Sanity vs Strapi

Sanity

The plugin page documents a graphical editorial calendar with at-a-glance schedule review and prioritization plus Schedule, Unschedule and Reschedule actions, but that plugin is marked deprecated, only runs on a deprecated Studio version, and its successor is merely named. The editorial workflows page adds stages, assignments and durable history, and a guide shows briefs with status, reviewer and usage-rights fields modeled by hand — but the publishing itself needs an external script or cronjob. We found no public information on campaign or channel grouping, per-person workload views, or recurring formats as scheduled series.

Strapi

The captured campaign-management page talks about content blocks, the content-type builder and review workflows rather than any calendar, brief or campaign view, and the strongest planning-adjacent evidence is the Content Manager list filtered by assignee and review stage. That is a status list, not a scheduled plan — I found no public information on due dates, topics, briefs or workload views.

Sanity

Comments and tasks on documents are documented with per-document-type configuration, mention permissions and plan availability, and the editorial workflows page gives configurable stages, assignments, decisions and durable history that can be driven from servers, the CLI or agents over MCP. A complete change history with plan-dependent retention is stated in the privacy policy, while the full audit trail and History API sit only on the Enterprise plan. We found no public information on mandatory reviewers such as legal or brand, guest access restricted to assigned pieces, or scheduled publishing that explicitly waits for approval.

Strapi

Review workflows are genuinely configurable: multiple workflows, up to 200 stages each, per-stage role permissions on who can move content in or out, a required stage before publishing, and every stage change, assignee change and workflow change written to Audit Logs. The captured pages do not show inline suggested edits, external contributor access, or publishing that waits on a schedule, so it stops short of the top band while the approval machinery and audit trail are documented in real depth.

Sanity

This is the part I can verify from the API side: content is stored as structured JSON in the Content Lake, queryable with GROQ and GraphQL, with a documented modeling approach, references between entries enforced in schema, and unlimited content types and locales on all plans. The terms promise an export of subscriber data in the standard database export format within thirty days after termination. We found no public information on field-level localisation with fallback languages, translation workflow statuses or named translation-service connectors, and no documented migration tooling that works in both directions.

Strapi

This is where the headless name is earned: collection types, single types, repeatable components, dynamic zones, six relation types and per-content-type internationalisation, all exportable through a documented CLI that packs schemas, entities, relations, assets and configuration together, with JSONL output and data import/export/transfer in the other direction. Field-level localisation with fallback languages and named translation-service connectors do not appear on the pages I read.

Sanity

The delivery story is real, not a logo swap: an API-first decoupled architecture serving any consuming surface, GROQ and GraphQL endpoints, a CLI, a plugin install mechanism, and Live CDN, Functions, an App SDK and an MCP server all named as products, with webhooks including GROQ-powered webhooks. Usage limits for API and CDN requests are published on the pricing page. We found no public information on API versioning or a deprecation policy, a public status page, or named connectors for CRM tools — the long integration list covers front-end frameworks rather than a marketing stack.

Strapi

The API story is the best-documented part of the product: auto-generated REST and GraphQL APIs per content type with an OpenAPI specification, bearer-token auth, a client library, and webhooks on create, update, publish, unpublish and delete with fully populated payloads, configurable headers, token auth and HMAC signing guidance. I found no public information on rate limits, API versioning or a deprecation policy, and the only preview capability shown is Live Preview on the Growth plan, so delivery is not yet dependable by contract.

Sanity

AI Assist features are described concretely — long-form drafting, document and field translation retaining formatting, image generation and variants — and the AI terms name OpenAI, Anthropic, Pinecone and Google as third-party providers while stating that all configurations explicitly prohibit or disable training on transferred input data. Beyond that the governance thins out: we found no public information on an admin switch to disable AI features per role or content type, EU processing locations for AI, ownership wording for generated output, or marking of generated content for EU AI Act transparency, and the terms allow the vendor to change AI partners and consumption rates without notice.

Strapi

The supplementary AI terms are candid that input is transmitted to third-party AI provider APIs whose policies apply, and that output may not be copyrightable and may be the same output given to others, with the integration provided as-is, outside the SLA and terminable without refund. No model is named, nothing states that customer content does not train models, and I found no public information on prompt retention, processing location, an admin off-switch or EU AI Act labelling. A bring-your-own-AI-key topic in the documentation is the one lever of control I can see.

Sanity

A data processing agreement is offered and the privacy policy commits to the GDPR with standard contractual clauses for transfers, but the terms are governed by California law with exclusive venue in San Francisco under Sanity US Inc., the vendor's own site is hosted in the United States, and customer AI input data goes to the US providers OpenAI, Anthropic and Pinecone. Google Cloud Platform is named as a subprocessor without a stated region, so EU hosting of content and assets remains unproven rather than disproven.

Strapi

The contracting entity is Strapi Inc., a Delaware corporation under California law, while the privacy policy names the United States and the EEA as storage regions, appoints an EU GDPR representative at a Paris address, and publishes a list of third-party service providers. Self-hosting under the MIT licence with full control over data residency, plus region selection on Strapi Cloud, are real options, but EU hosting for content, AI processing location and per-subprocessor locations are not established on the captured pages.

Sanity

The pricing page is unusually computable: monthly and per-seat figures with included limits, overage rates for API requests, CDN requests, bandwidth, assets and compute, AI credits priced at "$0.05 PER AI CREDIT" with published per-action credit costs, and add-ons like "+ $299 PER MONTH" for extended quotas and "$999 PER DATASET / MO", with Enterprise named as the plan where SSO and the full audit trail begin. Self-serve plans are monthly only with no minimum term, and renewal, price-increase and non-renewal notice terms are stated in the terms of service. Enterprise is custom-priced and we found no public information on the cost of exporting content, which is what keeps a full annual cost from being derivable for everyone.

Strapi

Community is "Free Forever" and Growth is $45 per month including 3 seats at $15/month/seat extra, with unlimited entries, API calls, locales and roles, AI priced at 1,000 credits per month plus $1.50 per 100 credits, and Single Sign-On as a $150 /month (+$50/month/seat) add-on. But Review Workflows and Audit Logs sit on the Enterprise plan, whose price is not published — a captured page even poses the question "Is Strapi Review Workflows free?" without a stated answer — so the buyer who needs approvals cannot compute their annual cost.