The AI Landscape
A map of tools and what they're good at
You’ve built the mental models. Now let’s look at the landscape. AI isn’t just ChatGPT — it’s a rapidly growing ecosystem of tools, each with different strengths. Understanding the landscape means you can pick the right tool for the job instead of forcing everything through one chat interface.
The big three (and what makes them different)
OpenAI
The most widely used. Strong general-purpose assistant with a large plugin ecosystem. Versatile, fast, and constantly updated with new features.
Best for: broad everyday tasks, web browsing, custom GPTs
Anthropic
Excellent at long-form writing, nuanced analysis, and document processing. Known for following complex instructions well and producing thoughtful, structured output.
Best for: writing, analysis, working with long documents
Deeply integrated with Google’s ecosystem — Gmail, Docs, Search. Especially useful if you already live in Google’s world and want AI that connects to your existing tools.
Best for: tasks connected to your Google workspace, search-grounded answers
These aren’t interchangeable. Each has different strengths — jagged intelligence applies to models too. Worth trying the same task across providers to see the difference.
Purpose-built AI tools
Beyond the big chat providers, there’s a growing world of tools designed for specific jobs. Here are some worth knowing about:
Presentations
- →Gamma — AI-generated slide decks from a simple brief
Image generation
- →Midjourney, ChatGPT, Flux — create visuals from text descriptions
Coding
- →Cursor, Claude Code, GitHub Copilot, Windsurf, Codex — AI-assisted programming, from inline code completion to autonomous coding agents
This landscape changes rapidly. The mental models from this course help you evaluate any new tool that comes along — you don’t need to memorize a list that’ll be outdated next month.
Four form factors
AI tools generally come in one of four shapes. Knowing which form factor you’re using helps you set the right expectations.
Chat interfaces (ChatGPT, Claude.ai) — general-purpose thinking partners. Good for brainstorming, drafting, and ad-hoc tasks. You bring the problem, the AI helps you think through it.
Embedded AI (Notion AI, Canva’s AI features) — AI built into tools you already use. Good for workflow integration. You don’t switch apps; the AI meets you where you work.
Connected AI (Gemini + Google Workspace, ChatGPT + Zapier, Microsoft Copilot + M365) — AI that reads from and writes to your other apps. Instead of copy-pasting between tools, AI pulls context directly from your email, calendar, CRM, or documents — and can push results back.
AI agents (Claude with tools, custom GPTs) — AI that takes multi-step actions on your behalf. The frontier of what’s possible. Instead of answering one question, agents can browse, research, and execute across multiple steps.
AI that talks to your tools
One of the most practical shifts in AI right now: connectors. These are bridges that let AI read from and write to the apps you already use — email, calendar, documents, spreadsheets, CRMs. No copy-pasting. No context-switching.
Instead of pasting meeting notes into a chat window and asking for follow-up emails, imagine AI that already has your calendar context and drafts those emails directly. That’s what connectors enable.
Gemini + Google Workspace
Extensions connect Gemini to Gmail, Docs, Drive, Maps, YouTube, and more. Ask Gemini to summarize your recent emails or find a document — it pulls from your actual Google apps.
ChatGPT + Actions & Plugins
Custom GPTs can connect to external services via actions — from Zapier automations to company-specific tools. Some GPTs can search the web, query databases, or trigger workflows.
Microsoft Copilot + M365
Deeply woven into Word, Excel, Outlook, and Teams. Copilot can draft documents from meeting transcripts, build presentations from reports, and summarize email threads — all within the Microsoft apps you already use.
Before you connect
Connectors are powerful, but they expand what AI can access. Ask yourself: What data does this connection touch? (Read-only is lower risk than read-and-write.) What happens if AI gets it wrong? (Auto-sending an email is riskier than drafting one for your review.) Does your organization allow it? Check with IT before connecting work accounts — many companies have policies about third-party AI access.
Takeaway: You now have the mental model to evaluate any AI tool. Ask: What is it predicting? Where are its jagged edges? Is it deterministic or probabilistic? Start with one or two tools that fit your workflow and expand from there.
When you encounter a new AI tool (and you will — there are new ones every week), ask these questions:
1. What’s it actually doing under the hood? Is it an LLM wrapper, a specialized model, or something else entirely? Many “AI tools” are just a nice interface on top of the same models you already have access to. That’s not necessarily bad — but it’s worth knowing.
2. What’s its jagged frontier? What is it great at and what does it struggle with? Every tool has edges. The marketing page won’t tell you about them, so try it on a task you know the answer to.
3. Does it connect to my existing workflow, or is it another thing to check? The best AI tool is the one you’ll actually use. If it requires switching contexts or learning a whole new interface, the friction might outweigh the benefit.
4. What’s the privacy story? Where does my data go? Is my input used to train the model? Can I opt out? Especially important for work-related tasks with sensitive information.
5. What can it connect to? More tools now offer connectors — bridges to your email, calendar, documents, and other apps. A tool that connects to your existing stack is more useful than one that lives in isolation. But more connections also mean more data access, so weigh the convenience against the privacy tradeoff.
Under the hood, many of these connections use open protocols like MCP (Model Context Protocol) — a standard that lets AI tools securely plug into external services. You don’t need to understand the protocol, but if you hear the term, now you know: it’s the plumbing that makes connectors work.
The mental models from this course give you a framework for asking the right questions — regardless of what the tool is or how impressive the demo looks.
Need help bringing this to your team?
We work alongside your team to build AI-native workflows — from one-week sprints to full engineering acceleration. No handoffs, no slide decks.
Talk to us