Carlos E. Perez retweeted
We switched 10 people from iPhone to Blue. Weeks later, none of them want their iPhone back.
Introducing Blue: a phone you just talk to. Tell it what you need, and it gets done.
It has all the context, can use every single app, and even triages your notifications.
We built it for people who always have a little too much to do. The ones buried in messages, notifications, and to-dos.
If that's you, get yours today.
Backed by Y Combinator + 70 other VCs & angels.
heyblue.com
Carlos E. Perez retweeted
What are you afflicted with?
15%AI psychosis
2%AGI denial
56%AI ADHD
27%AI time-warping
55 votes • Final results Carlos E. Perez retweeted
The CDL26 programme is now live!
Two days of real-world deployments, Business Value, Developer Evolution, Scale & Governance, Ontology Engineering & Practice, Rapid Insights.
Clinical trial protocols turned into knowledge graphs, Norway's railway mapped as one, and Siemens using graphs to power its industrial AI agents.
AstraZeneca, Mercedes-Benz, Bloomberg, Google, ServiceNow, and Scania are among those sharing what's working with knowledge graphs, ontologies, and AI right now.
William Tunstall-Pedoe opens with a keynote on neurosymbolic AI, and there are tracks for everyone, from beginners to advanced.
⭐ Take a look at the full programme and book your tickets now: 2026.connected-data.london/p… ⭐
📍11-12 November | London Tower Bridge
Keynotes from distinguished speakers
Presentations with leaders and innovators
Panel discussions exploring diverse views
Masterclasses with expert instructors
Lightning Talks to get you up to speed
#KnowledgeGraph #GraphRAG #Ontology #Graph #AI #DataScience #GraphDB #SemTech
Carlos E. Perez retweeted
Link to the slides for "Biological Naturalism and Biological Chauvinism," given yesterday at the excellent UCLA conference on biological naturalism about consciousness.
petergodfreysmith.com/wp-con… 1/
Carlos E. Perez retweeted
We're giving away a Hogwarts Legacy custom wrapped NVIDIA GeForce RTX 5090 to celebrate its official release with DLSS 5 🪄
To enter:
🟢 Share this post
🟢 Comment #RTXON
Terms and Conditions: nvidia.com/en-us/geforce/con…
Carlos E. Perez retweeted
Ontology-Driven Observability: Building an End-to-End Knowledge Graph at Netflix Scale
A home screen renders 1.5 seconds slower. An alert fires. Four hours later, the cause is found.
Very little of those four hours went into hard computation. Most went into reconstructing relationships the systems already held, but had never written down where a human or a machine could read them.
The scale behind it: ~65M concurrent streams during the Paul vs Tyson fight, 2B+ requests on an ordinary day, 38M+ real-time logging events per second.
The gap was not more data or better dashboards. Atlas, Radar, incident.io, PagerDuty, Spinnaker and Slack are each internally coherent and mutually isolated. An application in one is a service in another and a cluster in a third. Joining them depended on a person carrying the mapping in their head.
The response is the Observability Ontology, built on RDF, OWL, SHACL and SPARQL, and persisted in QuipuDB, a homegrown RDF quad store based on redb. Twelve namespaces each own one operational domain.
Design choices worth noting:
🔹 No generic Service type. Every ops:Application is validated against a software catalog of roughly 49,000 entries
🔹 Dependency edges are capped in both directions, so one popular service cannot dominate the graph
🔹 Incident Slack channels become ordered triples, preserving hypotheses tried and discarded, and the mitigation that worked
🔹 LLM output must fit a typed contract, so stochastic steps produce a deterministic graph that answers a SPARQL query the same way every time
On top of the graph, enrichers add typed triples. A correlation engine reads thirteen signal types and scores candidate chains within a ten-minute window, discarding anything below 0.6 confidence.
Each chain is stored with the records that produced it. The Knowledge Flywheel (Observe, Enrich, Infer, Adapt) lets patterns from past incidents become detection rules, and specialist agents in AutoRCA use the graph’s vocabulary as their shared interface.
The risks are named: hallucination from insufficient context, erroneous causal hypotheses, and overreliance on automation. The mitigations are strict type validation, evidence and confidence stored alongside every conclusion, and humans approving every proposal.
Next: connecting the ontology to Upper, the domain-modeling language at the center of Netflix’s Unified Data Architecture.
By Prasanna Vijayanathan Renzo Sanchez-Silva
See also Model Once, Represent Everywhere: UDA (Unified Data Architecture) at Netflix by Alexandre Bertails
Ontology-Driven Observability: Building an End-to-End Knowledge Graph at Netflix Scale netflixtechblog.com/ontology…
Model Once, Represent Everywhere: UDA (Unified Data Architecture) at Netflix 2025.connected-data.london/t…
#Observability #RDF #SPARQL #SHACL #AIOps #DataModeling #DataEngineering
--
Connected Data London 2026 | 11–12 November | Leonardo Royal Hotel London Tower Bridge
🎟 Tickets on sale now - 🔔 Full programme announced
sbee.link/pkjvrbngqd
📺 Sponsorship opportunities available. Contact [email protected] for details.
#KnowledgeGraph #GraphRAG #Ontology #Graph #AI #DataScience #GraphDB #SemTech
Carlos E. Perez retweeted
A proposed theory of physics reveals that the universe is a living process!
Carlos E. Perez retweeted
1/ Mercury → Gold via nuclear fusion is no longer just medieval fantasy.
A San Francisco startup just claimed they did it experimentally — at 10× CERN’s record.
Here’s the real story, the science, the catch, and why it could (or couldn’t) change both energy and gold. 🧵
2/ The idea from Marathon Fusion (2025 paper):
High-energy neutrons from a D-T fusion reactor hit mercury-198.
That knocks out a neutron → mercury-197.
Mercury-197 decays in ~64 hours into gold-197 — the only stable gold isotope.
Same gold as in your jewelry. Just made with fusion neutrons.
3/ Scale they modeled:
~2 tonnes of gold per GW_thermal per year.
That’s roughly 5,000 kg per GW of electricity.
At today’s prices, that revenue could roughly match or exceed the electricity revenue of the plant.
Double the value of a fusion power plant without cutting power output.
4/ The catch everyone skips:
Fresh product isn’t pure stable gold. Side reactions create radioactive isotopes (mainly from mercury impurities).
You need 7–18 years of storage before it’s “banana-safe,” depending on the standard and how pure the mercury feedstock is.
Enrich the mercury and you can shorten that a lot.
5/ Recent update (Oct 2026):
They announced an experimental demo of the mercury-to-gold reaction using fusion neutrons, beating CERN’s previous record by 10×.
Still tiny quantities. Not a factory. But the nuclear reaction itself is no longer just a simulation.
6/ Now the economic reality check people miss:
If you built a fusion plant *only* to make gold (no electricity sales), the cost would be ugly.
Full capital + ops amortization on a multi-billion-dollar plant → several thousand dollars per ounce.
Well above mining costs and current spot price. Uneconomic as a pure gold factory.
7/ As a *byproduct* of a plant already selling electricity?
Incremental costs (mercury feedstock, enrichment, extraction, handling) look much lower — potentially a few hundred $/oz once the plant is running at scale.
That’s the only version that makes economic sense.
8/ So is alchemy solved?
Nuclear transmutation of mercury into stable gold: yes, the physics works.
Industrial-scale, cheap gold that floods the market: not yet, and maybe never, because of fusion plant costs, mercury supply, cooling times, and the fact that fusion itself isn’t commercial.
9/ The real prize isn’t the gold.
It’s whether co-producing high-value materials (gold, medical isotopes, etc.) can make fusion economically viable sooner.
If fusion plants can sell power *and* precious metals/isotopes, the business case gets a lot stronger.
10/ Bottom line:
- Physics: real
- Experimental demo: claimed
- Cheap gold from fusion alone: no
- Fusion plants that also make gold as a bonus: interesting, still years away
Would you rather have fusion power or fusion gold first?
(Or both?)
Carlos E. Perez retweeted
Seems like they are getting ready to officially say that Sapiens, Neanderthals and Denisovans are all the same species...
Carlos E. Perez retweeted
The PHP security team is giving advance notice: Upcoming security release Wed, Oct 21st 2026, 14:00UTC, two critical vulnerabilities, both rated CVSSv4 9.2.
PHP users are encouraged to prepare to update. Earlier release if details become public or active exploit evidence.
Carlos E. Perez retweeted
I’ve always had a question: If I ever need to have my wisdom teeth removed, should I choose to undergo the procedure under general anesthesia? I’ve always been somewhat afraid of general anesthesia. At the core of that fear is the continuity of consciousness: after my consciousness is interrupted for a period of time, where will “I” awaken? Perhaps “I” is a localized concept, and there is no large-scale “self.”
Carlos E. Perez retweeted
A visceral visualization of the scale of AI relative to human cognition, explained through the analogy of a Rubik's Cube. @karpathy
Carlos E. Perez retweeted
Here is an explainer of the paper:
claude.ai/artifact/XbucMgr8N…
Carlos E. Perez retweeted
A visceral visualization of what AI can do relative to humans.
nitter.cf/Math_files/status/2110…
Carlos E. Perez retweeted
This is wild 🤯
Scientists just built tiny liquid robots made of water coated in Teflon particles. They can squeeze through cracks, swallow objects, merge with each other, and carry cargo — almost like living cells.
They start as ice cubes dusted with superhydrophobic powder, then melt into soft, armored blobs.
The future of soft robotics just got a lot weirder.
Carlos E. Perez retweeted
Always ask, "How does X change because of AI?" Here's an example of how ERP radically evolves in an AI context.
🤝 Paid partnership
For 25 years, Salesforce, SAP and enterprise software were built around one assumption: humans operate the software.
We’re changing that.
Introducing ERP•AI and Proto: business software built for agents, and an agent that operates business software.
Describe the system you need. ERP•AI builds it.
Then Proto takes over. It manages records, runs workflows and executes actions inside the software you just created.
This is Headless SaaS.
1: get ERP•AI
2: build your SaaS
3: post below
Best one wins a MacBook. First 100 decent entries get $100.
(a message from Steve Papa / founding team)


















