Lesson 11 — Orchestration: running multiple projects without losing your instructions
Standard: — · Bloom's: — · Structure: —.
Notice the Say-See-Do cycles running on Priya S.'s actual work — not a canned exercise — and that every capability claim is cited to the frozen doc-set. The exit ticket climbs Bloom's to —, and the lesson closes by writing to the ledger.
Learning objective
(Bloom's: Apply) Apply a multi-Project, multi-chat structure so that research, drafting, and admin work for more than one piece of writing stay separated but each still carry the standing instructions and context they need — without re-explaining yourself in every new chat.
Standard named: AIHC.1.D1 (Orchestration). At this scale, orchestration means structuring your own Projects and chats so nothing gets lost between them — the solo-scale form of P5.
Cycle 1 — One instruction, every project
SAY — Personalization layers combine rather than duplicate: Profile instructions ("Instructions for Claude") are "account-wide, applies to every conversation" [S13]; Project instructions are scoped to "a project you want Claude to hold" and apply "across all chats in that project" [S04, S13]. Both can be active at once, "each answering a different scope of 'how should Claude behave'" [S13].
SEE — A described diagram: a "Profile instructions" box at top with arrows fanning down into two Project boxes below, each Project box also containing its own Project instructions layered underneath the account-wide one.
DO — Confirm your confidentiality Profile Instruction from Lesson 09 is still active (Settings > "Instructions for Claude" [S13]). Then write one new Project instruction specific to The Last Mill project only (e.g., a tone or register rule for that piece) — without duplicating the account-wide rule.
⏸ Pause point. One layer confirmed, one new layer added. Stop here if needed.
Cycle 2 — A stray chat is a lost chat
SAY — Chat search "searches everything outside projects, plus each project's own chats, kept separate" [S12]; and recall from Lesson 08, "context is not shared across chats within a project unless the information is added into the project knowledge base" [S04]. A chat left outside the right project doesn't inherit that project's knowledge base or instructions. Referent (IPL cricket, flavor only): a fielder standing in the wrong position doesn't help the team, however good a fielder they are — positioning is the orchestration.
SEE — A described chat-history list showing one real chat currently sitting outside any project, that should be inside The Last Mill.
DO — Find one real chat of yours that's currently mis-filed, and move it in: "click the dropdown arrow next to a chat's name, choose 'Add to project,' then pick the target project" [S04].
⏸ Pause point. One chat correctly filed. Stop here if needed.
Cycle 3 — Separate the pitch tracker from the piece
SAY — Your pitch-tracker admin work and your longform drafting are different jobs with different context needs. A second Project keeps them from bleeding into each other's knowledge base, while starring — "from the Projects page, open the three-dot menu and choose 'Star'… Starred projects appear in the left-side panel for quick access" [S04] — keeps both reachable without hunting.
SEE — A described left-sidebar showing two starred projects side by side: "The Last Mill" and "Pitch Tracker / Admin."
DO — Create or confirm a second real Project for your pitch/admin work if it doesn't already have its own space, and star both projects [S04].
⏸ Pause point. Two projects, both starred. Stop here if needed.
Cycle 4 — Multiple chats, one shared knowledge base
SAY — Within one project, separate chats can each handle a separate job while all drawing on the same uploaded sources — because the knowledge base is project-wide, not chat-wide [S03, S04]. That's how you split "interview-question prep" from "drafting" without losing the shared context each one needs.
SEE — A described project view: two or three chat threads listed under one shared knowledge- base panel.
DO — Open a new chat inside your real The Last Mill project for a task you'd normally jam into your main drafting thread (e.g., prepping tomorrow's interview questions), and confirm it can already see your uploaded sources.
Independent at-bat
Unscaffolded: re-audit your full real project list. Move any remaining stray chats, confirm your instruction layering (Profile + each Project), and decide which chat goes where for the coming week of work — no checklist provided this time.
Exit ticket (climbing to Apply; graded against the doc-set)
- (Remember) Name the two personalization scopes from Cycle 1, broadest to narrowest.
- (Understand) Why does a chat left outside any project put your provenance and confidentiality habits (Lessons 08–09) at risk, even if you never meant to break them?
- (Apply) You start a brand-new chat to prep tomorrow's interview questions. Where should it live, and what will it automatically see because of where you put it?
- (Apply — objective level) Describe your real two-project structure right now: what's account-wide, what's project-specific, and which chat goes where, for the next week of work.
Ledger write
ledger_write:
learner_id: L3-CONS-PRIYA
lesson_id: L3-11-D1
standards: [AIHC.1.D1]
tags: [P5]
exit_ticket:
score:
bloom_reached:
auto_score:
self_score:
calibration_gap:
journal_prompt: >
Before this lesson, how many of your real chats were sitting outside the right project? What
would you have lost — provenance, confidentiality, or just time — if you'd kept working that
way for another month?
structure_used: PBL
referents_used: [cricket-IPL]
next_lesson_seed: "this project structure is exactly what Lesson 12 (D2 Regeneration) asks you to re-test whenever a new model or feature ships."
Rubric self-audit (R1–R13)
| # | Indicator | Verdict | Evidence |
|---|---|---|---|
| R1 | One Bloom's-leveled objective, ≥1 named AIHC standard, learner-visible | PASS | Objective names Apply + AIHC.1.D1, with the P5 scale note stated openly |
| R2 | Every product claim traces to the frozen doc-set; no invented UI | PASS | S03/S04 (projects, knowledge base, moving chats, starring), S12 (chat search scope), S13 (instruction layering) all quoted |
| R3 | 3–6 SSD cycles, complete | PASS | 4 cycles: instruction layering, chat filing, second project + starring, shared-knowledge-base chats |
| R4 | Each SEE anchors its SAY | PASS | SEEs describe only stated UI (instruction layering diagram, chat-history list, starred sidebar, project chat view) |
| R5 | Every DO acts on the learner's real work | PASS | Real Profile Instruction, real mis-filed chat, real second project, real new chat |
| R6 | Media doctrine | PASS | No motion; static/annotated only |
| R7 | Exit ticket 3–5 Qs, Bloom's-climbing, SSOT-graded | PASS | 4 Qs, Remember→Apply |
| R8 | Ledger write complete | PASS | anchor codes, scores, journal_prompt present |
| R9 | Scaffolding with fade; at-bat present | PASS | Cycles 1–3 scaffolded, Cycle 4 lighter, at-bat (full re-audit) unscaffolded |
| R10 | Referents elected-only, flavor-only | PASS | Cricket used once, positioning analogy only |
| R11 | Timing-tolerance honored | PASS | Pause points after Cycles 1–3 |
| R12 | Non-replication | PASS | Keyed to Priya's specific two real projects and her own chat history — no generic template |
| R13 | Consumer trust boundaries named, never over-reassured | PASS | Cycle 2 explicitly ties mis-filed chats to real confidentiality/provenance risk from prior lessons rather than treating filing as cosmetic |
Escalation verdict: all load-bearing indicators PASS → clears; auto-ships.