AI & Security Architect building production-ready Agentic AI systems, IAM solutions, and the protocols and frameworks that make enterprise AI implementations se
At best they can create a lossy snapshot of what people thought mattered at a particular moment, then context overload, stale retrieval, contradictions, undocumented judgments , and lost-in-the-middle effects take over.
And even if it did work, Its funny how the default fantasy is "do less with fewer people" and not "try things we couldn’t afford to try before."
Agree! Flagship models are not going to get cheaper. They will likely get more expensive.
because the real cost is being absorbed by funding. When that stops, the current pricing model breaks because the unit economics will not work.
Anthropic's CC changes are the proof point: weekly limits, separate credits for agent tools, and users burning tens of thousands in compute on $200 plans (once usage moves from one-shot to long-running multi-turn agent loops, the math breaks quickly).
GitHub's latest security incident shows where software supply chain risk is moving.
A poisoned VS Code extension reportedly led to an employee device compromise and unauthorized access to internal repositories.
Over the past few months, the same pattern keeps showing up across developer tools, npm packages, browser extensions, AI browsers, and coding agents.
Attackers are moving closer to where credentials, source code, and internal access already live. Agentic tools widens that path because they can read files, search directories, run commands, call tools, and act across the same developer environment.
If credentials, API keys, configs, scripts, packages, or internal files are reachable from the developer environment, treat them as potentially compromised.
1/ We are sharing additional details regarding our investigation into unauthorized access to GitHub's internal repositories.
Yesterday we detected and contained a compromise of an employee device involving a poisoned VS Code extension. We removed the malicious extension version, isolated the endpoint, and began incident response immediately.
Wrapping internal APIs as tools and piping raw responses back through the model is the wrong architecture for enterprise agents. It is also the default in most frameworks today.
The model ends up sequencing calls, holding state across them, interpreting backend errors, and reasoning over payloads it should never see. Part 1 walked through ten failures that follow from this choice.
This one is about what to do instead.
The model emits a typed intent: onboard this vendor, plan this trip etc and a runtime takes it from there. Capabilities are code files with decorators for dependencies, retries, scopes, and failure policies. The runtime walks the graph, manages state outside the context window, enforces permissions, and returns a small projection the model can reason about.
The model sees two tools regardless of how many capabilities exist behind them. Sensitive data stays in the runtime. Every execution leaves a structured audit record.
https://t.co/VVrBhUfHOX
#softwarearchitecture #systemdesign #distributedsystems #enterprisearchitecture #agentharness #llmops #aiengineering
Agreed! I wrote about where this goes next.
Unreadable work is only the first failure. The deeper damage is that people stop learning the craft, companies cut the people who were also the customers, and accountability gets reduced to a patch and the same system running again.
At some point "built by humans" may become the premium label. https://t.co/YbE6mBvZvn
Fei-Fei Li warns that AI may be staring too hard at language models.
The world is not just text on a screen.
It is physical, visual, spatial, and always changing. Most of the economy runs on seeing, moving, interacting, and embodied intelligence.
@rohanpaul_ai I wrote about the same problem here: AI can speed up the task while dumping more work on the person using it. More checking, more switching, more cleanup, more decisions. At some point these "productivity tools" start creating cognitive debt. https://t.co/hSKR9mF4PC
@trq212 Fully agreed! HTML is great for explaining ideas. https://t.co/WVv7lptcIn
Dual coding theory suggests people understand and remember ideas better when words and visuals work together.
https://t.co/ASuDoZivhf
HTML is great for explaining technical ideas because the page can become part of the explanation.
I used that approach in my LLM explainer. Tokenization, embeddings, attention, probabilities, and inference are easier to understand when you can see the pieces move and connect, instead of only reading about them.
AI is making the work faster. Workdays are not getting shorter.
That's the paradox. Execution got cheaper, so the savings got reinvested into more scope and more ambition, not more rest.
The challenge now is spending that leverage on purpose, and keeping enough people deep enough in the system to know what's actually getting shipped.
Read the full piece.
#ai #productivity #futureofwork #softwareengineering #aidevelopment #technology #engineeringleadership
https://t.co/j3BzL5fBky
Spot on!! Searle's Chinese Room experiment cuts through most of these claims. An LLM can manipulate symbols and generate coherent language without any understanding of what those symbols mean. It has no awareness of consequences, no lived experience behind its responses, and no evidence of subjective comprehension. Mimicking intelligent-seeming behavior is not the same thing as being conscious.