From workbench
to business.

Define the first offer.

For people already running several agent jobs and losing track of what happened.

Start with independent developers, technical consultants and small technical teams. The promise is a commission, activity record, cost evidence, retrievable artifacts and acceptance history in one workspace.

First demonstration

One real commission → execution → delivery → review → artifact download.

Evidence needed

A potential customer can explain why they would use it, connects a provider and comes back for another job.

Decision owner

Jay chooses the initial buyer and offer; implementation must support the demonstrated workflow.

Choose who sells Capsule.

The legal seller should match the publishing identity, payment account, invoices, support details and payout bank account. An existing FMVision entity may be suitable; a new product does not automatically require a new company.

1. Confirm the selling entity or individual, country and relevant state/province.
2. Check the proposed product name and applicable business-name/trademark requirements.
3. Complete the registrations, tax identification and banking appropriate to that seller.
4. Prepare terms, privacy, cancellation/refund policy and a working support contact.

Completion evidence: confirmed seller identity, appropriate registration/tax records, payout account and published legal/support pages.

Jurisdiction is unconfirmed. U.S. SBA launch guidance ↗ is one conditional reference, not a universal registration checklist.

Make the customer path complete.

Acceptance test: a new customer signs in, connects a provider, submits a job, leaves, returns, understands its outcome and downloads the correct artifacts. Failure recovery must be tested too.

The hosted operator workspace currently has no imported Mac study history or connected customer provider. A running site is not proof that customer onboarding is complete.

Connect payment to access.

Use hosted checkout and a customer billing portal. Start with Stripe Billing if its seller eligibility and commercial terms fit. Compare a merchant-of-record option when outsourcing supported transaction taxes and payment operations is valuable.

1. Verify the seller and payout account with the selected payment provider.
2. Create the subscription product, price, cancellation and refund settings in test mode.
3. Connect signed subscription/payment webhooks to account entitlements; handle duplicate events safely.
4. Test payment success, failure, cancellation, grace periods and customer exports.
5. Enable live checkout only after a complete test purchase and access-recovery test.

Failed payment should not silently erase a customer's work. Define a clear access and retention policy.

Stripe Billing · Subscription webhooks · Stripe Managed Payments · Paddle. Fees, seller eligibility and coverage remain to be confirmed.

Test what customers will pay.

Charge for Capsule's software, hosting, records, reports and support. Under the initial proposal, customer AI compute is billed separately by their connected provider.

These are unvalidated price experiments, not live offers. No customer count, revenue or profitability is claimed.

Evidence needed: paid pilot conversion, repeat use, support time, operating cost per workspace and reasons customers cancel.

Register the integration separately.

Business registration, payment onboarding and OpenAI plugin approval are three separate processes.

1. Verify the publishing individual or business in OpenAI Platform.
2. Provide website, support, privacy and terms URLs plus accurate listing text.
3. Choose a local or remote MCP distribution path. A website alone does not convert a local plugin to remote MCP.
4. For remote MCP, implement a stable public HTTPS endpoint and supported authentication. The current operator Basic login is not public MCP OAuth.
5. Prepare the ZIP, a demo recording, positive/negative test cases and a dedicated reviewer account with sample data.
6. Complete domain verification, tool scans, review and explicit publication after approval.

Subscription boundary. Existing paid customers may sign in through the plugin. Do not put plan sales, upgrade promotions or digital checkout in it under the current rules.

Submission steps · Local-MCP eligibility · Plugin guidelines

Earn the first repeat customers.

Proposed first milestone: five to ten paying pilot users, with assisted onboarding and a narrow, demonstrable workflow.

1. Publish a landing page with buyer, outcome, supported integrations, setup and pricing.
2. Record a 60–90 second real workflow. Show the result and artifact download, not just the interface.
3. Offer a sample report and a clear installation walkthrough.
4. Demonstrate to relevant developers, consultants and small teams; document objections and onboarding failures.
5. Share practical demonstrations in relevant communities and developer channels, observing their rules.
6. Test paid acquisition after activation and repeat use are observable; define an experiment and success criterion before buying traffic.

Evidence needed: activation and retention, paid conversion, support burden and the cost of acquiring a retained customer. A directory listing is an additional channel, not a customer-acquisition guarantee.

The inputs we still need.

Seller

Existing entity or individual; country and state/province; public publishing name.

Customer

Which initial buyer and one job worth paying to manage.

Offer

Monthly subscription experiment, pilot terms and what support is included.

Provider connection

Customer-owned provider account initially, or a different explicitly designed billing model.

Payment route

Standard payment processing versus a supported merchant-of-record arrangement.

Distribution

Direct website first; local plugin eligibility or remote MCP implementation as a separate workstream.

The commissioned team is tasked with challenging these assumptions, resolving policy ambiguities, and returning a stronger proposal, source register, economics model and exact operator walkthrough. Delivery will remain distinct from local review and acceptance.