This page compiles 3 files from the repository, verbatim, in reading order. The living version: this folder on GitHub.
level-1-essentials/00-start-here/README.md
AI Essentials Kit
A complete, opinionated system for taking a novice from zero to power-user across every major AI platform — and giving a company the governance, tooling, and rollout playbook to do the same at scale.
Built by Adam Abdalla · v1.0 · August 2026
What's in the box
ai-kit/
├── 00-start-here/ ← you are here
├── individual/ ← the personal track (build in order 01→09)
├── company/ ← the org track (build in order 01→10)
└── shared/ ← glossary, eval templates, checklists both tracks reuse
How to use this kit
If you're an individual
Read individual/01-foundation/README.md
Work sequentially through folders 01 → 09
Each folder has a README.md that tells you exactly what to do
Total time to complete the starter path: ~10 hours over 30 days
If you're a company
Read company/01-governance/README.mdfirst — governance before tools, always
Work sequentially through folders 01 → 10
Governance (01) and Accounts (02) are done by IT/Legal/Security
Department copilots (05) are built by department leads with the AI champion
Total time to a full rollout: ~90 days (see 10-rollout-playbook/)
If you want the jargon first
Skip to shared/glossary/master-glossary.md — it's the canonical 200-term reference.
The philosophy
Six rules this kit is built on:
Governance before tools. A team with policy but no AI beats a team with AI but no policy.
One prompt, three homes. Every good prompt lives as a Custom GPT, a Claude Project, and a Perplexity Space.
Evals or it didn't happen. No model change ships without running the eval set.
Route by strength, not by loyalty. Use the right model for the job — the cheat sheet in individual/05-model-routing/ tells you which.
Memory is a feature, not a magic trick. Curate what each model remembers; review monthly.
Jargon is leverage. Fluency in the terms is 30% of the skill gap.
Version and update cadence
v1.0 — August 2026 — initial release
Review release notes weekly (see individual/09-rituals/weekly-monthly-rituals.md)
Rebuild the eval set every quarter
Rotate API keys every 90 days
Quick index
I want to…
Go to
Set up my personal AI stack
individual/01-foundation/
Write my master system prompt
individual/03-system-prompts/
Grab the 20 starter prompts
individual/04-prompt-library/
Know which model to use when
individual/05-model-routing/
Learn the jargon
shared/glossary/master-glossary.md
Write the company AI policy
company/01-governance/acceptable-use-policy.md
Build a department copilot
company/05-department-copilots/
Roll it out in 90 days
company/10-rollout-playbook/
Set up evals
shared/eval-templates/
What to do next
Open individual/01-foundation/README.md or company/01-governance/README.md. Do not skip ahead. The order matters.
Same shape for individual and company — company items in [brackets].
Week 1 — Foundation
Day 1 — Gather foundation inputs (individual/01-foundation/foundation-inputs-checklist.md). [Company: publish AUP.]
Day 2 — Create accounts in priority order. [Company: sign DPAs, enable SSO.]
Day 3 — Write Master System Prompt. [Company: write Company System Prompt.]
Day 4 — Write Personal Style Guide. [Company: publish brand voice + disclaimer library.]
Day 5 — Build first Custom GPT with system prompt + 3 knowledge files.
Day 6 — Port same prompt + files to a Claude Project.
Day 7 — Port same prompt + files to a Perplexity Space.
Week 2 — Library and Routing
Day 8 — Copy in the 20 starter prompts. Pick your 5 most-used.
Day 9 — Rewrite those 5 in your voice.
Day 10 — Set memory in each model (ChatGPT, Claude, Perplexity, Gemini).
Day 11 — Learn model routing (individual/05-model-routing/). Print the cheat sheet.
Day 12 — Install local tools: Cursor or Claude Code, Raycast, ffmpeg, a clipboard manager.
Day 13 — First automation: one scheduled task or one Zapier scenario.
Day 14 — Weekly review: what worked, what didn't, prune 3 things.
Week 3 — Automation and RAG
Day 15 — Second automation. [Company: pilot department picks 3 workflows.]
Day 16 — Build first eval set (5 good, 5 bad artifacts per workflow).
Day 17 — Upload knowledge base to your Claude Project + Perplexity Space.
Day 18 — Test RAG: ask 10 questions, score answers against ground truth.
Day 19 — Third automation. [Company: instrument logging via LLM gateway.]
Day 20 — Try one model you don't normally use for a real task.
Day 21 — Weekly review.
Week 4 — Fluency and Ritualize
Day 22 — Jargon cluster 1: Model mechanics (see shared/glossary/).
Day 23 — Jargon cluster 2: Reasoning + Retrieval.
Day 24 — Jargon cluster 3: Agents + MCP.
Day 25 — Jargon cluster 4: Training + Multimodal.
Day 26 — Jargon cluster 5: Evaluation + Deployment + Governance.
Day 27 — Add 5 new prompts to your library based on real usage.
Day 28 — Rotate API keys, review spend.
Day 29 — Re-read your Master System Prompt. Update it.
Day 30 — Monthly review. Schedule next 30 days.
What "done" looks like at Day 30
All primary accounts active with SSO or password manager entries
Master System Prompt + Style Guide living in a canonical location
20+ prompt library, deployed as GPT + Claude Project + Perplexity Space
3 automations running
Eval set with 5×5 artifacts for at least 2 workflows
Model routing cheat sheet memorized (or printed above your monitor)
Glossary read, notes taken
Weekly and monthly rituals scheduled on your calendar
level-1-essentials/00-start-here/INDEX.md
Full File Index
00 — Start Here
README.md — this kit's philosophy and how to use it
INDEX.md — this file
30-day-starter-sequence.md — the 30-day plan for either track
Individual Track
01 — Foundation
README.md — what to gather before touching a model
foundation-inputs-checklist.md — the exact files to collect
privacy-line-worksheet.md — decide what you'll never paste into a model
02 — Accounts
README.md — which accounts to create and in what order
accounts-order.md — the priority list with cost estimates