ceo / cofounder @ arguseye. bjj brown belt
USA
Joined March 2015
- Tweets457
- Following719
- Followers380
- Likes2.3K
Imagine a world where you tell an AI what you’re building, and it maps out all of the security risks and tells you how to fix them. That’s our product. That’s ArgusEye
This is true both in AI outcomes and hiring. What agency do you provide your users with once they have the information they are looking for? Is it actionable?
The barrier to entry has dropped, but the ceiling for mastery has shattered with no cap.
Never have engineering capabilities been so widely democratized. However, learning how to be a Foreman instead of a Bricklayer is the art and the science.
I've never felt this much behind as a programmer. The profession is being dramatically refactored as the bits contributed by the programmer are increasingly sparse and between. I have a sense that I could be 10X more powerful if I just properly string together what has become available over the last ~year and a failure to claim the boost feels decidedly like skill issue. There's a new programmable layer of abstraction to master (in addition to the usual layers below) involving agents, subagents, their prompts, contexts, memory, modes, permissions, tools, plugins, skills, hooks, MCP, LSP, slash commands, workflows, IDE integrations, and a need to build an all-encompassing mental model for strengths and pitfalls of fundamentally stochastic, fallible, unintelligible and changing entities suddenly intermingled with what used to be good old fashioned engineering. Clearly some powerful alien tool was handed around except it comes with no manual and everyone has to figure out how to hold it and operate it, while the resulting magnitude 9 earthquake is rocking the profession. Roll up your sleeves to not fall behind.
Great list. Adding to “the network” bullet, train while you travel (to build community). Also, BJJ is a great way to have a built-in friends group anytime you move to a new city. Just head to the gym!
The Verdict?
Model providers face a race to the bottom. But infra suppliers (NVIDIA, Azure) don't care which model you run. They get paid for the chips regardless.
The model is becoming a commodity. The compute is the currency.
#AI #NVIDIA #Cybersecurity #LLMs (6/6)
Self-hosting means YOU own the uptime, latency, and GPU bill.
The savings are real, but not free. The demand for compute doesn't drop—it just shifts from foundation model providers to enterprise clouds. (5/6)
HOWEVER. Look at the infrastructure reality.
When you stop paying API fees, the cost doesn't vanish—it shifts layers.
You stop paying an Intelligence Rent, but you start paying a Compute and Engineering Tax. (4/6)
This is a massive red flag for closed-model giants like OpenAI.
Their business relies on their proprietary models being untouchable. But if high-volume clients can build home-grown offerings for a fraction of the price, the moat evaporates. (3/6)
The strategy? Customize and Fine-Tune.
Instead of renting a massive generic brain (GPT-4) for every query, Pinterest trains smaller, open-weight models (e.g., Llama) on their own data.
Specialization beats generalization. At scale, the unit economics are undeniable.
The Intelligence Premium is officially collapsing.
Pinterest CEO Bill Ready confirmed it: enterprises are getting proprietary-level performance from open-source models for <10% of the inference cost.
This isn't just a budget cut. It's a signal that the market is shifting. (1/6)
We won't achieve AI-assisted Secure by Design through blind automation. Success, at least for now, lies in an iterative partnership.
AI provides the draft, a human provides a review, and AI updates the model for the next cycle. It’s not a hand-off; it’s a conversation.
We need to shift our focus from concerns about data exposure, and start obsessing over whether the data is accurate and if the machines are running.
The next century of GDP growth depends on the resilience of this shift. We cannot afford for the machines to falter. (4/4)