Enterprise AI
Becoming a Full Stack Builder
The title is new; the anxiety is familiar. Start with one bounded project, keep your expertise visible, and ask employers what authority, support, and fair terms come with broader work.
Groktopus
Human-led AI transformation, agentic systems, governance, and earned autonomy.
Enterprise AI
Stop Paying Frontier Prices: Five Lessons from Coinbase
Coinbase's fraud-agent series makes the case for testing specialists against frontier defaults. My agent-skills repository provides the references, templates, and scripts to run that experiment.
Enterprise AI
Agentic SRE and the Autonomy It Earns
Agentic SRE should earn authority one capability at a time. Human owners decide the limits; SRE brings the operational evidence.
News
McDonald’s Pricing Tool Tests Human Control: Can Franchisees Say No?
A restaurant-level price recommendation is not the same as a price set for each customer. The harder question is whether franchisees can challenge the advice in practice.
Sunday Signal
Sunday Signal Oct 3, 2026 - AI Gets the Keys
An AI research task reached Australian government systems it was not authorized to access. Elsewhere, persistent agents and data-center promises put access, oversight, and local accountability in view.
Enterprise AI
Fiverr’s AI Pivot Is Unproven: Who Gets to Move Up?
Fiverr is chasing bigger projects as buyers leave. AI is shifting freelance demand, but its gains don’t reach every worker.
Sunday Signal
Sunday Signal Sep 26, 2026 - Jev Won't Chat. Meta Wants Muse Everywhere.
Jev brings typed decisions to software. Meta’s Muse Charm makes an adults-only agent pocket-sized, while Claude spots an overlooked DNA repeat beside a known enzyme. The story is what AI should decide, what we invite into daily life, and what still needs a human lab.
Enterprise AI
Policy as Code Isn't Enough: Prove the Control Ran
A policy file describes intent, not what happened in a live run. Follow Bernstein’s evidence paths to see what reaches the live route, what gets recorded, and what still requires human judgment.
Enterprise AI
Jev and Laya Put Probability Where Enterprise Workflows Need It
Rules work until messy humans touch them, and generative models ramble when code needs an exact answer. Jev and Laya turn that gap into a measurable, governable component.
Sunday Signal
Sunday Signal Sep 19, 2026 - Small Brains, Old Warnings
OpenAI filed six incident reports on its own models, a fruit-fly brain took up headline writing, and the regulation sermon turned out to be 160 years old.
research
Mozilla's 6 Open Source AI Recommendations Every Business Should Be Watching
Mozilla's new report turns the open-weight AI debate into six practical recommendations for businesses that want the freedom to change models without surrendering the harness, state, permissions, or economics around them.
Case Studies
Airbnb's AI Bet: Culture Before Cost-Cutting
Airbnb says culture, not model access, will decide who wins AI. Its flat-headcount experiment raises a harder question: where did the recovered capacity go?
Sunday Signal
Sunday Signal Sep 12, 2026 - Cheap Models, Louder Alarms
OpenAI's agents spent May seeding a public software registry, DeepSeek made a large class of agent work several times cheaper, and a safety researcher quit with an extinction warning. Capability moved on schedule; the explanations kept arriving late.