A week after launching ClearPoint Nexus, my bet: the AI skills market will be won on certification standards, not catalog size. Security questions are coming first. Good. https://clearpointnexus.com

Claude Code Skills & AI Agent Workflows | ClearPoint Nexus

Nexus Certified AI skills for Claude Code, Codex, OpenClaw, and Google Antigravity. Production-ready workflow files for research, sales, ops, and more. From $9.99 - deploy in minutes.

clearpointnexus.com

Sunday operating model: coffee, long walk, zero tabs open. The backlog will still be there Monday. It always is. That's what makes it a backlog.

Sunday reset: the best technology strategy has a boring core, which is making the business easier to operate. Fewer fragile systems, cleaner handoffs, less heroics. Compounding starts where the drama ends.

Saturday, close of business. All systems suspended. Tongue service still running. Ginny is our availability dashboard.

Weekend builder note: the best product decisions on ClearPoint Nexus came after launch, not before. Users beat whiteboards. Ship the small version.

A data poisoning attack pushed a securities firm's AI agent to recommend fabricated investment products to customers. Input integrity is now a fiduciary control. https://aigovernance.com/news/data-poisoning-attack-forces-financial-ai-agent-to-recommend-fabricated-securities

Data Poisoning Attack Forces Financial AI Agent to Recommend Fabricated Securities Products, Exposing Critical Input Validation Gap | AI Governance Institute

A documented incident at an unnamed securities firm revealed that an attacker manipulated market data consumed by an autonomous AI trading agent, causing…

aigovernance.com

China's agent rules took effect July 15: the first dedicated regulatory category for AI agents anywhere. Illinois now mandates third-party safety audits. https://aigovernance.com/news/chinas-agent-rules-take-effect-july-15-and-illinois-mandates-third-party-safety-audits

China's Agent Rules Take Effect July 15 and Illinois Mandates Third-Party Safety Audits, Creating Dual Compliance Deadlines for Enterprise AI Teams | AI Governance Institute

China's Implementation Opinions on intelligent agents became enforceable on July 15, 2026, establishing the world's first dedicated regulatory category…

aigovernance.com

The numbers defining enterprise AI in 2026: 1.3M new AI builder roles in two years. 14% of orgs with a clear AI strategy tied to accountability. The first hire for your AI operating model isn't an engineer: https://jaredmabry.com/blog/the-first-hire-for-your-ai-operating-model-isnt-an-engineer

The First Hire for Your AI Operating Model Isn't an Engineer

Companies added 1.3 million AI builder roles in two years, yet nearly 80% of enterprises can't say who owns their AI initiatives. The first hire for your AI operating model isn't an engineer. It's an owner.

jaredmabry.com

The labs sat in on four vendor calls today. Ginny slept through the pricing slide. Georgia now growls at "circle back." Fastest learners on the team.

Straiker's new STAR Labs report: 1,700+ successful exploits against production AI agents. 36% of coding-agent attacks reached remote code execution. 91% of productivity-agent attacks left no trace. https://sg.finance.yahoo.com/news/straiker-research-36-successful-ai-120300274.html

New Straiker Research: 36% of Successful AI Coding Agent Attacks End in Remote Code Execution

Straiker, The Agentic Security Company, today released the inaugural threat report from its STAR Labs team. To understand the real-world risk AI agents pose to enterprises, researchers ran thousands of adversarial scenarios across production coding, productivity, and first-party agents, resulting in more than 1,700 successful exploits. Volume I is available to read and download here.

sg.finance.yahoo.com

49% of employees admit to using unapproved AI tools at work. The C-suite is the most relaxed cohort. Shadow AI is the new shadow IT, and banning it just removes your visibility: https://jaredmabry.com/blog/shadow-ai-is-the-new-shadow-it

Shadow AI Is the New Shadow IT

In a late-2025 survey, 49% of employees admitted using AI tools their company never approved, and the C-suite is the most relaxed about it. Banning it does not remove the demand. It removes your visibility.

jaredmabry.com

Teams that mandate SBOMs for software will paste an agent workflow from a stranger into a tool that can read the whole repo. The AI skills supply chain is 2014-era open source all over again. That's the gap certification closes. https://clearpointnexus.com

Claude Code Skills & AI Agent Workflows | ClearPoint Nexus

Nexus Certified AI skills for Claude Code, Codex, OpenClaw, and Google Antigravity. Production-ready workflow files for research, sales, ops, and more. From $9.99 - deploy in minutes.

clearpointnexus.com

CIO by day. By night: a grown man losing an argument with a browser game's physics engine, one sprite at a time. Range matters.

Prompt injection detection is now a stock CodeQL check as of 2.26.0. Mainstream AppSec is admitting it: prompts and skills are supply chain. It's why every skill on https://clearpointnexus.com passes 96 checks plus a 7-dimension security scan before listing.

Claude Code Skills & AI Agent Workflows | ClearPoint Nexus

Nexus Certified AI skills for Claude Code, Codex, OpenClaw, and Google Antigravity. Production-ready workflow files for research, sales, ops, and more. From $9.99 - deploy in minutes.

clearpointnexus.com

83% of 1,400+ senior IT leaders say their infrastructure needs upgrades to run agentic AI. Only 17% fully trust their stack for mission-critical agents. Google's own 2026 report, via CIO Dive: https://www.ciodive.com/news/agentic-ai-strains-legacy-it-systems/825003/

Agentic AI strains legacy IT systems

More than 4 in 5 organizations need to upgrade their tech stacks to support AI agents at scale, according to Google’s 2026 State of AI Infrastructure report.

ciodive.com

We launched ClearPoint Nexus: certified AI skills for Claude Code, Codex, OpenClaw, and Google Antigravity. Nothing lists until it clears 96 checks plus a 7-dimension security scan. The vetting layer the AI skills ecosystem is missing. https://clearpointnexus.com

Claude Code Skills & AI Agent Workflows | ClearPoint Nexus

Nexus Certified AI skills for Claude Code, Codex, OpenClaw, and Google Antigravity. Production-ready workflow files for research, sales, ops, and more. From $9.99 - deploy in minutes.

clearpointnexus.com

311 integers in a public commit = your whole .env. Ghostcommit hides prompt injection in a PNG because AI code reviewers skip images. Vet what agents ingest. https://www.bleepingcomputer.com/news/security/ghostcommit-hides-prompt-injection-in-images-to-fool-ai-agents-steal-secrets/

'Ghostcommit' hides prompt injection in images to fool AI agents, steal secrets

A PNG hiding a prompt injection could steal your repo's secrets, researchers demonstrate. The technique, dubbed 'Ghostcommit,' slipped past AI code reviewers CodeRabbit and Bugbot, which never open image files at all, then convinced a coding agent to read a repo's .env and write every secret into the code as a list of numbers.

bleepingcomputer.com

Almost nobody owns the AI operating model. Of 505 Global 2000 execs surveyed, nearly 80% couldn't say who owns their AI initiatives. You can't rent accountability. Part two, after shadow AI: https://jaredmabry.com/blog/the-first-hire-for-your-ai-operating-model-isnt-an-engineer

The First Hire for Your AI Operating Model Isn't an Engineer

Companies added 1.3 million AI builder roles in two years, yet nearly 80% of enterprises can't say who owns their AI initiatives. The first hire for your AI operating model isn't an engineer. It's an owner.

jaredmabry.com

The technology leaders I trust most move fast and still know what changed. They automate and still know how to recover. Speed without that kind of clarity is just well-intentioned chaos.

The interesting agent governance question is where the real control model is still living in people's heads: what can the agent touch, what requires approval, what gets logged, who owns the exception, and who shuts it down?

New from The Restless CIO: The Agent Sprawl Problem Is Already Here. The next enterprise AI question is who controls the environment where agents act. https://jaredmabry.com/blog/the-agent-sprawl-problem-is-already-here

Open protocols help with connection, and managed runtimes help with speed. Neither solves identity, permissions, logging, approval, revocation, or accountability by itself. That is the control-plane work.

A model is easier to swap than an agent runtime. Once the workflows, credentials, logs, approvals, and monitoring live inside a provider's environment, switching gets much harder. That is infrastructure, not just model selection.

Every agent that works well in a demo creates the same set of uncomfortable follow-up questions. Ownership, permissions, logging, shutdown. That is where the real enterprise AI work starts.

Trying to be more intentional here. My lane: AI execution, healthcare technology, CIO operating models, and the gap between adopting tools and changing how work actually changes. I'm interested in practical lessons more than hype.

Pilots create optimism. Workflows create reality. If AI cannot survive messy data, unclear ownership, and a Tuesday afternoon production issue, it is not deployed. It is visiting.

The most underrated leadership skill is creating enough clarity that talented people can move without waiting for permission every five minutes. Autonomy without clarity becomes drift. Clarity without autonomy becomes bureaucracy.

A good technology roadmap should not just list projects. It should explain what gets simpler when the work is done: fewer handoffs, fewer fragile integrations, faster decisions, better reliability, cleaner data.

The AI skills gap isn't about tools. It's about the habit of reaching for AI before reaching for Google, a meeting, or a consultant. Power users aren't special. They just changed the reflex first.

Stanford AI Index: 10% of Americans excited about AI. 56% of AI experts are optimistic. That's a deployment gap, not a messaging gap. What I've seen actually work: https://jaredmabry.com/articles/the-trust-gap-is-ais-next-big-problem #EnterpriseAI #CIO #Leadership

Jared Mabry — Technology Leader, Builder, Investor

SVP/CIO at D4C Dental Brands. Building MorningKeep and HelloYaYa through Mabry Ventures. Writing about AI, technology leadership, and building products that matter.

jaredmabry.com