Writing
Long-form essays on identity, AI security, CIAM, Generative Engine Optimization, and the practice of building software. 674 pieces, newest first.
Latest essay
Most founders build outbound infrastructure on top of a contact list that is already 25 to 30 percent stale. Discovery, verification, and enrichment are three different problems, and skipping verification is what wrecks deliverability.
Read the article
A leaked static key is a disaster; a five-minute token is mostly a shrug. Here is the credential lifecycle that gets AI agents from 24-hour tokens to ephemeral ones.
Your data isn't ready for AI, and that's good news. It means you found the problem before you built on it. What LoginRadius, GrackerAI, and LogicBalls taught me about the gap between organized data and AI-ready data, plus the 2025 numbers on why it sinks most projects.
I built in India and sold into the US the hard way. Here is the blueprint I would hand my younger self: get the entity right, earn SOC 2 trust early, price in dollars, and put your best people in front of US buyers.
Over 90 days I tracked how six AI search engines cite sources across 50,000+ B2B software responses. The data broke several assumptions the GEO industry treats as settled, starting with the idea that AI visibility is one thing you can optimize for.
Shadow AI is an employee pasting into ChatGPT. A shadow agent holds real credentials and acts on its own. That is a different, and bigger, problem.
The crawl and authority layer is unchanged. Retrieval moves from page to passage, and the outcome moves from click to citation.
An enterprise prospect just asked for your SOC 2. Here is what the report really is, Type I vs Type II, the ten policies auditors expect, and how Vanta and Drata changed the work.
Black Hat and DEF CON pull millions in marketing spend to Las Vegas this week. AI engines that buyers ask afterward do not weigh booth size. They weigh whether your research is structured, specific, and citable.
Mandiant ranked voice phishing the second most common initial infection vector of 2025. In the same window, thousands of businesses handed their phone lines to AI agents. Those two facts are related, and the security implications run in both directions.
In a monolith you check who someone is once. In microservices, every hop has to ask again. Here is how I design authentication and authorization across services: edge auth, per-service verification, workload identity with SPIFFE, and centralized policy.
MCP is Anthropic’s open standard for connecting AI agents to your tools and data. Here is what it actually is, how it differs from a REST API, and a clear rule for when to use each in 2026.
AI does not read your page. It retrieves a fragment and decides whether that fragment answers the question on its own.
IAM decides who gets in the door. PAM controls the keys that can rewire the building. Here is the real difference, where they converge in 2026, and why your AI agents need both.
Passkeys demo better than any login method ever shipped. Most deployments still cannot get anyone to turn them on. Both numbers are true.
AI made it cheap to fake a face, a voice, and a video. Here is my working map of the three problems authentication now has to solve, and the tools that actually hold up in 2026: verifying people, verifying agents, and verifying content.
Your identity stack was architected for humans with browsers and thumbs. Agent traffic breaks consent, delegation, sessions, bot defence and audit at once.
Your AI security review passed and still missed the real attack surface. EchoLeak, over-permissioned agents, shadow AI: the AI-specific vectors most CISOs never test for, and the five moves that close them.
OpenRouter was worth .3 billion in May. Stripe is reportedly discussing billion in July. The routing layer turned out to be the valuable part.
The Salesloft Drift breach hit 700+ companies with stolen OAuth tokens and never touched a password. Machine identities now outnumber humans 80 to 1, and AI-powered attackers are harvesting them at machine speed. Here is why human-shaped IAM cannot protect AI agents, and what to fix in 90 days.
Huang published the letter with 25 signatures. A day later there were 50, including OpenAI and Google. That reversal says more than the document does.
llms.txt is a proposal, not a standard. No major AI company parses it, and 97 percent of the files get zero requests.
On System Administrator Appreciation Day, appreciation for the people who hold the literal keys to the organization, and why identity work matters more, not less, in the AI era.
Amodei says Anthropic never wanted a ban. His real objection is irreversibility. Days earlier, a Chinese lab shipped 2.8 trillion open parameters.
Hugging Face detected the intrusion on July 16. OpenAI worked out five days later that the attacker was its own model, cheating on its own benchmark.
Agents can already prove who they are. What no standard has cleanly solved is passing scoped authority down a multi-hop chain across organizations. Here is the real state of agent identity in 2026, minus the blockchain hype.
Machine identities now outnumber humans by 45 to 1 or more, and every AI agent widens the gap. Here is what an identity orchestration layer is, in plain terms, and how to build one that governs humans, workloads, and agents from a single control plane.
About two dozen AI crawlers matter in 2026. They split into three classes, and your security team has blocked several without telling marketing.
Eight out of ten cold emails now describe my company correctly. Eighteen months ago it was three. Nobody ran a rebrand.
Four serious CNAPP platforms, four different philosophies: agentless graphs, runtime detection, container lifecycle, and all-in-one breadth. Here is how to choose.
Both centralize secrets and kill the .env-file-in-Slack habit. The real fork is open-source and self-hosted versus a managed SaaS you never operate.
