RXed AI News

We x-ray the AI industry.
@RXed_EU
Latest auditsAI VersusMethodologyUpcoming & requests
SCORE 5/1014/20 elements Audited 2026-07-31 · RXed table v1.0

Framer AI

“AI website builder for designers and teams” — the vendor’s own words

The AI agent genuinely edits a live, publishable site instead of spitting out a mockup — rare in this category — but there's no open MCP, no code export, and the $10 headline creeps fast once seats and credit tiers stack up.

SMB toolAutomation & AgentsProductivityMarketingFreemium
Vendor
Framer B.V. · www.framer.com
Origin
NL — Amsterdam
Pricing
Free $0 · Basic $10/mo · Pro $30/mo · Enterprise Custom
Users (official only)
Over 500,000 monthly active users (self-reported at Series D) (source, 2025-08-28)
PRICING
Free$0500 agent credits to try, 1 GB bandwidth
Basic$10/mo1,000+ agent credits/mo, 2 CMS collections, custom domain, 50 GB bandwidth
Pro$30/mo3,000+ agent credits/mo, 10 CMS collections, staging, Branching with previews
EnterpriseCustomCustom credit volume, unlimited editors, SSO, SCIM, uptime guarantee

Additional editor seats +$20/mo each; Content Editor (CMS-only) seats +$10/mo each. Agent credits are also sold as add-on tiers within Basic/Pro (up to 15,000/mo on Basic, 100,000/mo on Pro), so the effective monthly price is often well above the $10/$30 headline.

checked 2026-07-30 · vendor pricing page

Element scores

Reactive
Retrieval & Memory
Orchestration
Validation
Models
Primitives
Pr7
Prompts
Em—
Embeddings
Cx7
Context
Tr4
Tracing
Lg8
LLM
Compositions
Fc5
Function calling
Vx—
Vector store
Rg4
RAG
Gr6
Guardrails
Mm5
Multimodal
Deployment
Ag8
Agents
Ft—
Fine-tuning
Fw4
Frameworks & harnesses
Ev4
Evaluations
Sm5
Small models
Emerging
Ma—
Multi-agent
Sy—
Synthetic data
Pc2
Protocols
In—
Interpretability
Th6
Thinking models
Tap or hover any element to see why it got that score.

Strengths

Genuine multi-frontier model access (Anthropic and OpenAI) through a transparent picker with published efficiency work; the agent operates natively inside a real design canvas with CMS and component context rather than producing throwaway mockups; a real publish gate (Branching, canvas-only edits) keeps AI changes reviewable before they go live; and the External Agents bridge lets Claude Code, Cursor and Codex work inside a live project without exporting.

Honest dings

No open MCP server — Framer built a proprietary bridge instead and says so explicitly; native image and video generation is outsourced to third-party marketplace plugins rather than shipped; and independent reviews (Trustpilot, G2-sourced roundups) show a consistent pattern of slow, bot-first support plus pricing that escalates fast once editor seats, localization and agent-credit tiers stack up.

Best for: Design-led teams and startups that want an AI agent editing a real, publishable site rather than generating a mockup, and who are comfortable with vendor lock-in (no code export) in exchange for design quality and speed.
Visit framer.com Prices and details change — this passport is re-verified at least quarterly.

Sources

Every audit lists the research it rests on — transparency and traceability are the product. Tools evolve: each audit is a snapshot of its audit date, and re-audits supersede older versions (kept below for reference).

RELATED ON RXED
LEARN Multi-agent (Ma) Several agents on one job — rarely worth the 15x token bill. LEARN Function calling (Fc) The model answers with a tool call instead of prose — so software can act. LEARN Frameworks & harnesses (Fw) Agent = Model + Harness. The scaffolding decides whether the agent works.