When a platform raises $85 million to evolve beyond a headless CMS and relaunch itself as a Content Operating System, it is worth paying close attention. Sanity has always been the developer’s CMS of choice — flexible schemas, a fully customizable React-based Studio, and a GROQ query language that gives developers granular control over every content request. But the headline question for most teams evaluating Sanity today is no longer purely about developer experience. It is about two capabilities that define how modern content teams actually operate: real-time collaboration and AI-assisted authoring.

This review breaks down both. We examine how Sanity’s collaboration layer compares to alternatives, how far its AI tooling has come with Canvas, AI Assist, and Content Agent, and where those features complement or compete with each other in a real workflow. Whether you are a product team managing complex content models, a digital marketing agency scaling content output, or an enterprise evaluating structured content infrastructure, this is what you need to know before committing.

What Is Sanity? From Headless CMS to Content Operating System

Sanity is a developer-first headless CMS built around a cloud-hosted, structured content store called the Content Lake. Instead of locking your content inside page templates, it stores everything as queryable JSON — meaning the same content can power your website, mobile app, digital signage, or third-party integration from a single source. Developers define their own content schemas in TypeScript, query data via GROQ (Sanity’s purpose-built query language) or GraphQL, and build their frontend with any framework they choose.

The editing environment is Sanity Studio, an open-source React application that teams deploy and own. Because it is built in React, every aspect of it can be customised — input components, editorial views, sidebar layout, and workflow states — without forking the platform or waiting for vendor updates.

The bigger story for 2025, however, is the platform’s strategic shift. In May 2025, Sanity unveiled its Spring Release 2025, backed by an $85 million Series C funding round, and formally rebranded its ambition as a Content Operating System. This release introduced Canvas (an AI-powered authoring environment), a unified Dashboard, a centralised Media Library, the App SDK for custom workflow applications, and Agent Actions for programmatic AI operations inside content pipelines. Sanity is no longer positioning itself as a place to store text. It is positioning itself as the operating layer for all content activity — creation, management, automation, and AI-driven delivery.

Real-Time Collaboration: How It Actually Works

Sanity’s collaboration engine is not a feature bolted onto an existing editing interface. The Content Lake stores documents in a real-time database using operational transforms — the same underlying technology that powers Google Docs. Multiple editors can work on the same document simultaneously, with cursor positions and active users visible at all times, and changes syncing instantly without requiring a page refresh. There are no save conflicts, no locked documents, and no “someone is editing this” warnings that interrupt workflows.

The practical effect for distributed teams is significant. Marketing can revise campaign copy while a developer updates schema fields. A legal reviewer can annotate a document while a copywriter continues drafting the next section. Every change is tracked, versions can be compared side-by-side, and anything can be restored. For time-sensitive editorial teams — those running news coverage, campaign launches, or product announcements across multiple markets — this is a genuine productivity multiplier, not a marketing claim.

Beyond real-time editing, Sanity’s collaboration layer includes:

• Presence indicators — colour-coded markers showing which users are active and which specific fields they are editing
• Inline comments and task assignment — reviewers can leave feedback, assign tasks with due dates, and resolve threads directly inside the document
• Content Releases — bundle changes across multiple documents, preview the entire release together, and publish on a schedule or on demand
• Role-based access control — custom roles per content type, with field-level permissions available on paid plans
• Version history — a field-by-field diff view before publishing, so mistakes are caught before they go live

It is worth noting that Sanity has been rated the number one headless CMS on G2 for four consecutive years, with seamless collaboration frequently cited as the primary differentiator. For agencies and brands running distributed teams across multiple time zones — a common reality for companies operating across Southeast Asia and beyond — this kind of built-in, zero-configuration collaboration removes a whole category of workflow friction that other platforms address through expensive add-ons or manual workarounds.

AI Authoring in Sanity: Canvas, AI Assist, and Content Agent

The AI authoring story in Sanity is split across three distinct tools, each serving a different part of the content workflow. Understanding which tool does what matters before assuming Sanity’s AI capabilities are either more comprehensive or more limited than they actually are.

Canvas: AI-Assisted Free-Form Writing

Canvas, introduced in the Spring 2025 release, is an AI-powered, context-aware authoring environment that lives inside the Sanity Dashboard. It is designed as a distraction-free writing space — closer in spirit to a focused editor like Notion or Google Docs than to a traditional CMS interface. Writers can draft freely without worrying about which content components to use or how to structure fields, because Canvas handles the translation automatically.

The AI component of Canvas functions as a ghostwriter: it provides autocomplete suggestions, generates first drafts from prompts, rewrites sections on request, and steps back the moment the writer takes control. Crucially, this AI is grounded in context. Teams can store notes — brand guidelines, tone of voice documents, product details, target audience profiles — and Canvas reads those notes when generating suggestions. The output is anchored in your actual brand context rather than generic AI output. Once a piece is drafted in Canvas, the platform automatically maps the free-form content into the structured fields of your Sanity Studio schema, eliminating the manual copy-paste step that currently sends most writers bouncing between tools.

Canvas also supports real-time, multiplayer collaboration and live data synchronisation, meaning writers, editors, and stakeholders can work in the same Canvas document together while a live preview of the structured content updates in parallel. For content marketing teams managing high-volume output, this closes the loop between free-form ideation and production-ready structured content in a single environment.

AI Assist: Inline Field-Level Automation

Where Canvas works at the document level, AI Assist operates at the field level inside Sanity Studio. Editors can trigger AI actions on individual fields — generating metadata, translating copy, rewriting a title to be more engaging, populating alt text for images, or summarising a long document into a short description. The actions are schema-aware, meaning the AI understands the purpose and constraints of each field based on the schema definition, and produces output that fits the content model rather than generic text that needs to be manually trimmed.

AI Assist also supports developer-defined custom instructions, so teams can encode brand voice rules, character limits, and style guidelines directly into the field configuration. For organisations running influencer marketing programmes or managing localised content across multiple markets, this means bulk metadata generation and translation at scale, without requiring editors to manually prompt the AI for each document. The free plan includes 100 AI credits per month; heavier usage requires the Growth plan or above.

Content Agent: Bulk Operations Through Conversation

Content Agent is the most operationally powerful of the three AI tools and serves a fundamentally different purpose. Rather than helping with the writing itself, Content Agent is designed to run content operations at scale through a conversational interface. Teams can ask it to audit hundreds of documents for missing metadata, identify content that contradicts recently updated legal dates, find gaps between published content and current industry trends, or apply bulk transformations across an entire content library.

The key distinction is that Content Agent understands your content structure. It can traverse document relationships, respect schema definitions, stage changes as drafts for human review, and integrate web research to fact-check claims while drafting — all without the content team needing to write a single line of code or export anything from the CMS. For teams managing large content libraries — franchise pages, product catalogues, multi-region marketing sites — this represents a meaningful shift in what a small editorial team can realistically maintain. All changes go through your existing approval workflow; nothing publishes without sign-off.

Collaboration vs. AI Authoring: Which Matters More for Your Team?

The honest answer is that these two capability sets serve different team archetypes, and the right emphasis depends on your current bottleneck. If your primary problem is that multiple editors are stepping on each other’s work, losing time to save conflicts, or managing review cycles through email threads and spreadsheets — real-time collaboration is the immediate win. Sanity’s collaboration layer is class-leading, and teams adopting it typically see workflow friction drop within the first few weeks of use.

If your primary problem is content volume — not enough hours in the week to produce the volume of SEO-optimised, structured, multi-channel content your strategy requires — then Canvas and AI Assist address that constraint more directly. The combination of AI-assisted drafting and automatic content mapping means teams can increase output without proportional headcount increases. One B2B analytics platform reported a 266% increase in CMS usage after implementing Sanity, alongside a 19% year-over-year increase in website traffic, driven largely by empowering non-developer content creators to publish independently.

For most growing teams, both capabilities matter and reinforce each other. A writer uses Canvas to draft with AI assistance, a second editor reviews and comments in real time, a manager approves via the built-in workflow, and Content Agent handles the metadata and audit layer that would otherwise consume hours of manual effort. This is the operational model Sanity is building toward with its Content Operating System positioning — and it is a more coherent vision than bolting AI onto an existing CMS infrastructure. If you are evaluating platforms to support an AI marketing workflow at scale, this structural integration matters.

Key Features Worth Knowing

• Content Lake — cloud-hosted structured content store with real-time sync and conflict-free merges
• Sanity Studio — open-source, fully customisable editorial interface built in React; deploy it anywhere or embed it in your application
• GROQ — Sanity’s purpose-built query language for JSON content; a single query can traverse references, filter conditions, and join related documents, reducing API overhead significantly
• Portable Text — rich text stored as portable JSON, renderable on any frontend without HTML lock-in
• Canvas — AI-powered, free-form writing environment with automatic schema mapping and real-time collaboration
• AI Assist — inline field-level AI for generation, translation, and enrichment, configurable per schema field
• Content Agent — conversational AI for bulk edits, content audits, gap analysis, and web-grounded research at scale
• Media Library — centralised asset management across teams and datasets, with role-based access, versioning, and AI-generated alt text
• Sanity Functions — serverless event handlers triggered by content changes; sync data, translate content, or send to third-party systems without additional infrastructure
• Visual Editing — click-to-edit on your live site with live-as-you-type previews, compatible with Next.js, Nuxt, Svelte, Remix, and Astro
• MCP Server — gives AI agents like Claude and Cursor governed, schema-aware access to your content for external AI workflows

Sanity Pricing: Free, Growth, and Enterprise Explained

Sanity’s pricing follows a three-tier model, and the free tier is one of the most genuinely capable in the headless CMS category. It is not a crippled trial — it runs in production for low-traffic sites and small projects indefinitely. The limits that push teams toward paid plans are specific and predictable.

Free ($0): 20 user seats, 10,000 documents, real-time collaboration, GROQ and GraphQL APIs, live previews, and a fully customisable Studio. The constraints are that scheduled publishing, comments and task assignment, custom user roles, and AI Assist are all excluded. The two user roles available are Administrator and Viewer, which limits editorial workflow control.

Growth ($15 per seat/month): Unlocks scheduled drafts, inline comments and task assignment, multiple user roles (Editor, Developer, Contributor), private datasets, AI Assist (100 credits per month included), and priority support. Viewer roles remain free and do not consume a paid seat, which makes the per-seat cost more manageable for teams with many stakeholders who need read access but not editing rights. For a five-person editorial team, this equates to $75 per month — considerably more affordable than Contentful’s entry-level paid plan, which starts at $300 per month flat.

Enterprise (custom pricing): Adds SAML SSO, advanced audit logs, custom data residency, SLA guarantees, dedicated onboarding support, and the ability to exceed Growth plan seat and dataset limits. Enterprise is a compliance and scale requirement, not a prestige tier. Teams under approximately 50 seats without strict security or regulatory requirements will find Growth sufficient.

One cost consideration worth flagging: Sanity’s pricing model is usage-based beyond the included quotas, and API request costs can escalate on high-traffic sites if caching is not configured carefully. Direct API requests run considerably more expensive than CDN-cached requests, making frontend architecture decisions a real cost driver at scale. For teams planning content-heavy builds, monitoring API usage from day one is practical advice.

Who Is Sanity Best For?

Sanity delivers the most value to teams where content flexibility, collaborative velocity, and long-term scalability are primary requirements. Specifically, it is the right choice when:

• The content model is complex — relationships, nested arrays, polymorphic content types, or content that needs to be reused across multiple channels and frontends
• Multiple editors need to collaborate simultaneously, particularly across distributed teams or different time zones
• The engineering team is comfortable with TypeScript, React, and either GROQ or GraphQL
• The organisation needs custom editorial workflows, approval chains, or role-based permissions that match how the team actually operates
• AI-assisted content generation and bulk content operations are part of the content strategy roadmap
• The project will evolve significantly over time and requires a content foundation that can change without painful migrations

This profile aligns closely with product teams at scale-ups, digital agencies managing content for multiple clients, e-commerce brands with complex product data, and media organisations with high editorial velocity. Notable companies running on Sanity include Nike, Figma, Spotify, Expedia, and Morning Brew — a range that reflects both the platform’s enterprise credibility and its performance at high content volume. For brands managing regional content across Asia with multi-language requirements, Sanity’s field-level localisation and AI translation via Agent Actions also make it a strong fit for multi-market operations — directly relevant for the kind of integrated digital marketing strategy that performance-driven agencies like those operating across Singapore, Malaysia, Indonesia, and China are increasingly asked to deliver.

What to Watch Out For

Sanity rewards investment. Teams that configure the Studio carefully, train their editors properly, and invest in clean schema design get outsized returns. Teams that treat it as a plug-and-play CMS will hit friction early and often.

The most common pain points to anticipate:

• Non-technical editor onboarding — a default Sanity Studio without customisation can feel overwhelming to editors accustomed to opinionated CMS platforms. Developer time spent configuring custom input components and focused document views is not optional; it is the price of the platform’s flexibility.
• GROQ learning curve — GROQ is expressive and efficient, but it is not as widely known as GraphQL or SQL. New developers need time to become productive with it.
• AI Assist limitations on the free tier — AI Assist, Canvas, and Content Agent features are either absent or metered on the free plan. Teams intending to leverage AI authoring at any meaningful scale should budget for the Growth tier from the outset.
• Dataset and staging workflow — Sanity uses multiple datasets for environment isolation, but promoting content between staging and production requires manual scripting unless you are on Enterprise, which adds dataset aliases for hot-swap workflows.
• Per-seat cost at scale — the $15 per seat model is competitive for small and mid-sized teams, but at 30 or more editors, the cost structure warrants a conversation with Sanity’s sales team about Enterprise economics.

Our Verdict

Sanity is the most developer-flexible headless CMS available, and with the Spring 2025 release it has meaningfully closed the gap on the editorial experience side that was previously its biggest weakness. The combination of real-time collaboration that is genuinely built in (not bolted on), and an AI authoring layer that respects content structure rather than ignoring it, makes it the strongest choice for teams that need both velocity and control.

The distinction between Canvas, AI Assist, and Content Agent matters in practice: Canvas accelerates drafting, AI Assist automates field-level enrichment, and Content Agent handles the operational overhead of maintaining large content libraries. Each solves a different problem. Teams that implement all three report significant productivity gains — not because the AI writes their content for them, but because it eliminates the mechanical work that was consuming editorial bandwidth.

Where Sanity still asks more of you than alternatives is in setup and editor configuration. The freedom the platform gives you comes with a corresponding responsibility to configure it well. For teams with developer resources and a genuine content strategy to execute, that trade-off is firmly worth making. For teams looking for a zero-configuration CMS that works out of the box for non-technical users, there are less demanding options. Understanding that distinction is the most important part of evaluating Sanity honestly.

If you are building or scaling a content-driven digital presence — whether that is an SEO strategy, a multi-channel brand experience, or an answer engine optimisation play for AI-driven search — Sanity gives you a structured content foundation that AI agents can actually work with. That is not a small thing in a content landscape increasingly shaped by how platforms like Google and AI assistants read and surface structured data.

Frequently Asked Questions

Is Sanity CMS free to use?

Yes. Sanity’s free plan is production-viable for small projects and personal sites. It includes 20 user seats, real-time collaboration, a fully customisable Studio, and GROQ and GraphQL API access. The main constraints are no scheduled publishing, no inline comments, no custom user roles beyond Administrator and Viewer, and no AI Assist. Paid plans start at $15 per seat per month on the Growth tier.

What is Sanity Canvas and how does it differ from AI Assist?

Canvas is a free-form, AI-powered writing environment where editors can draft content naturally before automatically mapping it into Sanity Studio schema fields. AI Assist operates at the field level inside Studio itself, allowing editors to generate, translate, or enrich individual content fields with AI on demand. Canvas is for drafting; AI Assist is for enrichment and automation.

How does Sanity’s real-time collaboration compare to Google Docs?

Sanity’s Content Lake uses operational transforms — the same underlying technology behind Google Docs collaboration — meaning changes sync in real time, cursor positions are visible, and there are no save conflicts. Unlike Google Docs, Sanity’s collaboration is tied to a structured content model, so every edit is schema-aware, version-tracked, and connected to a publishing workflow with role-based permissions.

Is Sanity good for non-technical editors?

With proper Studio configuration, yes. Without it, a default Sanity Studio can be confusing for editors accustomed to more opinionated platforms. A developer who understands the Sanity ecosystem can configure custom input components, conditional fields, and focused document views that make the Studio highly usable for non-technical teams. The Canvas AI writing environment is also significantly more accessible than the raw Studio for editors who primarily create content rather than manage schemas.

What is GROQ and do I need to learn it?

GROQ (Graph-Relational Object Queries) is Sanity’s purpose-built query language for fetching structured JSON content. A single GROQ query can traverse document references, filter by conditions, and project specific fields — reducing the number of API calls needed for complex content. Developers need to learn it; editors do not. It has a small learning curve but becomes a productivity multiplier for teams building content-heavy applications with complex data relationships.

How does Sanity handle multi-language content?

Sanity supports field-level localisation through its @sanity/language-filter plugin, which requires developer configuration to wire up. AI Assist and Agent Actions both support AI-powered translation as a native action, allowing teams to translate structured content field-by-field while preserving references, slugs, and schema integrity. Fallback logic for missing translations needs to be implemented in GROQ queries or the frontend rather than configured in the CMS itself.

The Bottom Line

Sanity is no longer just the developer’s CMS of choice. With Canvas, AI Assist, Content Agent, and the broader Content Operating System infrastructure launched in 2025, it is making a credible case as the structured content platform for AI-native content teams. The real-time collaboration engine remains class-leading. The AI authoring layer is now genuinely integrated rather than an afterthought. And the pricing model remains one of the most competitive in the headless CMS category for teams at the growth stage.

The platform rewards the teams that invest in it properly. Those who configure the Studio well, train editors thoughtfully, and use Canvas and AI Assist to eliminate mechanical workflow tasks will find themselves operating with a content infrastructure that scales as their ambitions do. The question is not whether Sanity is good — it clearly is. The question is whether your team has the developer resources to unlock its ceiling. If the answer is yes, it is one of the best-built content platforms available today.

For organisations evaluating how structured, AI-ready content infrastructure fits into a broader Generative Engine Optimisation strategy or a performance-led AI SEO programme, Sanity’s content-as-data model provides the kind of clean, queryable foundation that AI systems — from search engines to large language models — are increasingly built to consume.