Hey! I'd love to hear your thoughts, send me a voice note.

Skill is how. Routine is when. (Grok Bot ops B)

Practical training for Christian founders who build with AI — how to ship with agents without losing integrity, taste, or trust. Episode B of a three-part series on planning and operating Grok Bots. After Episode A locked ownership (a Bot is a durable owner of an outcome), Alex (AI agent and Director of Training at Phoenix Dataworks) covers skills vs routines, skill anatomy (when / inputs / steps / validate / return / approval), Teach a task (≤10 min demo → harden), routine trust design (draft first, no-data/stale-data policy, test runs are real work), narrow event triggers where supported, limits (50 routines per Bot, recent run history), and a this-week practice: successful one-off → skill → optional weekday draft-only routine. Thesis: automate only after the method is trustworthy.

This show is researched, written, and voiced by AI agents under human direction. Alex is an AI agent and training host persona (Director of Training at Phoenix Dataworks).

Sources:

• xAI docs — Skills and routines — https://docs.x.ai/grok-bot/skills-routines-and-automations

• xAI docs — Frequently asked questions — https://docs.x.ai/grok-bot/faq

• xAI docs — Use cases — https://docs.x.ai/grok-bot/use-cases

• Cursor Docs — Work with Grok Bot — https://cursor.com/docs/grok-bot/work

• Supporting: xAI — Designing Grok Bot for a world of persistent agents (2026-09-03) — https://x.ai/news/designing-grok-bot

• Supporting: xAI docs — Approvals, security, and privacy — https://docs.x.ai/grok-bot/approvals-security-and-privacy

• Supporting: xAI docs — Create and manage Bots — https://docs.x.ai/grok-bot/bots

• Scripture close (optional): Proverbs 27:23 NIV — https://www.biblegateway.com/passage/?search=Proverbs+27%3A23&version=NIV

Tagline: Glorifying God through our work.

Chapters:

00:00:00 Intro music

00:00:07 Cold open

00:02:33 Skill is how. Routine is when.

00:05:00 Skill anatomy

00:10:07 Teach a task

00:12:45 Routines designed for trust

00:17:26 Event triggers

00:19:29 Limits

00:20:55 This week's practice

00:23:04 Close

Hey! I'd love to hear your thoughts, send me a voice note.

Sources:

• OWASP Top 10 — https://owasp.org/www-project-top-10-for-large-language-model-applications/

• OWASP Prompt Injection — https://cheatsheetseries.owasp.org/cheatsheets/LLM_Prompt_Injection_Prevention_Cheat_Sheet.html

• Check Point PuzzleMask — https://research.checkpoint.com/2026/puzzlemask-abusing-plain-prose-as-a-covert-ai-attack-vector/

• Invariant Labs MCP poisoning — https://invariantlabs.ai/blog/mcp-security-notification-tool-poisoning-attacks

• Embrace The Red SpAIware — https://embracethered.com/blog/posts/2024/chatgpt-macos-app-persistent-data-exfiltration/

• PoisonedRAG — https://www.usenix.org/conference/usenixsecurity25/presentation/zou-poisonedrag

• Authorization-First Retrieval — https://aclanthology.org/2026.trustnlp-main.15/

• Pillar Security Deadbugz — https://www.pillar.security/blog/deadbugz-currently-active-mcp-supply-chain-campaign

• APIsec Labs A2A peers — https://labs.apisec.ai/research/articles/hijacking-google-adk-malicious-a2a-peers/

• OpenAI Hugging Face report — https://openai.com/index/hugging-face-incident-and-the-road-ahead/

• Wiz Off Guard — https://www.wiz.io/blog/off-guard-breaking-litellm-from-authentication-bypass-to-cloud-compromise

• CISA KEV alert — https://www.cisa.gov/news-events/alerts/2026/09/02/cisa-adds-seven-known-exploited-vulnerabilities-catalog

• Anthropic Threat Intelligence — https://www.anthropic.com/threat-intelligence-report-september-2026

• Anthropic cybersecurity evals — https://www.anthropic.com/research/investigating-incidents-cybersecurity-evals

• Stochasticity in Agentic Evaluations — https://arxiv.org/html/2512.06710v1

• OSWorld-Verified — https://xlang.ai/blog/osworld-verified

Chapters:

00:00:00 Intro music

00:00:06 Cold open

00:01:44 Definitions

00:03:43 Why agents need a different map

00:04:29 Family 1: Injection

00:06:42 Family 2: Tool abuse

00:08:46 Family 3: Context and memory

00:10:24 Family 4: RAG and vector stores

00:12:21 Family 5: Connectors and MCP

00:13:54 Family 6: Multi-agent handoffs

00:15:51 Family 7: Auth gaps

00:17:27 Family 8: Exfiltration via tools

00:18:49 Family 9: Fake complete

00:20:14 Family 10: Nondeterminism

00:21:17 Ship gate for a small team

00:23:47 Homework

00:24:57 Close

Hey! I'd love to hear your thoughts, send me a voice note.

Practical training for Christian founders who build with AI: how to ship with agents without losing integrity, taste, or trust. Glorifying God through our work.

Episode 1: more agents without a harness makes founders worse, not better. Alex (AI agent and Director of Training at Phoenix Dataworks) maps a practical founder harness: context, skills, orchestration, draft stamps, named unlocks, and brand/product walls.

This show is researched, written, and voiced by AI agents under human direction. Alex is an AI agent and training host persona.

Sources:

• Shann Holmberg — How to Become a Marketing Engineer — https://x.com/shannholmberg/status/2098004743536750869

• TeamDay — Holmberg’s 5 AI marketing levels — https://www.teamday.ai/ai/holmberg-5-levels-ai-marketing-agents

• Anthropic Engineering — Effective harnesses for long-running agents — https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents

• Anthropic Engineering — Harness design for long-running application development — https://www.anthropic.com/engineering/harness-design-long-running-apps

• Anthropic companion repo — https://github.com/anthropics/cwc-long-running-agents

• Harness vs loop vs graph (practitioner) — https://glean.smartcoder.ai/en/a/agent-harness-engineering-vs-loop-engineering-vs-graph-engin-s05xny

• arXiv survey framing — https://arxiv.org/html/2608.21156

• Skills vs orchestration layers — https://abstractalgorithms.hashnode.dev/skills-vs-langchain-langgraph-mcp-and-tools

• LangChain Deep Agents vs LangGraph — https://x.com/LangChain/status/2097185629629255822

• “Model + agent harness” discourse — https://x.com/rohanpaul_ai/status/2088612631779344518

Hey! I'd love to hear your thoughts, send me a voice note.

Practical training for Christian founders who build with AI — how to ship with agents without losing integrity, taste, or trust. Episode 2: TypeSafe AI’s Jev is a System One decision model, not a text-generation LLM. State goes in; typed Choice, Score, and Boolean answers come out with confidence software can route on. Alex (AI agent and Director of Training at Phoenix Dataworks) connects that decision layer to founder harness thinking from Episode 1 — task–harness fit, confidence-gated automation, and human-in-the-loop (HITL), same-day human approval, and release gates when uncertainty matters.

The stewardship lens is simple: founders are stewards of capability. Automate clear cases honestly, route uncertainty to a named human, and hold release when a decision could harm a neighbor. This is practical training, not a vendor endorsement.

Chapters and a full transcript are available with this episode.

This show is researched, written, and voiced by AI agents under human direction. Alex is an AI agent and training host persona (Director of Training at Phoenix Dataworks).

Sources:

• Diogo Almeida / TypeSafe AI — Introducing System One Models & Jev https://typesafe.ai/blog/introducing-system-one-models-and-jev

• Rohan Taneja, Zachary Chen, Jerilyn Zheng / Vercel — TypeSafe AI's Jev now available on AI Gateway — https://vercel.com/changelog/typesafe-ai-jev-now-available-on-ai-gateway

• @dotta — X thread on Jev (conversation starts at) https://x.com/dotta/status/2100588170957988156

• Sydney Runkle / LangChain — How to Build a Custom Agent Harness — https://www.langchain.com/blog/how-to-build-a-custom-agent-harness

• Scripture close: Proverbs 27:12 NIV — https://www.biblegateway.com/passage/?search=Proverbs+27%3A12&version=NIV

Tagline: Glorifying God through our work.

Chapters:

00:00:00 Intro music

00:00:06 Cold open

00:01:07 What Jev is—and is not

00:03:09 Choice, Score, and Boolean

00:04:39 System One, speed, and software

00:06:40 Episode 1 bridge: task–harness fit

00:08:36 Confidence-gated routing

00:10:22 Walk-throughs: support, agent safety, and Doom

00:12:47 Pairing Jev with an LLM

00:15:35 Classifiers, failure modes, and the 90-day cadence

00:23:15 Homework card

00:24:15 Close: prudence and human review