https://xitter.cf/t.co/L4vR6UuF8z curated research for being more human and AI tools for creation and expansion
Joined March 2024
- Tweets294
- Following81
- Followers23
- Likes312
Pinned Tweet
This time is critical for families and kids.
I raised three kids.
I was a public schoolteacher.
Schools have gone off the rails and are not serving the humanity or knowledge for kids.
I'm back to provide resources for the family and the kids in this age of AI.
You said yes to one volcano video.
YouTube had other plans.
I built Focus Player that lets you lock in exactly what your kids watches — no ads, no sidebar, no rabbit hole. Just the videos you choose.try it. focusplayermvp.netlify.app?s…
Stop building harnesses. Start building fabrics. 🧵
Beyond the Harness
Instead of the traditional LLM+harness process+external+infra setup, Agent-o-rama provides an LLM client as an object inside a distributed, stateful compute fabric
solve the state nightmare in AI #aiagent
An AI agent just wiped 2.5 years of student data.
Here’s the terrifying part — and why the database architecture was the real villain. 🧵
tomshardware.com/tech-indust…
Whiteboard DB = one bad prompt from disaster.
Rama = agents can only append facts through defined paths. Deletion is a deliberate, human-controlled act. History is protected by design.
We’re building agentic systems on top of Rama for exactly this reason.
Agent-o-rama is designed so agents submit inputs → topologies process them → authoritative depots stay protected.
Agents get sanitized views. Not master keys.
🔴 Red Planet Labs / Rama — tag @nathanmarz
They literally cannot run migrations or destroy modules. The architecture enforces it.
School records in a StudentModule? No agent gets near that depot.
Here’s the deeper protection layer.
Rama gives you Depot.disallow() — a flag that blocks ALL external clients from writing to a depot.
Agents using foreign depot clients? Read and append only.
The open AI government surveillance machine deal is just another reason and demonstration.
Why I will never touch a single product from open AI.
Trust will be the number one value going forward
No trust of them at all!
businessinsider.com/anthropi…
The Canary in the coal mine. Just the beginning of possibilities within workforce. Increase output, reduced employee account, but new roles to play in new hires in different areas.
Last year, it took an Axios engineer 3 weeks to complete a project. This past week, a similar project was completed with the same amount of work in 37 minutes with AI, writes Axios CTO Dan Cox.
This is the reality of the AI hype you read and hear about.
axios.com/2026/02/15/ai-codi…
The change train is chugging and picking up steam. Do you hear it coming over the rise?
The new software development paradigm: the purity of the codebase is no longer the holy cow protected by angry senior devs.
Companies will use agentic code systems like Claude Code Opus 4.5 to fix messy codebases as long as they are fixable.
Once each new fix starts taking too much time or consuming too many tokens, the entire non-fixable piece will be rewritten from scratch based on the spec and the unit tests remaining from the previous non-fixable version.
No human will be involved in the code writing. Only in the spec and test writing.
Tor Keeling retweeted
Announcement blog post
blog.redplanetlabs.com/2025/…
Tor Keeling retweeted
We just open-sourced Agent-o-rama, a library for building, tracing, evaluating, and monitoring stateful LLM agents in Java or Clojure!
It closes the gap with Python tooling by expanding upon the ideas from LangGraph and LangSmith.
Link in replies.
It begins quietly, but like a flash flood, tearing down a narrow canyon, the gush of wind hits you in the face first. The wind is howling people! Who knows what this future hold.
