Zero-to-Hero Agentic Literacy — Historical Track Audit

This is historical evidence, preserved on 2026-09-09 from source checkpoint 827dc2c. The matching Astro checkpoint is 9fa94cc. The original outline below preserves planning inputs, not a second curriculum inventory or a commitment to build every idea.

The snapshot’s AI Literacy link is rebased for this audit’s location; its wording and the remaining original body are preserved.

The current Track Guide and shared Track Map carry the composition: bounded delegation and verification as the center; From Chatbot to Agent as the proposed signature; permissions/tools, context/memory, and review/recovery as developing routes; organizational delegation and building as optional depth.

Next Artifact Pass — Evidence and Unresolved Questions

• From Chatbot to Agent has no module source or canonical Lab package. The proposed first practice compares an answer with a tool-using plan for one small, low-risk task, then records permissions, approval points, verification, and a stop condition. The module, complete Activity, standalone Notebook, reviewed answer/plan examples, and facilitator guide are missing. The Notebook must carry the whole activity without the module page.
• Combine the historical comparison and simple-delegation-checklist ideas in this first practice. “Compare a chatbot answer to a tool-using assistant workflow” is an activity direction, not a settled Lab title. Do not claim an action ran when the learner examined only a plan or supplied example.
• Permissions and Sandboxes, Memory in Agentic Systems, Tools, Actions, and Connectors, Planning and Review Loops, Why High-Value Automation Can Still Be a Bad First Move, and Deploying Version 2 have no canonical module sources. The permission-identification and pre-action plan-review Lab ideas likewise have no source packages.
• The first safe live environment remains undefined. Start with vendor-neutral concepts and reviewed plan comparisons; selecting and testing a concrete environment is later work. It needs disposable practice material, enforced least-necessary access, visible action records, denied-action examples, tested stop/reset behavior, and results learners can verify. A sandbox label or a restrictive prompt alone does not establish containment.
• Keep proposed, authorized, attempted, completed, blocked, and verified actions distinguishable. A plan review is not execution permission. Narrowing or declining delegation, or using a simpler chatbot, can be successful outcomes. The existing Track title does not make full autonomy or expertise a promised destination.
• Conversation Engineering remains Sketch / Planned, owned by Renaissance Solutions, primary to AI Literacy, with no related Tracks. Its metadata dependency is Prompting Is the Door, Not the Destination. Its Beta Lab Stay, Steer, or Start Over is owned by Mike, primary to Conversation Engineering, related to AI Literacy, and has a standalone Notebook. Required documents are marked Coming Soon in that Lab’s metadata; no separate example packet is claimed here.
• Context Engineering remains Sketch / Planned, owned by Renaissance Solutions, primary to AI Literacy, with no related Tracks. Its metadata dependency is Conversation Engineering. Its Beta Lab Build the Brief Before the Work is owned by Mike, primary to Context Engineering, related to AI Literacy, and has a standalone Notebook and existing supporting materials.
• Conversation and Context learner copy describes lighter entry preparation than their metadata dependencies. Preserve both sources in this Track-only pass and resolve the mismatch during the artifact review; do not manufacture a new required sequence or silently drop a dependency.
• Safety: Share Less, Notice More and Reality Check for AI remain Draft / Drafting, owned by Mike, primary to AI Literacy and related to Curiosity, with no dependencies. Their featured Labs What Else Came Along? and Reality Check for AI: Six Experiments remain Beta, owned by Mike, with existing standalone Notebooks and unchanged primary-module relationships.
• These four support pairs provide conversation, context, privacy, and evidence practice. They do not demonstrate live-agent competence, prove tool execution, or substitute for a reviewed agent environment. Agentic Literacy’s new support selections do not update reciprocal module/Lab metadata; review those relationships in the subsequent artifact pass.
• Decision Ownership Before Delegation already exists despite the old outline’s unlinked name. It remains Seed / Planned, Business level, owned by Mike, review status Ideation integration; not reviewed, primary to AI Readiness, with no related Tracks and a Friction Mapping dependency. Its additional named relationships remain candidate relationships, not established prerequisites beyond the metadata dependency.
• Decision Ownership’s Decision Ownership Map and Delegation Ladder and Control Plan remain concepts without a canonical Lab or Notebook. Its small-business audience, staged authority model, governance examples, privacy review, and domain guidance make it later organizational depth, not the beginner signature. Assist or Recommend may remain the best operating point.
• The historical Before You Paste and Trust, but check dependency wording is retained below as evidence. The current Track points to the AI Literacy entry and existing Safety/Reality Check practice rather than declaring every historical topic a new required module. Module metadata remains untouched.
• Account-free supplied-example, paper, partner, observer, and facilitator comparison routes are intended participation paths; the signature comparison packet is not built. A live extension comes after environment review. Do not connect personal accounts, operate on real sensitive files, send messages, make purchases, or require irreversible public actions in the entry pilot.
• First pilot after package review: compare answer and plan, identify permissions and approval points, produce a short delegation checklist, and explain a human decision. Record actual timing, friction, and access needs. Checklist quality is not proof of tool success. All facilitator route demonstrations remain Coming Soon.
• Preserve Additional Reading, Track Draft / Drafting status, and Unassigned ownership. No module, Lab, Notebook, example packet, demonstration, environment, dependency, permission boundary, or source status is created or rewritten.

Original Track Guide Snapshot

Zero-to-Hero Agentic Literacy — Original Outline

Status: Working track outline, not finished curriculum.

This track follows basic AI literacy and introduces learners to AI systems that can plan, use tools, remember context, ask for permission, and complete multi-step work. It should not begin with autonomous-agent hype. It should begin with responsibility: what the human owns, what the tool can do, what permissions matter, and how to verify the result.

The invitation is not “let the AI take over.” The invitation is “learn how delegation changes when the assistant can act.” That includes knowing when to stop, when to review, when to narrow scope, and when a simple chatbot is the better tool.

Audience: Learners who have basic AI literacy and want to understand assistants, agents, automations, and tool-using systems.

Purpose: Teach safe, practical delegation to AI systems that can take actions or coordinate tools.

Likely Modules

• From Chatbot to Agent
• Decision Ownership Before Delegation
• Permissions and Sandboxes
• Memory in Agentic Systems
• Tools, Actions, and Connectors
• Planning and Review Loops
• Why High-Value Automation Can Still Be a Bad First Move
• Deploying Version 2

Starter Lab Ideas

• Compare a chatbot answer to a tool-using assistant workflow.
• Build a simple delegation checklist.
• Identify which permissions a task would require.
• Review an agent’s plan before allowing any action.

Additional Reading

No additional reading is listed yet.

Dependencies

• AI Literacy for the Rest of Us.
• Prompting as conversation.
• Before You Paste.
• Trust, but check.
• Decision ownership.

MVP Holes

• Define the first safe tool-using environment.
• Decide whether this track teaches one harness or vendor-neutral concepts first.
• Define permission examples that are real enough to matter but low-risk enough for class.
• Decide when agentic literacy becomes systems engineering.

---

Source: ops/track-audits/zero-to-hero-agentic-literacy.md