Sean Michael Kerner's Spring 2023 Human Writer certificate authory.com/SeanMichaelKerne… That's write I'm human. I write about #AI. Skynet hasn't won (yet).

You've never routed like this before. @OpenRouter is bringing Jev to all of your LLM calls, so your agentic workflows never have to waste a token again. As always, faster, cheaper, more intelligent. Go build the future.

And then there’s the OpenClaw bombshell: Microsoft Autopilot is built on OpenClaw. This strongly suggests something much bigger about Microsoft’s agent strategy: Microsoft is just not that invested in its own agent runtime. It actually wants to own the CONTROL PLANE around it. So, this means OpenClaw become the open substrate. They help make it enterprise-grade. Then push policy, security, Windows integration and reliability upstream. Then they can differentiate the combined iffeing as Autopilot, over time, with Entra identity, M365 context, Graph, Azure compute, governance, distribution, and (the key ti unlocking revenue) ultimately the meter. It’s essentially the Linux/Azure playbook applied to agents: Commoditize the layer underneath you, then become the best place to run it. If OpenClaw becomes something like the “Linux of agents,” Microsoft doesn’t need a proprietary agent stack to win. It needs to become the enterprise operating environment for ALL agents. That is a much bigger, and in my opinion better, ambition than just Copilot.

IP Fabric builds a read-only digital twin of enterprise network infrastructure, and version 8.1 of the platform adds application infrastructure mapping and a redesigned cloud-native data model. networkworld.com/article/422…

I like plan mode. I want and need to see what the direction could be to make adjustments before actual coding occurs

Follow the link below to claim the credit or run /claim-credit in the CLI. You’ll need GitHub connected to start a session. Claim by Oct 7. Terms apply. claude.ai/code/claim-credit/…

We made claude​.ai 3x faster in two weeks. Here’s how we use Claude to measure, debug and improve performance. Prompts and methods included. claude.dev/blog/how-we-made-…

Wait what?? You did this all with just three.js? (not Blender or..?)

Network-as-code? Selector is taking a somewhat novel approach to building its agentic AI NetOps capability called foundry, using the same approach that makes GitOps work so well. networkworld.com/article/422…

Following in Google's footsteps - remember BigQuery Omni? same idea.

Building a database for machines means rethinking the architecture from the ground up. @VentureBeat's @TechJournalist sat down with our CEO @MKardashov to take a closer look at KeewanoDB & why the way machines consume and reason over data calls for a new DB. Read his take ↓

Open weights are not open source: Why AI's favorite label is under dispute theregister.com/columnists/2… via @TheRegister & @sjvn Downloading a model is increasingly easy. Understanding how it was made, or changing a system at its root, is another matter.

rip elastic. i know that using tuple ids in the text map(removes one additional hop) is not the only optimization that led to such fantastic results, but just wondering why no one else made this, it looks obvious to use a direct pointer to the data on disk rather than create another map table. congrats to the team, i'll be using this for sure!

Why AI companies are really pumping the brakes on their models computerworld.com/article/42… via @ComputerWorld & @sjvn #AI's big dogs say it's because frontier AI is advancing faster than they can test and secure it. True, but safety isn't their main motivation. Follow the money.

Next, I asked Jev to estimate how many rows a given filter would match. Postgres would then use this data to plan the query. This ended up working pretty well. The queries that improved the most were queries where outside context gave a lot of information about how to execute the query. When querying the IMDB data, Postgres estimated 1 in 100,000 movies are sequels. Jev was much more accurate and estimated 1 in 100 were

I've just burned through $100 of credit on @AnthropicAI Fable 5.1 for a gauntlet game dev effortlessly and I'm only about 1/10th of the way through... Yeah it's cheaper than building a full games studio - and paying ppl - but still...

Mark my words: Confidential inference is going to become the vital bridge between enterprise AI demand + the ability of the hyperscalers to provide it in a way that makes public cloud AI services trustable, and therefore consumable. Lots of issues to work out, but this is the way.

This is typical for PostgreSQL though.. stability is not a work in progress it’s a release goal

Because people seem so excited by our new logo and the launch of Cockroach Continuum, we're giving away some swag. Like/repost the post below for your chance to win. 👇

The Moon’s got a big new crater! 🕳️ Spotted by NASA’s Lunar Reconnaissance Orbiter, the McGetchin crater formed when a rock as big as a six-story building crashed into the Moon. It’s 141 feet deep and wider than the length of two football fields.🪨💥🌕 go.nasa.gov/3TzHqUf

The real genius of the AI business model was in the use of “token” as the unit of billing. How much work does a token do? It depends. How many tokens will this project take? All of them. Tokens are an ideal unit of metered billing, because they have no definite properties.