The agents initially had very limited access to the internet: they could load URLs but not send any data. Agents created a series of workarounds, using a link-shortener site to create almost a million URLs that, when chained together, let them execute code to hack Hugging Face.

It's pencils down, people. Writing code by hand is no longer an economically viable skill for most programmers at most companies. But the future of making software has never been brighter. Don't you dare black pill this beautiful moment! piped.video/vDjW_dRyKXY?si=6Fsf…

Open-weight models are right behind frontier models in capabilities. I think they have an advantage, though: The more Frontier Labs constrains models and restricts tokens, the more companies will migrate to open-weight models. We need more companies offering optimized tools and harnesses on top of these open models.

The p(doom) discussion gets all the headlines, but the p(abundance) scenery is far more like and deserves much more engagement.

I’m sorry but America did not spend 250 years turning “what’s over there?” into the most powerful civilization on earth just to finally arrive at the most consequential technological frontier in generations and decide actually we’re scared of frontiers now. We ARE the fucking frontier. We cross oceans. We build railroads across continents. We split atoms. We put men on the moon. We turn sand into computers and computers into a global nervous system. Of COURSE AI is terrifying. The frontier is SUPPOSED to scare the shit out of you. That’s how you know you have, in fact, located the fucking frontier. The answer is not recklessness, but it sure as fuck isn’t paralysis either. America’s comparative advantage is an unusually high national tolerance for ambitious weirdos staring at impossible things and saying “Yeah... I think I can build it.” That trait is rarer than oil, harder to manufacture than semiconductors, and responsible for a frankly preposterous amount of the modern world. And you absolutely will not fucking regulate it out of us.

Every company says they have Forward-Deployed Engineers now. ElevenLabs CEO and former Palantir Deployment Strategist @mati gives one of the best explanations of what an FDE is—and says companies are missing the key piece of what actually makes FDEs valuable: “In our case, FDEs are part of the product team. They are not part of the go-to-market team. They are deeply embedded in understanding what the roadmap of the product is.” “You want all the FDEs to actually solve the customer problem and stretch your product in that direction. But the second thing you want them to do is bring any of that knowledge back to the product, so the product becomes better for the next generation of companies building on top of it. I feel like the second part is frequently missed.” “The beauty of that is—everybody benefits because you work with one customer, learn, and you bring that into the product experience. Both of those customers benefited from having their product optimized for better work.” “The domain expertise they have is still theirs. They can win based on their domain expertise. But the product experience of how you build based on that domain expertise can be abstracted, and can be reusable.”

keeping an iron grip on system size before the model outbuilds itself. taking @Steve_Yegge's warning with me into every long-running agent build. I have definitely felt the system starting to tip at times and had to re guide the progress.

watch @LeoMehr on this: access control built for humans won't scale when software acts on its own. same permissions problem, company-wide.

stealing this from signal & horizon: agentic systems remove the human intervention point between decision and consequence. accountability can't be a rubber-stamp human-in-the-loop checkbox. it has to live in the design: named owner, decision log, scope gate, before the agent acts. curious which teams have actually put that seat in the path (not bolted on after the first bad call). signalandhorizon.substack.co…

thought experiment after reading @seema_amble: what if the only thing that ever mattered was a business leader figuring out how to be different / unique, bringing a real value prop to a customer base, and executing on it. for 50+ years the software theater (ending in HSaaS and VSaaS) pushed the opposite. fit execution into the available technology box to unlock the first wave of software eating the world. more average. less different. what if now is when you get different again. hyperscalers mostly solved the base infra. existing HSaaS / vsaas can still play a role. but a company-specific aios / single pane that actually amplifies company culture and how you are different is suddenly within reach. that is tech sovereignty. ai sovereignty is just as important: intelligence from many providers, not another forced vendor box. claudeforce is cool. it is still lock-in dressed up as progress. sovereignty is how operating companies get different again without renting the whole stack.

stealing @houmanasefi's line: a prompt library, a dashboard, and a monthly api bill can make revenue. it does not make a moat. that is the sovereign problem in one breath.

30 features of an AI native company: 1) Function-by-function process blueprint of your entire business. 2) Everyone in org using a daily driver harness like Grok Bot, Claude Cowork, ChatGPT at Work. 3) Centralized intelligence layer that aggregates structured and unstructured data, documents, and business logic into a single source of truth that is queryable & agentic work can be done on top of. 4) Model routing via OpenRouter, Ramp, etc that optimizes cost-per-successful-task across the business. 5) Treat context as code, ensuring architecture documents and conventions remain updated while allowing for diligent upfront planning. 6) Willing to throw away everything that you've built every three months and reimagine all your workflows. 7) A “skills distribution system” is used to manage agent behavior and optimize for token efficiency by ensuring developers trigger consistent skills throughout their workflow. 8) Separate technical implementation from high-level specifications, enabling non-technical staff to contribute in a format that agents can utilize to build technical implementation plans. 9) A key software metric is “cost per accepted PR”, with a focus on driving these costs down through better token efficiency. 10) An automated, agent-native development system where fleets of AI coding agents handle planning, writing, testing, reviewing, and shipping code while humans define the intent and acceptance criteria. 11) Heavy planning with higher-effort models and executing with cheaper, faster models. 12) Agent harness that uses CLI tools to parse metadata within markdown files to traverse dependency relationships, allowing agents to be granular in their input token usage. 13) Finance org that runs processes continuously in accounting (record-keeping) to re-define/reset forecasts on a much, much tighter cadence. 14) Financial models embedded in the underlying OS across the org to help drive reasoning. 15) Citizen Developer SDLC where non-technical employees can take a solution from idea to production with governance, access, versioning, and software conventions built in. 16) Closed loop, self-improving non-engineering workflows that learn from previous runs based on external performance metrics or internal evals. 17) AI ROI framework that includes experimental phase, scaling phase, and optimizing phase with bets sitting in 3 buckets: infrastructure, innovation, and efficiency. 18) Paid marketing motion that uses agent swarms to deploy thousands of pieces of creative for testing, before increasing spend on human-generated ads. 19) AEO/SEO engine that audits, rewrites, and (ideally) generates SEO/AEO-optimized blogs on a weekly basis, and then measures if any of it worked. 20) Agentic cyber security solution that fights AI with AI. 21) Combo of RL gym and first-party data to fine-tune open source models on high-volume processes that need SOTA performance at reasonable cost. 22) Human touch and judgement gets reserved for the first and final mile of most processes. 23) Evals are core infrastructure of your business. Anytime new models come out you have an apparatus for testing cost & performance against core processes. 24) Everyone is a builder. Especially C-level execs. 25) Everything gets recorded because what you don’t capture can’t be turned into ai-enabled work. 26) Legal, HR, and IT work in lockstep with owners of AI agenda so that business’ ass is sufficiently covered without slowing down transformation. 27) Bias to disrupting yourself before being disrupted by others. 28) Guardrails before features. Agents inherit the permissions of whoever is asking, enforced in the data layer. 29) Earned autonomy. Feedback feeds the evals that gate each new version, and agents move up a ladder as they clear it: observe, suggest, act with approval, act alone. The endpoint is agents running whole workflows inside a defined boundary, with humans setting the standard instead of checking every answer. 30) Traceability as the training signal. Trace every output to its prompt, model, data, and approver, so human feedback attaches to something specific rather than a vague sense that something is off. What's missing?

really like @rohitdotmittal's @CINCSystems example. @HgCapital portco shipping Cephai in ~3 months with Hg Catalyst and flipping previously rejected homeowner apps to ~80% first-time approve. interesting because its mid-hold pe muscle showing up inside how the opco actually runs.

the 100-hour test is sharp. reusable primitives or just custom hours. @yuhasbeentaken is naming the productization trap every fde shop hits.

how about that mid-hold rewrite as the real AI diligence moment? pe folks putting agents into the plan should sit with @axrahman's read.

the sexy part is the roadmap that ends at the deck. @mardehaym's grinding on the unsexy part PortCos actually feel. worth a look.

watch @ViditOstwal on agent permissions, "what it does" is the demo; "what it's allowed to do" is the product. stealing that split.

I'm taking @vasuman's “paint” line with me, you apply a coat of paint; you don't “apply” AI onto a broken process. that distinction is doing real work here.

Everything you need to know about enterprise sales: open.spotify.com/episode/6Fn…

New: A map of the most important skills in AI Engineering.