Here’s a crisp, end-to-end operating playbook for the aiConnected Marketplace—from a developer submitting an Engine, to a customer turning it on, to monthly developer payouts. I’ve split it into phases so you can hand it straight to product, ops, and engineering.

​

0) Key definitions (shared language)

• Engine: A packaged automation (primarily n8n template–based) that runs on aiConnected infrastructure. Billed per run at $0.05.
• Run: One complete execution of an Engine (success or fail after retries).
• Revenue share: 80% to developer ($0.04/run), 20$0.01/run) on paid runs.
• Free tier: Every customer gets 250 free runs/month platform-wide (non-revenue to developers unless a promo subsidy is explicitly offered).
• Credentials model: Customers bring their own API keys/OAuth; secrets stored in aiConnected’s vault; developers never see user data.

---

​

A) Developer onboarding (before any submission)

1. Create developer account & profile
   • Identity/KYC, tax forms (W-9/W-8), payout method (ACH/wire).
   • Public profile page (bio, portfolio, industries served).
2. Accept terms
   • IP ownership, security, data-handling, support expectations, refund & chargeback policy, service deprecation rules.

---

​

B) Engine submission (what a developer uploads)

1. Package
   • n8n template (or supported blueprint) + Engine Manifest (name, category, version, inputs/outputs, required connections & scopes, expected runtime, webhooks).
   • Setup assets: logo, 3–5 screenshots, 90-second demo video (optional), quickstart, changelog, support contact.
   • Test kit: sample data, unit test flows (where relevant), success criteria.
2. Declare requirements
   • Required integrations (e.g., Gmail, Shopify, Stripe), OAuth scopes, environment variables.
   • Data classification (PII present? retention needs?) and compliance attestations.
3. Submit via Developer Console and select Release channel (Private, Limited/Pilot, Public).

---

​

C) Automated checks (gate 1)

• Schema & linting: template validity, manifest completeness.
• Security scan: disallowed endpoints, key exfiltration patterns, unsafe code nodes.
• Dependency allowlist: only approved connectors.
• Resource profiling: estimated runtime, memory, concurrency footprint.
• Synthetic tests: run with sandbox creds & sample data.
  Outcome: Pass → Human Review; Conditional Pass → Fixes requested; Fail → Rejection with reasons.

---

​

D) Human review (gate 2)

• Functionality check: does it do what it claims?
• UX & docs: clear activation steps, sane defaults, error messaging.
• Policy/IP: no scraped/illegal content, license checks.
• Market fit: avoids duplicative listings without differentiation.
  Outcome: Approve / Approve-with-changes / Reject (all reasons logged).

---

​

E) Staging & listing

• Deployed to staging cluster with isolated secrets.
• Marketplace listing assembled (SEO summary, use cases, industries, required integrations, $0.05/run price badge, expected monthly run estimate).
• Quality badges (Security Checked, Performance Tier, Popular in X Industry) applied when earned.
• Optional Pilot: limited customer cohort for 1–2 weeks with feedback loop.

---

​

F) Public launch

• Promotion slots (New & Noteworthy / Trending).
• Developer notified; followers of the developer get updates.
• Version pinned (e.g., v1.0.0); support contact active.

---

​

G) Customer discovery → activation (happy path)

1. Find & evaluate
   • Search, filters, “recommended for you” (usage-based & problem-based suggestions).
   • Read listing; see run count, ratings, industries usage chart, recent updates.
2. Click “Activate”
   • Pre-flight wizard: shows required connections & scopes, estimated monthly runs, cost controls.
3. Connect services & secrets
   • OAuth/API key flow; tokens stored in vault; scopes shown plainly; test connectivity.
4. Configure & test
   • Minimal inputs (mapped fields with sensible defaults).
   • Test Run on sample or real data; view logs, outputs, and data mappings.
5. Set guardrails
   • Choose Free plan (uses remaining 250 runs) or Pay-as-you-go.
   • Budget cap, per-day run limits, alert thresholds, pause-on-error toggle.
6. Go live
   • Enable. A success card confirms: last run time, last status, runs remaining, spend-to-date.

---

​

H) Runtime & reliability (what happens on each run)

• Orchestration: job queued with idempotency key; fetched by a worker in a secure sandbox.
• Secrets: pulled just-in-time from vault; never exposed to developer.
• Metering: run recorded at start; billable only on completion (success or terminal fail after retries).
• Retries & DLQ: exponential backoff; push to dead-letter with clear error codes.
• Observability: structured logs, step timings, external API latencies.
• Post-run UX: quick 1–click rating; surface common fixes if errors recur.

---

​

I) Billing & quotas (customer view)

• Monthly 250 free runs draw down first (platform-wide).
• After free tier, $0.05/run pay-as-you-go; receipts & invoices issued.
• Real-time usage dashboard; alerts at 50/75/90% of budget; auto-pause when cap hit.

---

​

J) Support & incident handling

• Tiered support (community → standard → premium).
• In-app ticketing pinned to Engine & specific run IDs.
• Status page & incident comms; automatic crediting rules for platform-wide incidents (where applicable).

---

​

K) Versioning & change management

• SemVer: Patch = safe bugfix; Minor = backward-compatible features; Major = breaking change.
• Rollouts: canary (1–5%), staged (25/50/100).
• Migrations: config diffs shown; one-click revert to previous working version.
• Deprecations: ≥60-day notice, auto-migration where feasible.

---

​

L) Compliance & privacy (platform guarantees)

• Data minimization; PII tagging in flows; field-level redaction in logs.
• Encryption at rest/in transit; secrets never logged.
• OAuth token lifecycle with refresh & automatic rotation.
• Customer export & deletion (per Engine and global).
• Developer never accesses customer data; n8n templates run inside aiConnected tenancy only.

---

​

M) Developer analytics & lifecycle

• Dashboard: daily runs, active installs, conversion funnel (views→activations→paying), error rates, median runtime, average cost per customer.
• Customer feedback: ratings & reviews; Q&A tab.
• Comms: release notes, pinned notices, scheduled maintenance.
• Quality triggers: automated re-review if crash-looping or spike in error rate; possible delisting after warnings.

---

​

N) Revenue calculation & payouts

1. Gross revenue per Engine
   • Count paid runs only × $0.05.
2. Share
   • Developer: **80% ($0.04/run)**; aiConnected: 20$0.01/run).
   • Free runs: no revenue share (unless aiConnected runs a clearly announced subsidy promo).
3. Adjustments
   • Refunds/chargebacks/credits netted out in the same period.
4. Schedule
   • Monthly, net-30 from month-end; minimum payout threshold (e.g., $50).
5. Delivery & docs
   • Payout via ACH/wire; downloadable statements (runs, gross, adjustments, net).
6. Disputes
   • Developer can dispute metering within 60 days with run IDs; audited by ops.

---

​

O) Takedown, sunset, and handover

• Voluntary sunset: developer requests delist → 90-day support window; customers notified with alternatives or migration guides.
• Policy takedown: immediate unlisting for egregious violations; remediation path defined; payouts held pending resolution.
• Data handling: all Engine-scoped data can be exported/deleted by customers on sunset.

---

​

Quick checklists

​

Developer (per Engine)

1. Package: template + manifest + assets + test kit.
2. Declare integrations, scopes, data classes.
3. Submit → resolve automated/human review notes.
4. Pilot (optional) → Launch.
5. Maintain: updates, support, changelog.
6. Monitor analytics; iterate on conversion & reliability.
7. Receive monthly net-30 payouts.

​

aiConnected (internal)

1. Run automated checks; human review; sign-off.
2. Stage, list, badge, and (optionally) pilot.
3. Operate runtime: queue, workers, vault, metering, billing, observability.
4. Enforce security/compliance; handle incidents.
5. Calculate revenue; issue statements & payouts.
6. Govern quality; manage deprecations & disputes.

​

Customer

1. Discover Engine → Activate.
2. Connect OAuth/API keys; configure inputs.
3. Test run → set budgets/limits → Go live.
4. Monitor runs, costs; get alerts; pause/resume as needed.
5. Export/delete data anytime.

---

​

“Happy-path” timeline (first public release)

• Day 0–2: Developer packaging & submission.
• Day 2–4: Automated checks & fixes.
• Day 4–7: Human review & staging.
• Day 7–14: Optional pilot → Public launch.

---

If you want, I can turn this into a one-page SOP for each team (DevRel, Review, Runtime Ops, Billing) and a minimal Engine Manifest template you can drop into the Developer Console.