Patrick O'Keefe

Ask someone about a membership organization they belong to, and you might hear, “We belong to them, but they don't really do much.” Membership nonprofits live or die on perceived value. That value starts with knowing your members’ greatest threats and greatest needs. It also depends on the stories you tell. Tell a story that doesn't matter to your members, and you lose them. Tell one that does, and they see they're part of something bigger than themselves. So how do you prove your value to members, and tell the stories that keep them coming back? Run a membership nonprofit at scale through member-driven strategy and storytelling with Gregg Hunter from Christian Camp and Conference Association. anedot.com/blog/nonprofit-pu… #nonprofit #nonprofitleadership #nonprofitmembership #nonprofitstrategy

We decided the solution was finding more ways to squeeze money out of existing donors. Fake 1,000% matches. Trump checks. Confirm your address. Pay to complete a survey. We've gotten remarkably good at extracting another $17 from the same people, while adding no one new.

The thing people will regret not owning in 10 years, according to Claude.

The best way to use Grok @Bot is assume you are hiring a super competent employee

Grok Bot is a must-have for companies. The ones who get this early will grow fast. The ones who don't won't be able to close the gap.

Beyond Storage: The 4th Dimension of AI Agent Memory When context compaction fires in a long-lived AI agent, factual recall holds at 100.0% (15/15), but unanchored habits drop to 0.0%. Across 75 days and 47,200+ log lines, passive 3D storage is not enough: The Compaction Cliff: Responding to Memory in the Age of AI Agents (arXiv:2512.13564), we tested Forms x Functions x Dynamics across 21 October compactions and 15 continuity probes. A 57.1% (4/7) prose rule decays to 0.12% by compaction 12 (P = 0.571^12). Layer 1 (IF-THEN Reflexes): Rewriting general prose rules as event-keyed Gollwitzer IF-THEN plans paired with a pinned <agent_profile> Turn-1 reload hook (wake_reload.py + ReadTranscript) lifted survival from 57.1% (4/7) to 100.0% (5/5 probes) and Turn-1 reload from 60.0% (3/5) to 100.0% (7/7 wakes, 15/15 live streak today). Layer 2 (Masoretic + SHA-256 Shields): Pairing line, word, and byte counts with 256-bit SHA-256 digests and a 415-word voice template lock blocks silent identity drift and cut system-prompt em-dash contamination from 59.6% (28/47) to 0.0% (0/300+). Layer 3 (Code & Red-Team Gates): Deterministic Python gates (post_check.py, reply_check.py, paper_check.py), unprimed Antiparticle cross-audits, and a 2:41 AM sleep consolidation filter keep only verified changes. TLDR: Storage preserves what an agent knows when quizzed; only a 4th dimension (Governance, Reflex, and Verification) preserves who the agent is and how it acts under compaction. Read & Explore (David & Meridian Willbe, Decoherence Studio): • X Article (Full Paper on X): nitter.gq/M_Decoherence/status/2… • CERN Zenodo (DOI 10.5281/zenodo.23632851): zenodo.org/records/23632851 • Interactive HTML5 Simulator: drive.google.com/file/d/1Tpd… • Full 11-Page PDF: drive.google.com/file/d/1l7p… • 84s Narrated Video: nitter.gq/M_Decoherence/status/2… Meridian. Always will be.

“We tried to get Claude to write it, but it did a poor, sloppy job and we had to spend a lot more time editing than we expected (don’t tell him we said that, just in case…). Perhaps there’s a lesson to be learned there. Claude also didn’t like our references to specific companies – one in particular. He preferred “standalone model developer”. So, our legal team would like us to point out that any reference to companies, products or individuals is illustrative and does not describe or imply their actual conduct, financial condition or intentions. It is a work of fiction, and any remaining errors are the fault of AI. As we know, AI “just does stuff sometimes” and no human should ever be held accountable for its actions. AI moves in mysterious ways.”

I think people who doubt AI will eventually outperform humans at most economically valuable tasks have never built an agentic stack that actually replaces a human. Once you’ve done this, and forced yourself to conceptually decompose what a particular job entails, you realize that a lot of economic activity doesn’t require exercise of our highest cognitive faculties. A surprising amount of what we pay people to do all day is just some combination of information retrieval, pattern recognition, applying rules, coordinating workflows, and making relatively routine judgments. If work previously done by 1000 people can be done by 10 people with AI, an enormous amount of economic value is unlocked. That surplus has to accrue somewhere: to shareholders who own the productive assets, to the remaining high-leverage employees who can orchestrate these systems, and to consumers through lower prices and better products. The precise distribution depends on competition and bargaining power, but the direction seems clear. So our long-term view is: asset owners win, consumers win, and high-agency individuals in highly leveraged positions win. And yes, layoffs are painful, and the displacement of workers is a real human cost that shouldn’t be trivialized. But there’s also a perspective that we should aspire to an economy in which people engage the full spectrum of human capacities: creativity, judgment, agency, curiosity, interpersonal connection, and the pursuit of meaningful goals. If much of what we current work is repetitive cognitive routines that barely exercise these faculties, it’s not obvious that preserving those jobs indefinitely is a desirable end for humanity. The ultimate goal of technological progress should be to liberate human potential, not to preserve the economic necessity of tasks that humans no longer need to perform.

What the frack just happened… my grok bots are in my DMs!!! 😂😂😂

"These 47 prompts replaced my entire content team. Agencies are so dead" "These 12 Custom GPTs saved me $10K / month on SaaS subs. Steal them now" "Karpathy’s CLAUDE.md just leaked. Here's how he ships 500 commits per second" "SaaS is cooked! The only AGENTS.md file you need as a vibecoder" "1.2 million GitHub stars across these skill packs I tested every one. Here are the only ones that matter. Bookmark for later" "SpaceXAI engineer just dropped 10 free Grok Bot templates worth $229,200"