Designing Your Personal AI Stack
Most people use AI haphazardly — one tool for everything, adding others occasionally when something catches their attention, never quite sure which to use for what. The result is a fragmented experience where they're never getting the best from any single tool. This lesson is about doing the opposite: deliberately designing a personal AI stack — the specific combination of tools you'll use, for specific purposes, with clear reasoning for every choice.
What an AI stack is
Your AI stack is the set of tools you've chosen for your specific life and goals. It covers every AI use case you have — and nothing more. The goal is not to have every tool — it's to have the right tools, each earning its place by being the best option for a specific set of tasks.
A well-designed personal stack typically has three layers:
- Core AI assistant: The tool you use for 80% of your AI work — general writing, thinking, research, conversation. Usually ChatGPT, Claude, or Gemini.
- Specialist tools: Tools that are genuinely better than your core assistant for specific tasks. Perplexity for research, DALL-E or Midjourney for images, Otter.ai for transcription.
- Embedded tools: AI features built into software you already use — Microsoft Copilot in Word, Notion AI in your notes, Grammarly in your browser.
Choosing your core assistant
| Tool | Strongest at | Consider if... |
|---|---|---|
| Claude | Long documents, nuanced writing, careful reasoning, admitting uncertainty | You write a lot, work with long texts, or value epistemic honesty |
| ChatGPT | Breadth of tasks, image generation (Plus), coding, the largest community | You need a general-purpose tool with the widest ecosystem and most tutorials |
| Gemini | Current information via Google Search, Google Workspace integration | You live in Google's ecosystem or frequently need up-to-date information |
| Perplexity | Research with cited sources, current information synthesis | Research is your primary AI use case and you want references you can verify |
Start with one core assistant and use it for everything for at least a month before adding specialist tools. Most people add tools too early — before they've exhausted what their core assistant can do. Depth in one tool beats breadth across five.
Mapping your personal use cases
Before finalising your stack, map your actual AI use cases. For most people they fall into four broad areas:
- Communication: Emails, messages, letters, proposals, presentations
- Research: Learning new topics, finding information, analysing documents, preparing for decisions
- Creativity: Writing, ideas, side projects, content creation
- Admin and organisation: Planning, scheduling, note-taking, information management
Help me design my personal AI stack. My main use cases are: [list your top 5–8 AI tasks]. I currently use [current tools]. My budget for AI tools is [free / £X per month]. I work primarily in [Google Workspace / Microsoft 365 / neither]. My most time-consuming tasks that AI could help with are [list 3]. Based on this, recommend my ideal stack: one core assistant, up to three specialist tools, and any embedded tools worth adding. For each, explain why it earns its place specifically for my use cases — and what I should cut if I'm currently using more.
The free vs paid decision
A clear framework for when upgrading is worth it:
- Stay free if: you use AI fewer than 5 times a day, your tasks are short and simple, or you're still building the habit.
- Consider paid if: you hit usage limits regularly, you need image generation, you work with long documents, or you want priority access during busy periods.
- Definitely upgrade if: AI is a core part of your daily work and the time saved by better access is worth more than £20/month — which, for most regular users, it is within weeks.
A personal AI stack has three layers: core assistant, specialist tools, embedded tools. Choose your core assistant based on your primary use cases. Map your actual use cases before choosing tools. Start with one tool and use it deeply before adding more. The free vs paid decision comes down to whether the time saved justifies the cost — for regular users, it usually does quickly.
Your Daily AI Routine
The difference between people who occasionally use AI and people who consistently get extraordinary results from it is not skill — it is habit. A daily AI routine turns occasional use into compounding productivity. This lesson shows you how to build one that fits your life.
Why routines matter more than skills
Most people use AI reactively — when they remember, when they're stuck, when someone mentions it. Reactive use means you never build the muscle memory, the prompt library, or the mental model that makes AI genuinely transformative. A routine changes this.
A good daily routine does three things: it makes AI use automatic rather than deliberate, it creates a consistent feedback loop so you keep improving, and it ensures your highest-value recurring tasks always benefit from AI assistance.
The morning briefing
A morning AI briefing takes 5-10 minutes and sets up your day. The goal is not to consume more information — it is to process what matters faster so you can focus on doing rather than reading.
I want a morning briefing on: [3-5 specific topics relevant to your work and life]. For each topic, give me: one key development from the past 48 hours, why it matters to me specifically, and one question I should be thinking about. Keep each section to three sentences maximum. Skip anything that is noise rather than signal.
The mid-day productivity block
A dedicated 20-30 minute mid-day AI block handles the tasks that accumulate during the morning — emails to draft, documents to summarise, decisions to think through. Batching these prevents context-switching and lets you use AI with full attention rather than squeezing it between other things.
The end-of-day review
A brief end-of-day AI conversation serves two purposes: processing the day and preparing for tomorrow. Asking AI to help you reflect on what happened, identify what got in the way, and plan tomorrow takes 5 minutes and compounds over weeks.
I'm doing a quick end-of-day review. Today I was working on: [brief summary]. What got done: [list]. What didn't: [list]. Help me: identify one thing that slowed me down today and how to remove that friction tomorrow, and draft a three-item priority list for tomorrow morning.
Routines beat skills for consistent AI results. Start with one daily touchpoint — not three — and expand once the habit is established. The morning briefing, mid-day productivity block, and end-of-day review are the three natural integration points for most people's working day.
Personal Automation: Saving Hours Every Week
Personal automation is the highest-value application of AI for individuals — but most people think it's only for developers or businesses. It isn't. With tools like Zapier and Make, you can build automations that connect AI to the rest of your digital life without writing a single line of code. This lesson shows you how.
What personal automation actually means
Personal automation is not about building complex systems. It's about identifying the small, repetitive tasks in your life that happen on a predictable schedule or trigger — and setting up a one-time workflow that handles them automatically.
The best personal automations share three qualities:
- They happen frequently. A task you do once a year isn't worth automating. A task you do every day or every week is.
- They follow a predictable pattern. Automation works when the input-to-output relationship is consistent. Creative, judgment-heavy tasks don't automate well — administrative, templated ones do.
- They're low-stakes enough to run without review. Or high-stakes enough that catching one error is worth the time saved on fifty correct outputs.
The highest-value personal automation opportunities
Getting started with Zapier or Make
Both tools offer generous free tiers. The concepts are identical: you define a trigger (something that starts the automation), one or more actions (things that happen in response), and connect them.
For AI-powered automations, both integrate directly with OpenAI and Anthropic via their APIs. You can pass text to an AI model, give it instructions, and receive the output — all within a no-code workflow that runs automatically.
I want to build a personal automation for the following task: [describe the task]. Currently I do this manually by [describe your current process]. It happens approximately [frequency]. I use [list your apps — Gmail, Notion, Google Calendar, etc.].
Help me design a simple automation for this. What should the trigger be? What are the steps? What AI instruction would produce the output I need? Flag any places where the automation might fail and how to handle them. Keep it as simple as possible — I'd rather have something that works reliably than something sophisticated that breaks.
Notion AI: the easiest entry point
If you already use Notion, Notion AI is the fastest way to experience AI-powered automation without setting anything up. It can summarise pages, convert voice memos to structured notes, write from templates, and clean up messy text. It's not a replacement for Zapier or Make — but it's the simplest possible starting point.
The biggest automation mistake is trying to automate everything at once. Pick one task — the most frequent, most predictable, most tedious — and build one automation. Run it for two weeks. Fix the edges. Then consider a second. Automation debt (broken workflows you don't maintain) is worse than no automation.
Personal automation is accessible to non-developers via Zapier, Make, and embedded AI like Notion AI. The best automations are frequent, predictable, and sufficiently low-stakes to run without constant review. Start with one automation: meeting notes → action items, or daily content → morning digest. Use AI to design your automations before you build them.
Managing Information Overload with AI
We live in an era of information abundance and attention scarcity. The average knowledge worker receives more information in a day than someone in the 1800s encountered in a lifetime. Most of it is noise. The question is not how to consume more — it's how to extract the signal that actually matters, efficiently and reliably.
AI changes the economics of this problem fundamentally. What used to take an hour of reading can now take five minutes of AI-assisted synthesis, with better comprehension and retention. This lesson covers building an AI-powered information system that keeps you current without consuming your attention.
The daily document digest
One of the most practical immediate applications: paste any long document — a report, a newsletter, a lengthy article — into your AI assistant and get a structured summary in seconds.
[Paste document here]
Give me a structured digest of this document:
1. The main argument or purpose in one sentence
2. The three most important points or findings
3. Anything that surprised you or that contradicts common assumptions
4. What I should do with this information, if anything
5. What it doesn't say that would be important to know
I am [brief description of your role/context]. Calibrate the relevance of points accordingly.
Building your personal intelligence briefing
Rather than reading everything, build a weekly intelligence briefing — a curated synthesis of what's happening in the areas that matter to your work, career, and interests. This doesn't require automation; you can do it manually in ten minutes once a week.
I want a weekly briefing on the following topics: [list 3–5 areas — e.g. "AI developments that affect personal productivity", "UK tax changes", "the publishing industry"]. For each topic:
— What are the most significant developments this week?
— What should I be paying attention to that most people are missing?
— Is there anything that changes how I should be thinking about this?
Base your response on your training knowledge and flag clearly which parts may be out of date. I'll verify time-sensitive claims independently.
AI-assisted reading: books and long-form content
One of the most underused applications of AI is as a reading companion. Before reading a book or long report, use AI to give you the conceptual landscape — the main argument, the key frameworks, the debates the author is engaging with. This pre-reading makes your actual reading significantly more efficient and more effective.
I'm about to read [book/report title] by [author]. Before I start, give me:
1. The central argument and why it matters
2. The key frameworks or models I should understand going in
3. The main criticisms of this work — what do thoughtful critics say is missing or wrong?
4. Two or three questions I should be asking as I read
5. What background knowledge would help me get more from this?
The goal of an AI-powered information system is not to process more information — it's to extract more signal from less time. Every tool and habit in your system should be evaluated on whether it reduces noise and surfaces what actually matters, not just whether it helps you consume more.
AI changes the economics of information processing — a one-hour read becomes a five-minute digest. The document digest prompt is your most immediately useful tool. Build a weekly intelligence briefing for the areas that matter to your work. Use AI as a reading companion — pre-reading preparation makes your actual reading faster and more effective. The goal is signal extraction, not volume processing.
Part 2 Exercise: Design Your System
This is your Part 2 exercise lesson. You have covered your AI stack, daily routine, personal automation, and information management. Now you put it together as the first draft of your Personal AI System document.
Open a new document and write the first three sections of your Personal AI System:
Section 1 — My AI Stack: List every AI tool you use regularly. For each one, write one sentence explaining exactly what task or need it serves that no other tool does as well.
Section 2 — My Daily Routine: Describe your actual AI touchpoints during a typical day. Be specific — what time, what task, which tool.
Section 3 — My First Automation: Describe one automation you have built or are planning to build. What does it do, what does it connect, and what does it save you?
This document will grow throughout the course. What matters now is starting it — even an imperfect first draft is infinitely more valuable than a perfect document that doesn't exist yet.
Mark this lesson complete to unlock your Part 2 completion. Keep your system document somewhere you will actually use it — not buried in a folder, but accessible daily.