Track Pack
Lifetime access to one of four tracks of the PickBits University curriculum. Read at your pace, build at your weekends, ship what you learn.
See the four tracks → Agentic AI training · Phoenix, AZ & remote worldwide
Agentic AI training
that ships.
Built in Phoenix, Arizona by Mark Pickering — the operator who runs PickBits.AI. Self-serve as the PickBits University curriculum, live as a workshop in your room, or embedded as a multi-week consulting engagement that wires real agents into your real stack. No lecture slides. No "future of work" handwaving. Production patterns from someone who ships agents into production for a living.
Workshop
Half-day or full-day workshop on agentic AI patterns. On-site across the Phoenix metro, or remote-delivered for distributed teams worldwide.
Scope a workshop →Consulting
Multi-week engagement that audits, builds, and embeds agentic workflows into how your team works. Four weeks to first production win.
See consulting → Free weekly workshop · Phoenix-led, remote-open
A themed, operator-led
workshop every week.
Sixty minutes, one theme, one operator. No vendor pitches, no panels, no "future of work" handwaving — just the patterns that worked last week, the patterns that broke last week, and your questions. Free, live, recorded for those who can't make it.
Inaugural session
Don't ship the AI prototype that worked on the first try.
The Vibe Coder pipeline — Explore, Validate, Harden — and the transition triggers most builders miss. Walk through a real prototype on screen; leave with the pre-ship checklist that keeps it from paging you at 2 a.m.
Saturday, July 11, 2026 · 10:00 AM Phoenix (MST) · 1pm ET · 6pm UTC · 7pm CET
RSVP free →
via Luma · calendar invite + reminders
What's coming up
// themes are subject to swap based on what's live in the room
Wk 02 · Jul 18
Prompt → Workflow → Agent. Pick the right tier.
Wk 03 · Jul 25
Build your first MCP server (and avoid the tool-use security holes).
Wk 04 · Aug 01
Cost control for production AI. The twelve-line rate-limit that saves $1,500.
Wk 05 · Aug 08
Evals: stop swapping models on vibes. The 20-row file that gates every release.
Wk 06 · Aug 15
Agent security: when someone else's document becomes your prompt.
Wk 07 · Aug 22
Claude vs. ChatGPT vs. Cursor vs. Copilot. Picking the right tool for the job.
Wk 08 · Aug 29
Build a Second Brain with Claude Cowork.
Wk 09 · Sep 05
Agentic AI for solo operators & small business.
RSVP & follow the series →
Following the calendar adds you to reminders for every future session — no spam, just the next link.
Definitional
What is agentic AI training?
Agentic AI training is structured instruction on building and operating AI systems that take multi-step actions on a user's behalf — calling tools, reading and writing files, running queries, making decisions — rather than just generating text. It covers prompt patterns, the Model Context Protocol (MCP), the workflow-vs-agent decision, evals, security review, cost control, and the production discipline (observability, rollback, audit) that keeps an agent from paging you at 2 a.m.
Generative AI is a calculator. Agentic AI is a junior employee. They share the same model underneath, but the engineering discipline around them is different — and that discipline is what gets taught here.
vs. generative AI training
Generative AI training teaches you to prompt for one good output. Agentic AI training teaches you to design a system that loops — plans, calls tools, checks itself, and recovers. Different failure modes, different evals, different security model.
vs. AI coding bootcamps
A bootcamp teaches someone to write code with AI assistance — useful for beginners. Agentic AI training assumes you already ship code and shows you how to make an AI ship code on your behalf, safely. Senior engineering discipline, not intro track.
vs. chatbot / prompt training
Prompt training stops at the conversation window. Agentic training continues through tool design (MCP), workflow tiers (prompt → workflow → agent), evals, security review, cost guardrails, observability, and the rollback plan when the agent does something unexpected at 2 a.m.
Outcomes · in plain operator voice
What you (or your team)
will stop doing.
Real outcomes look like fewer 2 a.m. pages, smaller cloud bills, and code review your senior engineer doesn't quietly rewrite on Monday. Here's the short list.
- Outcome 01Stop shipping the AI prototype that worked on the first try. The one that pages you at 2 a.m. is the one that worked too easily.
- Outcome 02Stop letting an agent burn $1,500 of API spend overnight. A hard rate-limit and a per-user daily cap is twelve lines of code.
- Outcome 03Stop shipping code your senior engineer throws away in the rewrite. Four artifacts make AI-built work inheritable.
- Outcome 04Stop swapping models on vibes. A 20-row eval file is the difference between "this is better" and "this regressed forty percent of our flows."
- Outcome 05Stop reaching for an agent by default. Most production AI work belongs in the middle tier — a workflow — at one-tenth the cost.
- Outcome 06Stop pretending your agent's tool list is a security model. Someone else's document is the prompt now.
Who it's for
Four audiences. Four tracks.
The curriculum splits four ways because the people building agents, the people running them in production, the people integrating them into knowledge work, and the people operating a business around them all need different things. Same instructor, different room.
// engineers
Builders shipping AI features
Indie hackers, app founders, first-time AI builders, and senior engineers whose prototype just got real users. Vibe Coder track.
// IT & ops
Teams running AI in production
Engineering leads, DevOps/SRE, security, FinOps — anyone responsible for cost, reliability, audit, or vendor risk on an AI surface. Vibe Management track.
// knowledge work
Professionals using AI to think
IT pros, BAs, architects, consultants, marketers, lawyers, healthcare, finance — anyone whose job is information work. Field Guides track.
// solo operators
Operators building around AI
Solo studios, indie creators, consultants, and one-person shops standing up an AI-augmented business. Solo Business track.
Three ways in
Self-serve, live workshop,
or embedded engagement.
Pick the format that matches the budget and the urgency. Same curriculum, different velocity.
01
Self-serve · Track Pack · $45
Lifetime access to one of four tracks of the PickBits University curriculum. Read at your pace, build at your weekends. Best for solo builders and individual contributors deciding whether agentic AI is worth their time.
02
Live workshop · groups · half- or full-day
The operator in your room, teaching your whole group hands-on — on the tools they already run, or the University curriculum delivered as a live class. On-site across the Phoenix metro, or remote-delivered worldwide. Everyone walks out with the prompt vaults and configs from the session.
03
Embedded engagement · teams · 4–12 wks
Multi-week consulting that audits, builds, and embeds agentic workflows into how your team actually works. Four weeks to first production win, ten to twelve for a full rollout. Priced against hours-recovered and labor-shifted-to-compute, not flat day rates.
04
1:1 coaching · individuals · 60 min
Sixty focused minutes on your own codebase with the operator. Best when you're stuck on a specific stack problem — choosing between agent vs workflow, designing an MCP server, getting an eval suite past the first ten rows. Free with any Track Pack.
The curriculum
Four tracks. Pick one,
or all four.
Each track is a self-contained mental model plus a working stack of prompts, configs, and patterns you keep for life. A Track Pack unlocks one. A Substack Experimenters subscription unlocks all four plus community.
Track · vibe coder
Build agents that ship.
17 modules. For engineers and indie hackers whose AI prototype is starting to have real users.
- When to vibe, when to agent, when to write it yourself
- The prompt → workflow → agent decision tree
- AI-first TDD, code review, security review
- MCP server design and tool-use security
- Observability, cost guardrails, release engineering at AI velocity
- Audit, compliance, continuous evals
Track · vibe management
Run AI in production.
10 modules. For engineering leads, DevOps, security, and FinOps owners of an AI surface.
- Cost control and attribution across models and vendors
- Reliability and SLO patterns for agentic workloads
- Vendor management — Anthropic, OpenAI, Google, open-source
- Risk envelope, audit, and incident response for agent failures
- Buy-vs-build for agent platforms
- Negotiating capacity and pricing with vendors
Track · field guides
AI as your second brain.
16 modules. For IT pros, BAs, architects, consultants, and anyone whose job is information work.
- Personal knowledge database — your private second brain
- Team Brain — shared context for the whole department
- Claude Cowork patterns — delegating real work to an agent
- Microsoft 365 Copilot recipes (Pages, Studio, Teams, Prompt Gallery)
- GitHub Copilot in VS Code patterns
- Print-and-pin reference cards for every module
Track · solo business
Operate around AI.
8 modules. For solo studios, indie creators, and one-person shops standing up an AI-augmented business.
- Pricing AI services without becoming a commodity
- Positioning and differentiation in an AI-saturated market
- Personal economics of an operator-led business
- Workflow design for a one-person team
- Being the operator — the role, the discipline, the day
Compared
How PickBits agentic AI training compares.
There are good options for learning agentic AI. PickBits sits in a specific corner of that map — built and taught by one operator who ships agents into production every week, available as both a $45 self-serve curriculum and a multi-week embedded engagement. Here's the shape of it.
| Dimension | PickBits | Anthropic Academy | Coursera AI courses | Big-4 AI Academy |
|---|---|---|---|---|
| Self-serve curriculum | Yes · $45 lifetime / track | Yes · free | Yes · subscription | No |
| Live workshop on your stack | Yes · half/full day | No | No | Yes · enterprise only |
| Embedded multi-week engagement | Yes · 4–12 wks | No | No | Yes · 6 figs+ |
| Phoenix, AZ on-site | Yes · operator-led | No | No | Yes · associate-led |
| Tool-agnostic (Claude, ChatGPT, Cursor, Copilot) | Yes | Claude-only | Mostly | Yes |
| Production discipline (cost, evals, security) | Core curriculum | Yes | Light | Yes |
| Built by someone shipping agents weekly | Yes — Mark Pickering, operator | Anthropic engineers | Academic instructors | Practice partners + associates |
| Entry price | $45 / 1 track | Free | ~$50/mo | Custom (5–6 figs) |
Comparison reflects published offerings as of 2026. Anthropic Academy refers to the Anthropic-published learning materials; Big-4 AI Academy is a category label for the in-house AI training programs operated by major consultancies (Deloitte, EY, PwC, KPMG, Accenture).
Proof · what graduates & clients say
Real engineers,
real IT leads, real quotes.
Lifted from emails, Slack DMs, and the end-of-engagement notes. Six voices across the four tracks — builders, IT teams, knowledge workers, solo operators. Names go on once each person signs off on attribution.
Vibe Coder · engineer
// placeholder · awaiting client sign-off
[QUOTE 1 — placeholder. Likely a Vibe Coder track grad describing the pre-ship checklist or the moment the eval suite caught a regression.]
??
Vibe Coder · indie builder
// placeholder · awaiting client sign-off
[QUOTE 2 — placeholder. Indie hacker or first-time AI builder; the "I shipped without the 2 a.m. page" angle.]
??
Vibe Management · IT lead
// placeholder · awaiting client sign-off
[QUOTE 3 — placeholder. Engineering lead or DevOps owner; the cost-control + observability angle.]
??
Vibe Management · security
// placeholder · awaiting client sign-off
[QUOTE 4 — placeholder. Security or audit-side stakeholder; tool-use security, MCP, prompt-injection angle.]
??
Field Guides · knowledge worker
// placeholder · awaiting client sign-off
[QUOTE 5 — placeholder. IT pro / BA / consultant; second-brain or Claude Cowork delegation angle.]
??
Solo Business · operator
// placeholder · awaiting client sign-off
[QUOTE 6 — placeholder. Solo studio / indie creator / one-person shop; pricing AI services or workflow-design angle.]
??
Looking for a longer write-up? See the case studies — including the full Excellens.us build.
Where we work
Built in Phoenix.
Trains teams worldwide.
PickBits.AI is operated from Phoenix, Arizona. The studio ships globally — the Daily Signal newsletter and four-track University curriculum reach readers in every major timezone — and the workshops travel with it. Locally in the Phoenix metro we deliver on-site for engineering teams, IT departments, and leadership cohorts; remotely we run the same workshop over video for distributed teams anywhere in the world.
Phoenix metro on-site: Tempe, Scottsdale, Mesa, Chandler, Gilbert, Glendale, Peoria, Surprise — any company or venue within the Valley. On-site travel time is included for engagements scoped on the strategy call.
Remote & worldwide: Workshops, coaching, and consulting all run over video for distributed teams. Sessions land between 7 a.m. and 7 p.m. Phoenix time — comfortable for US, Latin America, and same-day windows for Europe, Africa, and the Middle East. Asia-Pacific is asynchronous-friendly with scheduled video checkpoints.
Why a one-operator studio: The same person who writes the curriculum, teaches the workshop, and ships the daily AI artifact is the one who shows up in your room. No handoff to a junior. No "I'll loop in a colleague." It is what it looks like.
Asked & answered
Common questions.
Phrased the way people actually ask them.
What is agentic AI training?+
Agentic AI training is structured instruction on building and operating AI systems that take multi-step actions on a user's behalf — calling tools, reading and writing files, running queries, making decisions — rather than just generating text. It covers prompt patterns, the Model Context Protocol (MCP), workflow vs. agent decisions, evals, security review, cost control, and the production discipline (observability, rollback, audit) that keeps an agent from paging you at 2 a.m.
How is agentic AI different from generative AI?+
Generative AI produces a single output from a single prompt — a paragraph, an image, a code snippet. Agentic AI plans, loops, and uses tools — it calls APIs, reads files, browses, runs code, and decides what to do next. Generative AI is a calculator; agentic AI is a junior employee. Same model underneath, very different engineering discipline around it.
Where can I get agentic AI training in Phoenix, Arizona?+
PickBits.AI is operated from Phoenix and runs in-person workshops across the Phoenix metro — half-day or full-day, on-site at your office or a venue you choose. The same curriculum is also available as a self-serve Track Pack at $45 one-time or as a remote-delivered workshop for distributed teams.
Who is agentic AI training for?+
Engineers shipping AI features into production (Vibe Coder track), IT teams automating internal workflows (Vibe Management track), knowledge workers who use AI to think for a living (Field Guides track), and operators running an AI-augmented business (Solo Business track). PickBits assumes professional working knowledge of your tools — no intro sessions.
Can you train my remote team on agentic AI?+
Yes. Live workshops run over video for distributed teams in half-day and full-day formats. Material, prompt vaults, and configs from the session are shared with everyone in the room — yours to keep and extend. Scope on the free 30-minute call.
What tools and platforms does the training cover?+
Claude (Anthropic), Claude Code, ChatGPT, Cursor, GitHub Copilot, Microsoft 365 Copilot, NotebookLM, Gemini, and the Model Context Protocol (MCP) for tool integration. The curriculum is tool-agnostic where it can be and tool-specific where it has to be — agent security, MCP server design, and CI integration are platform-specific by necessity.
How long does agentic AI training take?+
Self-serve: a Track Pack is structured for one focused week of evening reading plus a few weekend builds. Workshop: half-day (4 hours) or full-day (8 hours) with hands-on labs. Embedded engagement: four weeks to a first working production workflow, ten to twelve weeks for a full rollout across a team.
Do you offer corporate or enterprise agentic AI training?+
Yes. Corporate workshops run on-site (Phoenix metro) or remote (worldwide) for groups of any size — engineering teams, IT departments, leadership cohorts. For deeper rollouts the consulting engagement embeds AI workflows into a team's actual stack over four to twelve weeks, with measurable hours-recovered and cost-shifted metrics.
What's the difference between agentic AI training and an AI coding bootcamp?+
A coding bootcamp teaches someone to write code with AI assistance. Agentic AI training teaches someone to design, ship, secure, and operate AI systems that act autonomously — which is a senior engineering discipline, not an intro track. PickBits assumes professional working knowledge of code and tooling; we don't run intro sessions.
How much does PickBits agentic AI training cost?+
Self-serve Track Pack: $45 one-time, lifetime access to one track. Workshop: scoped on a free 30-minute call based on group size, format (half/full day), and on-site vs. remote. Consulting: scoped to the engagement after a free strategy call — typically priced against hours-recovered and cost-shifted ROI rather than flat day rates.
Who is Mark Pickering?+
The operator of PickBits.AI. Mark ships games and AI products in public from a one-person studio in Phoenix, Arizona — then teaches exactly how each one was built. Workshops, consulting, and University curriculum are all written and taught by the same person who's doing the work the rest of the week.
Why focus on agents specifically, not just "AI training"?+
Because by 2026 the interesting work — and the interesting failure modes — moved from "can the model write the thing" to "can the system take the action safely, at cost, under audit." The gap between teams that ship agentic systems well and teams that ship them once and then quietly turn them off is wider than the gap between teams that use AI at all and teams that don't. Agentic is where the discipline pays.
I'm not in Phoenix. Should I still consider PickBits?+
Yes — most clients are remote. Phoenix is where the studio is operated from; the work travels. Self-serve curriculum, live workshops over video, and consulting engagements all run worldwide. The Phoenix anchor is a fact, not a requirement.
Pick your door
Three ways to get
your team building.
Forty-five dollars and an evening, a workshop in your room, or a multi-week engagement that ships the workflows for you. Each one starts with a real conversation.