A new creator economy is building on user-made AI assistants — big opportunity, platform risk, and a likely reshaping of small software.
Illustration by IMF Alpha editorial · Reviewed by Pedro Marini
The big idea — two sentences
Custom GPTs — small, user-built AI assistants — have moved out of the lab and into everyday use. They’re replacing single-purpose SaaS in many niches, and that shift is already changing who ships software and how creators earn from it.
What’s happening now
Over the past year, major AI vendors put simple tools in people’s hands: assemble prompts, hook in APIs, publish an assistant. The result is a flood of focused GPTs — everything from contract-clause helpers to classroom tutors and Shopify description writers. These aren’t enterprise suites. They’re lean, conversational tools that solve one job very well — basically the old micro‑SaaS playbook, but with a model doing the heavy lifting.
What’s interesting here is how fast iteration happens. You don’t rewrite a codebase; you tweak prompts, swap a model, or add a connector. That changes the development rhythm.
Why this matters (beyond the hype)
A few concrete examples
The risks — why this might be a bubble (or worse)
How investors and incumbents are reading this
VCs are watching the usual SaaS metrics — retention, ARPU, acquisition costs — to see which GPTs scale into repeatable revenue. Big tech is not standing still: Microsoft, Google, Apple (indirectly) are building tooling or integrations. Winners could be platform-agnostic tooling companies, the platforms themselves, or a few independent creators who crack distribution.
What creators should do now
Why this feels familiar — and why it’s not the same
It echoes the early app-store era: new distribution, lots of low-cost creators, unpredictable winners. But the product isn’t just code anymore — it’s behavior and knowledge. That makes iteration fast but also fragile: a prompt tweak or model update can materially change what users get.
The short version
Custom GPTs are creating a new, low-cost creative economy: people can ship software-as-conversation and find ways to monetize it. That opens room for nimble innovation — and brings fresh problems around discovery, platform lock-in, and safety. For founders, the play is simple in principle: find a repetitive, stubborn task, build a lean assistant that genuinely outperforms templates, and spread your distribution bets.
If you’re building one, treat your GPT like a boutique product — focused, service-minded, and ready to change fast.
Draft guidance would require model audits, vendor controls and investor disclosures — a fast-moving shakeup for fintechs, banks and Big Tech.
From AutoGPT experiments to production pilots, autonomous agents are changing how companies automate knowledge work. The upside is real — so are the governance headaches.
SECURE 2.0 now forces Roth treatment on catch-up 401(k) contributions for higher earners — a stealth tax change many retirees will feel. Here’s what to do next.