a rucksack still.
Though my wife does weigh less than the program minimum, so I might need to add a rucksack still.
Really getting that AltaVista feel with Google these days what with even half the search results being ads.
June 14, 2025 at 3:52 PM
Really getting that AltaVista feel with Google these days what with even half the search results being ads.
The strangest aspect of being a fast sleeper is daily being the last to bed and the first up, both by margins of several hours. For the kids it creates the illusion and family joke that daddy never sleeps at all.
June 9, 2025 at 4:42 AM
The strangest aspect of being a fast sleeper is daily being the last to bed and the first up, both by margins of several hours. For the kids it creates the illusion and family joke that daddy never sleeps at all.
The Emirates long haul (15h45m) from SF to Dubai is quite the scenic route, what with overpasses of Greenland, Iceland, and Denmark along the way.
Not quite the shortest great-circle distance, though, given current geopolitical circumstances.
Not quite the shortest great-circle distance, though, given current geopolitical circumstances.
March 29, 2025 at 9:33 AM
The Emirates long haul (15h45m) from SF to Dubai is quite the scenic route, what with overpasses of Greenland, Iceland, and Denmark along the way.
Not quite the shortest great-circle distance, though, given current geopolitical circumstances.
Not quite the shortest great-circle distance, though, given current geopolitical circumstances.
We got the friendliest welcome to Silicon Valley from the inimitable, one and only Robert Scoble!
His X Pro dashboard on his living room TV is a sight to behold: the den of a Mr. Universe, a reclusive techno-wizard sifting through the noise to find pure signal.
His X Pro dashboard on his living room TV is a sight to behold: the den of a Mr. Universe, a reclusive techno-wizard sifting through the noise to find pure signal.
March 8, 2025 at 3:30 PM
We got the friendliest welcome to Silicon Valley from the inimitable, one and only Robert Scoble!
His X Pro dashboard on his living room TV is a sight to behold: the den of a Mr. Universe, a reclusive techno-wizard sifting through the noise to find pure signal.
His X Pro dashboard on his living room TV is a sight to behold: the den of a Mr. Universe, a reclusive techno-wizard sifting through the noise to find pure signal.
✈️ My travel hack: get an Amazon Basics cast iron kettlebell delivered upon arrival in a new city. (Greets from San Francisco, btw 👋)
That's a one-time cost of about $1.1/lb ($2.4/kg) for a full-body workout on your own schedule. That's less than a one-time entry to a swank San Francisco gym. 💪
That's a one-time cost of about $1.1/lb ($2.4/kg) for a full-body workout on your own schedule. That's less than a one-time entry to a swank San Francisco gym. 💪
March 5, 2025 at 10:30 PM
✈️ My travel hack: get an Amazon Basics cast iron kettlebell delivered upon arrival in a new city. (Greets from San Francisco, btw 👋)
That's a one-time cost of about $1.1/lb ($2.4/kg) for a full-body workout on your own schedule. That's less than a one-time entry to a swank San Francisco gym. 💪
That's a one-time cost of about $1.1/lb ($2.4/kg) for a full-body workout on your own schedule. That's less than a one-time entry to a swank San Francisco gym. 💪
My ASIMOV Protocol sneak peek on the NEAR Town Hall: enriching user-owned AI with high-quality structured, linked, and verified knowledge in order to realize the promise of the neurosymbolic synthesis.
(Alternative tagline: Building positronic brains since 2024!)
youtu.be/oJmAkEctnOk
(Alternative tagline: Building positronic brains since 2024!)
youtu.be/oJmAkEctnOk
ASIMOV Protocol on the NEAR Town Hall (February 2025)
An ASIMOV Protocol sneak peek on the NEAR Town Hall in February 2025: enriching user-owned AI with high-quality structured, linked, and verified knowledge in...
youtu.be
March 5, 2025 at 10:24 PM
My ASIMOV Protocol sneak peek on the NEAR Town Hall: enriching user-owned AI with high-quality structured, linked, and verified knowledge in order to realize the promise of the neurosymbolic synthesis.
(Alternative tagline: Building positronic brains since 2024!)
youtu.be/oJmAkEctnOk
(Alternative tagline: Building positronic brains since 2024!)
youtu.be/oJmAkEctnOk
My ASIMOV Protocol sneak peek on the NEAR Town Hall: enriching user-owned AI with high-quality structured, linked, and verified knowledge in order to realize the promise of the neurosymbolic synthesis.
(Alternative tagline: Building positronic brains since 2024!)
https://youtu.be/oJmAkEctnOk
(Alternative tagline: Building positronic brains since 2024!)
https://youtu.be/oJmAkEctnOk
ASIMOV Protocol on the NEAR Town Hall (February 2025)
An ASIMOV Protocol sneak peek on the NEAR Town Hall in February 2025: enriching user-owned AI with high-quality structured, linked, and verified knowledge in...
www.youtube.com
February 20, 2025 at 11:43 PM
My ASIMOV Protocol sneak peek on the NEAR Town Hall: enriching user-owned AI with high-quality structured, linked, and verified knowledge in order to realize the promise of the neurosymbolic synthesis.
(Alternative tagline: Building positronic brains since 2024!)
https://youtu.be/oJmAkEctnOk
(Alternative tagline: Building positronic brains since 2024!)
https://youtu.be/oJmAkEctnOk
Monitoring account creation on the NEAR Protocol testnet in real time with a simple 15-line NEAR.rb script:
February 3, 2025 at 3:08 PM
Monitoring account creation on the NEAR Protocol testnet in real time with a simple 15-line NEAR.rb script:
Monitoring $NEAR payments on the NEAR Protocol in real time with a simple 15-line NEAR.rb script:
February 3, 2025 at 5:19 AM
Monitoring $NEAR payments on the NEAR Protocol in real time with a simple 15-line NEAR.rb script:
📦 Released NEAR.rb 0.3, a Ruby client library for the NEAR Protocol.
✅ Now supports parsing block, chunk, transaction, and action data from the neardata.xyz API. It's never been easier to track the chain or write an indexer!
👉 Most useful for rapid scripting and prototyping on NEAR Protocol!
✅ Now supports parsing block, chunk, transaction, and action data from the neardata.xyz API. It's never been easier to track the chain or write an indexer!
👉 Most useful for rapid scripting and prototyping on NEAR Protocol!
February 2, 2025 at 1:31 PM
📦 Released NEAR.rb 0.3, a Ruby client library for the NEAR Protocol.
✅ Now supports parsing block, chunk, transaction, and action data from the neardata.xyz API. It's never been easier to track the chain or write an indexer!
👉 Most useful for rapid scripting and prototyping on NEAR Protocol!
✅ Now supports parsing block, chunk, transaction, and action data from the neardata.xyz API. It's never been easier to track the chain or write an indexer!
👉 Most useful for rapid scripting and prototyping on NEAR Protocol!
💡 If you have young children and haven't yet tried using an LLM to generate endless, novel variations of bedtime stories on their current favorite obsessions, then you're really in for a treat.
January 19, 2025 at 9:33 AM
💡 If you have young children and haven't yet tried using an LLM to generate endless, novel variations of bedtime stories on their current favorite obsessions, then you're really in for a treat.
Needed to explain phone booths to my kids while watching old Superman cartoons. (“No, those aren’t actually changing rooms on the street.”)
That led to an explanation of phreaking which led to an explanation of hacking. The question what happens to Clark Kent’s discarded clothing still remains.
That led to an explanation of phreaking which led to an explanation of hacking. The question what happens to Clark Kent’s discarded clothing still remains.
January 17, 2025 at 5:46 AM
Needed to explain phone booths to my kids while watching old Superman cartoons. (“No, those aren’t actually changing rooms on the street.”)
That led to an explanation of phreaking which led to an explanation of hacking. The question what happens to Clark Kent’s discarded clothing still remains.
That led to an explanation of phreaking which led to an explanation of hacking. The question what happens to Clark Kent’s discarded clothing still remains.
🔀 Comparing and contrasting efficient binary serializations for RDF knowledge graphs, I know about:
RDF/HDT (by Javier Fernández), RDF Thrift (in Jena), CBOR-LD (by Manu Sporny), RDF Jelly (by Piotr Sowiński), and RDF/Borsh (by yours truly). And CBL (from IIT).
👉 Any others I might have missed?
RDF/HDT (by Javier Fernández), RDF Thrift (in Jena), CBOR-LD (by Manu Sporny), RDF Jelly (by Piotr Sowiński), and RDF/Borsh (by yours truly). And CBL (from IIT).
👉 Any others I might have missed?
January 15, 2025 at 8:10 PM
🔀 Comparing and contrasting efficient binary serializations for RDF knowledge graphs, I know about:
RDF/HDT (by Javier Fernández), RDF Thrift (in Jena), CBOR-LD (by Manu Sporny), RDF Jelly (by Piotr Sowiński), and RDF/Borsh (by yours truly). And CBL (from IIT).
👉 Any others I might have missed?
RDF/HDT (by Javier Fernández), RDF Thrift (in Jena), CBOR-LD (by Manu Sporny), RDF Jelly (by Piotr Sowiński), and RDF/Borsh (by yours truly). And CBL (from IIT).
👉 Any others I might have missed?
Boy, looks like 2024 was a productive year for me over on GitHub! (This is a 96% certified alpha leak.)
January 13, 2025 at 9:17 PM
Boy, looks like 2024 was a productive year for me over on GitHub! (This is a 96% certified alpha leak.)
📦 Released Borsh.rb 0.1, a Ruby library for encoding and decoding data in the Borsh binary serialization format designed for security-critical projects where consistency, safety, and performance matter.
👉 Most useful for prototyping on NEAR Protocol!
https://github.com/dryruby/borsh.rb
👉 Most useful for prototyping on NEAR Protocol!
https://github.com/dryruby/borsh.rb
January 12, 2025 at 12:59 AM
📦 Released Borsh.rb 0.1, a Ruby library for encoding and decoding data in the Borsh binary serialization format designed for security-critical projects where consistency, safety, and performance matter.
👉 Most useful for prototyping on NEAR Protocol!
https://github.com/dryruby/borsh.rb
👉 Most useful for prototyping on NEAR Protocol!
https://github.com/dryruby/borsh.rb
The easiest way to recognize LLM-generated text is simply that my eyes glaze over while reading the bland bullet points.
More advanced prompting mitigates this, but that's perhaps 1% of LLM users so this remains a pretty useful heuristic.
More advanced prompting mitigates this, but that's perhaps 1% of LLM users so this remains a pretty useful heuristic.
January 6, 2025 at 10:13 AM
The easiest way to recognize LLM-generated text is simply that my eyes glaze over while reading the bland bullet points.
More advanced prompting mitigates this, but that's perhaps 1% of LLM users so this remains a pretty useful heuristic.
More advanced prompting mitigates this, but that's perhaps 1% of LLM users so this remains a pretty useful heuristic.