Why Claude-native
Built for Claude, by commitment.
Planar is built for Claude — not for "any AI assistant that happens to support MCP." That's a deliberate product commitment, not a hedge. The trade-off accepted: smaller short-term TAM in exchange for a deeper, more opinionated product surface that converts the addressable audience at a higher rate.
The framing
Software written by talking to a model.
A new wave of software is emerging where the primary authoring surface is a conversation with a language model. Not a chat bolted onto an app — the conversation is how the work gets done. The model holds context, runs tools, drafts artifacts, and the user reviews and steers.
For deal documents — proposals, SOWs, engagement letters, estimates — this conversational surface is a much better fit than a WYSIWYG editor. A proposal is a structured negotiation, not a layout problem. The structure (scope, pricing, payment schedule, terms) is small and repetitive; the content (this specific client, this specific deliverable, this specific number) is what changes.
A model is good at filling in the variable part while preserving the structural part. A WYSIWYG editor forces the user to do both at once. Planar is built on the first observation, not the second.
One platform, deeply
Why we picked Claude.
Three reasons:
- The wedge audience lives in Claude. AI implementation firms, AI-forward agencies, modern service businesses with a reason to ship faster — they're already in Claude every day. The deeper Planar integrates into how they already work, the lower the activation friction.
- Claude is materially better at the work. Drafting and managing proposals is structured, sustained, tool-calling-heavy work. Claude handles that profile better than the alternatives, and the gap isn't shrinking on the dimensions that matter for this product.
- Building on one platform deeply produces a better product than building for every platform shallowly. Every "AI-agnostic" tool that ships ends up at the lowest common denominator on every integration surface. Going Claude-native means we can use Custom Connectors, Skills, Artifacts, Projects, and Claude Code as first-class primitives — not as adapters bolted onto a generic abstraction.
Architecture vs. experience
Claude-native experience, MCP-standard architecture.
Architecturally, the Planar MCP server is built to the standard MCP spec. Protocol portability is hygiene — the integration layer shouldn't be tied to one client's quirks. Non-Claude MCP clients may technically connect.
Experientially, the product is optimized exclusively for Claude. Demos, docs, installation flows, integration surfaces, and tested clients all assume Claude. Non-Claude clients are undefined and unsupported.
The frame: Claude-native experience, MCP-standard architecture.
What we use that nobody else can
Claude-specific integration surfaces.
These are the integrations that exist only because the product is Claude-native. Each makes Planar feel native to how the wedge audience already works, and each is hard to replicate without committing to the platform.
Custom Connector install
v1Planar installs as a Custom Connector in Claude's native Connectors UI (Settings → Customize → Connectors). User pastes the MCP server URL and their workspace API key; Claude handles the rest. This is the primary onboarding path and what every install screenshot, demo, and doc reflects.
Planar Skill (planar-drafter)
v1A companion Anthropic Skill that ships in the Skill directory. Knows proposal structure, terminology per vertical, pricing conventions, and how to invoke the Planar connector. Senders activate the Skill in their Claude project; drafting becomes guided.
Artifacts preview
v1When drafting in Claude, the staged content renders as a live Artifact preview before opening. The Artifact is the exact rendered Planar page, minus the recipient-specific actions. Sender iterates in the Artifact before triggering `open_planar`.
Claude Projects integration
v1.5Each client or deal lives as a Claude Project with the Planar connector enabled. The Project remembers every prior interaction, draft, version, and Planar associated with that client. Drafting follow-on Planars becomes context-aware.
Claude Code workflow
v1.5Power users open Planars from the command line via Claude Code, piping proposal drafts from their terminal-based workflow directly to Planar.
What recipients need
Nothing.
Recipients of Planars are universal. They open the URL in any browser. They sign, pay, and schedule on the page. No account creation, no app install, no Claude login, no API key — nothing.
The "Claude-native" choice constrains the sender: the deeper product surface lives inside Claude. It doesn't constrain the recipient. A sender's clients may use ChatGPT, Gemini, no AI at all — it doesn't matter.
For senders not yet on Claude, the dashboard at app.openplanar.com offers a web-based markdown draft form that produces the same Planar. Same deal surface, same integrations, same engagement tracking. The Claude-native experience is the deeper product; the web form is the floor.
Stack-agnostic, by design
Bring your existing tools. Planar isn't trying to be the whole stack.
A core architectural commitment of Planar: customers bring their existing stack. The same logic that makes MCP the right protocol bet — don't lock users into one client — applies to integration providers. Every integration category exposes multiple v1 providers plus an adapter interface for additional providers later.
Signature: DocuSign, Dropbox Sign, native click-to-sign. Payments: Stripe Connect in v1; Plaid ACH, Square, PayPal in v1.5. Scheduling: Cal.com, Google Calendar direct; Calendly in v1.5. Email: Resend with verified sender domains.
Planar is the workflow layer that connects your stack — not the workflow layer that forces you onto our stack.
Built for Claude. Open to everyone.
Add Planar as a custom connector in Claude. Your recipients don't need a thing.