Pick one AI seat this week.
Cancel the rest on evidence.

Don’t buy three coding AI seats — pick one primary agent

Copilot Business + Cursor Pro + Claude Max on the same tickets is a quiet $100+/mo hole. One primary agent surface; everything else is overflow with an owner.

Put Midjourney on campaign kickoff week — not always-on

Taste seats shine when you need direction. They are expensive wallpaper the other three weeks of the month.

Ship AI copy only after a named human signs the risk

Hallucination is cheap to generate and expensive to publish. Gate anything that touches brand, money, or customers.

AI game NPC / quest dialogue brief pack

Scene-oriented prompt pack for NPC voice, quest beats, and greybox dialogue — ties to AI-for-games pipelines.

SEO brief → outline with evidence placeholders (anti-farm)

Turn a keyword brief into a source-backed outline — every claim gets a citation slot before anyone drafts prose.

Switch-or-not verdict for any AI tool decision

Force a blunt keep / switch / trial answer with money risk — paste into ChatGPT or Claude before you renew.

AI freelancing claims: who actually gets paid

$10k/mo with ChatGPT gigs is usually client acquisition theater. Here's the claim, the evidence bar, real costs, and who should walk away.

ChatGPT

writing Free (capped) · Plus-class ~$20/mo · Pro 100 ($100) / Pro 200 ($200) / Pro 500 ($500). Pro 200 usage steps down ~Oct 30 2026 for non-grandfathered/new allowance; Pro 500 is the Ultrafast tier. Verify on chatgpt.com / OpenAI Help — voice, images, agents, and Ultrafast burn allowance faster than plain chat.

Pay for one familiar general seat this month — not a second writing IDE or image studio “just in case.”

Claude

writing Free · Pro (~$20/mo class) · Max (usage multipliers) · Team/Enterprise. API: Sonnet 5.5 list $2/$10 per MTok (cache read $0.20) — same sticker as Sonnet 5; Opus 5.5 $4/$20. Verify on claude.ai / Anthropic pricing — weekly/agent limits and effective $/task matter more than sticker.

Long-context judgment seat — pay when tone and multi-doc packs are the bottleneck, not for status.

Midjourney

design Paid GPU plans (Basic → higher tiers; Stealth on upper plans). No free tier. Verify on midjourney.com — Fast vs Relax hours decide whether the month felt cheap or punishing.

Buy Midjourney for campaign taste in a kickoff week — pause after; it is not your free doodle box or Brand Kit OS.

Canva Magic

design Free · Pro · Teams. Verify on canva.com — Magic features gate by plan.

When marketing needs assets tonight—not a design system tomorrow

v0

design Tied to Vercel/v0 plans (free tier + paid). Verify live on v0.dev / vercel.com — quotas and model access move.

Prompt-to-React UI for builders who ship—not a Figma philosophy club

Cursor

coding Free hobby · Pro-class individual · Teams Standard/Premium with dual usage pools. Verify on cursor.com — agent loops and premium model picks drive the real bill more than the seat sticker.

Buy Cursor only if multi-file agents are weekly work — not a third seat beside Copilot and Claude Max.

Claude Code

coding Usage / Claude subscription envelopes (Max common for heavy). Verify on Anthropic — weekly ceilings and unattended runs dominate cost.

Terminal/CI agent overflow — not a silent third daily driver beside Copilot and Cursor.

GitHub Copilot

coding Individual Pro/Pro+/Max · Business/Enterprise per seat + AI Credits (completions typically free on paid). Verify on GitHub — budgets beat hope.

Keep Copilot when GitHub is the factory — don’t also buy Cursor and Claude Max for the same tickets.