AI tools every studio should use (a 2026 stack)
A practical AI stack for a small creative studio. The ones we use, why we use them, and what we have stopped using.
AI tools are everywhere in 2026. Most are forgettable. A few are genuinely useful for a small studio. Here is the stack that actually moves the needle.
Writing and content
Claude (Anthropic). First draft of long-form content, client emails, proposals. The output sounds the most natural of the major models — less AI tell than competitors. Used for thinking out loud as much as for finished drafts.
ChatGPT (OpenAI). Better for code-related tasks and structured output (JSON, lists, schemas). We use both Claude and ChatGPT for different roles rather than picking one.
Notion AI / Notion Q&A. Internal documentation search. When the studio knowledge lives in Notion, AI-assisted search beats traditional search.
Design
Figma + AI plugins. Diagram and Magician are useful for generating placeholder content and rough icon variants. Real design still happens in Figma proper.
Midjourney. Mood boarding, concept exploration, hero image direction. We rarely ship Midjourney output directly but use it heavily for getting concepts in front of clients in the first hour.
DALL-E / Stable Diffusion (via Hugging Face). For specific image needs that need licensing clarity.
Vectorizer.AI. Converting rasters to vectors. One of those tools that does one thing and does it well.
Development
Cursor / Claude Code. AI-assisted IDE. Genuinely useful for accelerating code production. The studios doing this well treat AI as a junior pair, not as an autopilot.
v0 (Vercel). Generating React component scaffolding. Good first drafts, hand-finishing required.
GitHub Copilot. Inline completion. Still the baseline for any developer in 2026.
Project management
Linear. Not AI per se, but its AI-powered features (auto-prioritisation, summary, search) actually work.
Granola. Meeting notes. Transcribes calls, summarises, extracts action items. Replaces the role of "person taking notes" in client meetings.
Research and competitive intelligence
Perplexity. Faster than Google for research that needs synthesis. We use it for industry research, competitor analysis, finding obscure references.
Exa (formerly Metaphor). Semantic web search. Better than Google for "find sites like this one" queries.
Things we have stopped using
AI website builders for client work. Output is too generic for studio-quality projects.
Stock photo upscalers. Real photography or Midjourney has replaced stock for most use cases.
AI logo generators. Output is recognisable as AI logo. Bad for brand work.
Heavy AI copy editors. Tools like Grammarly's AI mode flatten voice. We've gone back to manual editing.
Stack cost
For a small studio (2-5 people), the total AI tool spend runs $200-500/month. That covers ChatGPT Pro/Claude Pro/Cursor/Midjourney/Notion AI. Roughly the cost of one good freelancer-day per month, in exchange for 20-30% productivity uplift across the team.
Honest principle
Pick tools that integrate into existing workflows. Tools you have to remember to open get forgotten in a week. The best AI tools are the ones that show up where the work already happens.