Plain
A Python web framework that grew out of a real company's Django app, built explicitly so AI coding assistants (and humans) can understand and safely modify the code.
🔗 Visit Plain
Description
If you've ever handed a messy Django or Rails codebase to an AI coding assistant and watched it get confused about which of five different ways to do something it should use, you'll get why Plain exists. Plain is a Python web framework designed from the ground up so there's exactly one obvious way to do most things, with types and structure explicit enough that both a human teammate and an AI agent can read the code and know exactly what it does.
Plain evolved out of Django through years of production use at PullApprove (the company behind it, Dropseed), stripped down to PostgreSQL-only storage, a typed ORM, htmx and Tailwind CSS on the frontend, and Jinja2 templates, with 30+ first-party packages covering auth, REST APIs, background jobs, email, caching, and admin panels so you're not stitching together a dozen third-party libraries. It ships with built-in OpenTelemetry observability and slow-query detection, a pytest-based testing setup, and agent-specific tooling: rules files, accessible docs, and slash-command workflows meant to help AI coding agents produce correct Plain code on the first try. It's open-source under a BSD-3 license, requires Python 3.13+, and there's an optional commercial add-on (plain.connect) for exporting observability data to a hosted dashboard. It has over 1,000 GitHub stars but is still a young, single-company-backed project.
💬 Our review
The short version: if your team is leaning hard on AI coding agents and tired of Django's flexibility working against the AI instead of for it, Plain's opinionated, typed-and-explicit design is a genuinely different bet worth evaluating, just know you're adopting a framework with one primary maintainer-company behind it, not a decade-old ecosystem.
Against Django, Plain trades away a lot of flexibility (PostgreSQL only, no swap-in ORM alternatives) for predictability, exactly the tradeoff that helps an AI agent generate correct code instead of guessing between equally valid patterns. Against FastAPI, Plain is a full batteries-included web framework (auth, admin, background jobs) rather than a minimal API layer, so it's a different category matchup: pick FastAPI for a thin API service, Plain for a full web app you want an agent to help maintain long-term. The catch is maturity: Plain is a relatively young open-source project maintained primarily by Dropseed/PullApprove, so you're taking on more single-vendor risk than with Django's much larger contributor base, and the commercial observability tier (plain.connect) is unproven at scale. Free and open-source to start, so the downside of trying it is mostly your own time.
💰 Pricing
📊 Global score
🤖 AI-enriched data
Framework open-source sous licence BSD-3, gratuit. Option payante : plain.connect (export observabilité vers Plain Cloud), tarifs non publiés.
Pros
Une seule façon évidente de faire les choses — pensé pour être lisible par un agent IA
30+ packages first-party (auth, API REST, jobs, email, cache, admin)
Observabilité OpenTelemetry et détection de requêtes lentes intégrées
Open-source, gratuit, licence BSD-3
Cons
PostgreSQL uniquement, pas d'alternative d'ORM
Projet jeune, porté par une seule entreprise (Dropseed)
Écosystème de contributeurs bien plus restreint que Django
Exige Python 3.13+