A Simple AI Design Workflow If You Feel Behind
For designers who feel behind on AI tools · Based on Wooldridge Connected AI Design Workflow
// TL;DR
Designers who feel overwhelmed by AI tooling can use the Wooldridge Connected AI Design Workflow to escape tool sprawl with a lean, connected stack. Instead of collecting fifteen disconnected tools, you master a few — ambient meeting capture, a Notion hub, an AI agent with MCP connectors, and voice dictation — where each feeds context into the next. Capture meetings passively, store structured summaries in one place, let your agent pull live context, and dictate prompts by voice. This gives you a calm, repeatable system that ships first-pass work faster, replacing the anxiety of falling behind with a clear, connected process.
Why do so many designers feel behind on AI?
Because they confuse having more tools with being ahead. Collecting fifteen AI apps that don't connect to each other leaves you overwhelmed and, ironically, feeling further behind. The Wooldridge Connected AI Design Workflow reframes the goal with its first principle: Connection Over Collection. The real value is never in any single tool — it's in how they connect, because each connected tool makes the next one more powerful. You don't need every AI tool; you need a few that talk to each other.
What's the minimum stack I actually need?
Five things: an ambient meeting transcription tool (Granola), a documentation hub as your Single Source of Truth (Notion), an AI agent with MCP connectors (Claude Code), your design and code tools (Figma, GitHub), and a local voice-to-text app (Spokenly). That's it. Everything else is optional. This lean stack covers the whole arc of a design project — capturing context, organizing it, and executing against it — without the anxiety of juggling a dozen disconnected apps.
How do I start without getting overwhelmed?
Start with one project and one meeting. Let your transcription tool run quietly in the background during a call — don't join a bot, just let it capture. Afterward, retrieve the structured summary of decisions, action items, and open questions, and drop it into a single Notion page. That's the entire first step, and it already gives you a clean foundation instead of scattered notes. You've now practiced Context-First Execution — capturing context early so your AI never starts from a blank slate.
How do MCP connectors make this easier, not harder?
MCP connectors sound technical but remove work rather than adding it. In Claude Code, go to Customize → Connectors → Browse Connectors, connect Notion and Figma, and authenticate — a one-time setup. From then on, your agent reads your project context directly, so you never copy-paste again. Instead of manually feeding the AI, you just say "Based on my notes in Notion, build a first-pass screen flow," and it pulls everything itself. Less friction, not more.
How does voice dictation lower the barrier further?
Speaking is three to four times faster than typing, and it removes the intimidation of crafting long, perfect prompts. Hold the hotkey, speak naturally like you're explaining to a colleague, and release. Your words appear in the prompt field. When the agent's first pass misses, you don't type a wall of corrections — you just talk through the changes. This makes the whole system feel conversational instead of technical.
What should I not do?
Don't try to adopt every tool at once — that's the exact trap that made you feel behind. Don't feed raw transcripts to your agent; only structured summaries belong. Don't skip the single hub. Don't expect the first pass to be final — iteration is the whole point. And don't use inaccurate cloud dictation that garbles your words; choose a local model.
Next step
Stop collecting. Pick one current project, set up just the five-tool stack, and run a single meeting through the workflow this week — capture it, store the structured summary in Notion, connect the tools once, and dictate one context-grounded prompt. Prove to yourself that a small connected system beats a big disconnected one.
// FREQUENTLY ASKED QUESTIONS
I'm overwhelmed by AI tools — where do I even start?
Start with one project and one meeting. Run an ambient transcription tool during a call, retrieve the structured summary, and drop it into a single Notion page. That one habit already gives you a clean context foundation. Add MCP connectors and voice dictation only once the capture step feels routine. The goal is a lean connected stack, not more tools.
Do I need to be technical to set up MCP connectors?
No. In Claude Code you go to Customize → Connectors → Browse Connectors, pick your tools, and authenticate — it's a guided, one-time setup per workspace. After that, connectors remove work by letting the agent pull context automatically, so you never copy-paste again. It's designed to reduce friction, not add technical overhead.
Will this actually make me feel less behind?
Yes, because it replaces tool anxiety with a repeatable process. Once you have a few connected tools that feed each other, you stop chasing every new app and start shipping first-pass work faster. Feeling behind usually comes from disconnected collection — a lean connected workflow fixes the root cause, not just the symptom.