The UX Designer's AI Stack: 7 Tools That Actually Work Together

You finish a user interview. Fathom has the transcript. Figma has the wireframe. ChatGPT has your synthesis notes. Notion has last week's research. None of them know about each other. That 45-minute reconnection task every Tuesday morning isn't an AI problem — it's a stack problem.

Seven UX practitioners who documented their complete AI setups, including a Senior Designer at Wise who shipped three production apps in 2025 on a $40/month stack, all converged on the same pattern independently: route messy thinking through one LLM, pass structured output to one specialist tool per stage, automate the handoffs. None learned this from a tutorial. They each cancelled five to ten tools first.

The verdict upfront: for most solo UX designers, the working stack is ChatGPT Plus plus Fathom plus Figma plus Notion plus Make, totaling $52–61/month — or $0 on free tiers. The tools most commonly cancelled after three months are Dovetail, Otter purchased alongside Fathom, and the second LLM subscription. The single highest-ROI first move is Fathom, free, connected to your next interview.

Everything after that is expansion, not foundation. Here's how the stages connect.

The Five-Stage Stack, Tool by Tool

Stage 1 — Research: Your LLM and Your Transcription Tool

The stack's brain is whichever general LLM you choose, and this is the most consequential decision in the system. ChatGPT Plus ($20/month) and Claude Pro ($20/month) handle roughly 80% of the same tasks — discussion guide drafts, screener questions, synthesis from pasted notes, stakeholder email framing. Practitioner consensus from the research: pick one for 60 days before deciding. ChatGPT wins for breadth, including image generation and a native Figma integration. Claude wins for long-document work — its 200K context window handles 20 interview transcripts in one paste without losing the thread. For most UX designers starting out, ChatGPT Plus is the default. Subscribing to both simultaneously is the first thing practitioners cancel.

Worth noting: ChatGPT's free tier (GPT-4o mini) handles early-stage research tasks well enough to test the workflow before spending anything.

Fathom is the most under-the-radar tool in this stack and the one with the clearest immediate ROI. The free tier gives individuals unlimited transcription — genuinely unlimited, not a trial — for Zoom, Google Meet, and Microsoft Teams. It joins calls automatically, transcribes with roughly 95% accuracy on clean audio, produces a summary and action items, and emails a recap within five minutes of the call ending. Compared to Otter ($8.33/month on annual billing) and Fireflies, Fathom's free tier is the most generous, and it doesn't put a visible bot in the room.

The honest limitation: Fathom's free tier doesn't function as a research repository. It captures and summarizes — organizing and cross-referencing sessions is Notion's job.

Stage 2 — Create: Figma and Its AI Layer

Figma needs no introduction, but Figma Make does. It's Figma's native text-to-UI generator, included with paid plans and partially available on the free tier. Type a prompt — "3-column dashboard for a B2B analytics tool, using our existing component library" — and get three wireframe variants in under a minute. Every practitioner in the research who used it said the same thing: "It gets me to 60%, then I drive." Figma AI separately handles layer renaming, microcopy generation, and variant creation inside existing files — features that save the most time in daily practice, not the headline generation feature.

Free tier covers three files, which is adequate until you're juggling multiple active projects. Upgrading to Pro ($12/month) is the right call the moment you hit that file limit, not before.

Relume (free tier, Pro $20/month) solves a specific problem: the blank-canvas moment before wireframing begins. Its AI generates a complete sitemap from a one-sentence brief, then expands into wireframe-level page structures that export directly into Figma. Practitioners who compared it to Uizard settled on Relume because it respects information architecture rather than jumping straight to visual UI. It shines for marketing sites and landing pages; it's less useful for complex product application design with custom interaction patterns. If your work skews heavily toward app UX, you can skip Relume entirely.

The tools that stuck are the ones that solve a specific friction point, not the ones that promise to replace a workflow.
— Nurkhon, Senior Product Designer

Stage 3 — Review: Turning Research Into Shareable Decisions

Every UX designer has the same recurring tax: converting research synthesis into a stakeholder deck. Gamma (free tier, Plus $8/month) is the most consistently cited solution. Paste in a Notion summary or a Claude synthesis, describe your audience, and Gamma produces a full slide deck in under five minutes — one that looks considered, not AI-generated. The free tier allows ten AI deck generations; the $8/month Plus tier removes limits.

The data flow is simple: Notion synthesis → Gamma → shareable stakeholder link. No PowerPoint reformatting. Gamma decks work best for research readouts and project updates. They're less useful for complex design presentations where custom visual hierarchy matters.

Stage 4 — Collaborate: Notion as the Stack's Memory

Notion is where the stack accumulates knowledge over time — research notes, design rationale, meeting summaries, action items. The free tier is functional and genuinely generous. Plus ($10/month) with the Notion AI add-on (an additional $10/month) enables Q&A across the entire workspace ("What did users say about the checkout flow in Q3?"), AI-powered summarization, and automatic theme clustering across multiple research sessions.

The most important workflow pattern in the whole stack: Fathom summary lands in email → Make pushes it to a Notion database → Notion AI clusters themes across all sessions → Claude or ChatGPT writes the synthesis from the Notion summary. That chain is what makes the stack feel like a system rather than five separate apps.

One honest limit practitioners raised: Notion AI can't access Figma files or external documents. Everything that needs synthesizing has to live in Notion first.

Stage 5 — Automate: Make as the Connective Tissue

Make (free tier, Core $9/month) is what turns five separate subscriptions into one integrated workflow. The free tier includes 1,000 operations per month — enough to run three or four basic automations at low volume. Two automations eliminate the most manual work for UX designers specifically: Fathom transcript ready → automatically create a new Notion database row with the transcript, summary, and date; Figma comment tagged with a keyword like "?question" → create a Notion action item. Once these run, the daily copy-paste tax disappears.

Zapier does the same thing with a more polished interface at a higher price ($20/month for the equivalent tier). n8n is free and self-hosted but requires technical comfort to set up. For a solo UX designer who wants low friction, Make Core at $9/month is the right call. Add Make after you've been using the rest of the stack for four to six weeks and know exactly which handoffs are costing you time — not before.

What This Stack Costs at Three Budget Levels

At $0/month, the free tiers of ChatGPT, Fathom, Figma (three files), Notion, Make (1,000 operations), and Gamma (ten AI decks) cover roughly 80% of a solo UX designer's daily workflow. Gaps appear quickly: GPT-4o mini hits a quality ceiling on complex synthesis, Figma's file limit becomes constraining within weeks on active projects, and Gamma's deck limit runs out fast if you present weekly. The free tier is genuinely useful for running one full project through all five stages before spending anything. It's not a long-term solution for anyone billing clients.

The mid-range solo stack runs $52–61/month: ChatGPT Plus $20 plus Figma Pro $12 plus Notion Plus $10 plus Notion AI add-on $10 plus Make Core $9, with Fathom staying free. This is where practitioners in the research settle after three to six months of experimentation. If budget is tighter, drop the Notion AI add-on and use Claude's free tier for manual synthesis — the workflow still runs with one more paste step. Benhur Senabathi at Wise shipped three production apps on a $40/month stack (Claude Pro plus Cursor), while comparable practitioners spent $150–250/month on larger stacks, with no clear correlation between spend and output quality.

The premium solo stack runs $110–130/month and makes sense when you're regularly synthesizing 10+ interview transcripts at once (Claude Pro's context window becomes genuinely useful at that volume) and building more than ten stakeholder decks monthly. Beyond that, you're buying overlap.

The tools most commonly cancelled after three months: Dovetail or Looppanel (powerful research repositories at $30–39/user/month, but only valuable at three or more research studies per quarter); Otter or Fireflies purchased alongside Fathom; and both ChatGPT Plus and Claude Pro in month one before either has proven its specific value.

Where the Stack Still Requires Human Work

Three friction points are predictable enough to name before you hit them.

The research-to-design gap is the most-requested workflow improvement that doesn't yet exist cleanly. Dovetail and Figma have a native embed, but connecting a specific research insight to a specific design decision in a Figma frame is still manual — a two-minute task, not a two-hour one, but a manual step that surprises people who expect full automation.

AI synthesis hallucinations are real. NN/g specifically warns that AI summaries can misattribute quotes, merge distinct participant perspectives, or overstate pattern strength. The rule every practitioner applying AI to research learned: always verify quoted insights against the original Fathom transcript before including them in a stakeholder document. Treat AI synthesis as a first draft, not a final output.

Coding is about syntax. Orchestration is about intent, systems, and knowing what 'done' looks like.
— Benhur Senabathi, Senior Designer at Wise

Context doesn't travel between tools. When you switch from ChatGPT to Figma to Notion, the AI in each tool has no memory of what you did in the others. Benhur Senabathi's solution — a plain-text project brief he loads at the start of every AI session — is the most practical workaround available. Keep a one-page Notion document with the current problem, research questions, and design constraints, and paste the relevant section into each AI tool at the start of a work session. Thirty seconds of overhead, significant improvement in output quality.

Which Stack Is Yours

If you run user interviews and spend hours on notes and synthesis, start with Fathom free → Notion free → ChatGPT Plus in that order. These three eliminate the two biggest manual tasks in research workflows before you spend a dollar on anything else.

If you design more than you research and your main pain is blank-canvas wireframing, Figma Pro plus Figma Make plus Relume's free tier is the right starting point. Add ChatGPT Plus for prompt writing support once you've hit the LLM quality ceiling.

If you present research findings to stakeholders weekly, add Gamma to whatever stack you already have. At $8/month, it pays for itself in the first deck.

If you're running all of the above and the copy-paste between tools is the tax that costs you the most time, Make Core at $9/month with the Fathom-to-Notion and Figma-comments-to-Notion automations is the highest-leverage upgrade at any budget level.

Watch for one shift in the next 12 months: AI that travels between your tools — a Figma session that knows what your Notion research said, a ChatGPT conversation that knows what your Figma file looks like. Every major tool in this stack has published roadmap intent toward connected context. When that arrives, the manual project brief paste disappears. Until then, this seven-tool stack — with deliberate handoffs between stages — is the most reliable system working UX designers have built and documented in practice.

Recommended Tools & Resources