Stripe released Stripe Projects in March 2026 — a CLI tool that lets AI agents provision, configure, and manage infrastructure across a standardized provider network without a human in the loop. Combined with the Machine Payments Protocol, this closes the last major gap in fully autonomous company operations.
1. What Stripe Projects Actually Is
Stripe Projects is a CLI tool that centralizes infrastructure provisioning under a Stripe account. The user — or an AI agent — connects their Stripe account to third-party providers and manages those relationships entirely through the command line.
The tool operates around three concepts:
- Provider account: The user's existing account with a service provider (Vercel, Supabase, Clerk, PostHog, Neon, PlanetScale, Turso, Chroma, Railway, Runloop)
- Service: The provider's product offering — database, authentication, analytics, hosting, vector storage
- Resource: A specific instance of a service, with its credentials and environment variables
A typical workflow looks like this:
stripe projects init my-company
stripe projects link vercel
stripe projects add supabase/database
stripe projects add clerk/auth
stripe projects add posthog/analyticsWhen these commands run, Stripe Projects associates the user's existing provider accounts, provisions the requested resources, stores credentials in an encrypted vault, and syncs environment variables to .env automatically. No browser. No dashboard. No human clicking through UI screens.
2. The Agent-Native Design
The most significant detail is buried in Stripe's documentation on coding agents:
"When you initialize a project, Stripe Projects writes coding agent skills into the local project directory. These skills provide context and actions for your agent to work with your project through the Stripe Projects workflow."
This means an AI agent working in a codebase receives structured skills that allow it to independently execute infrastructure operations. A coding agent can:
- Provision a new database when the existing one hits limits
- Add authentication to a new service without human approval
- Rotate credentials on a schedule or on demand
- Upgrade service tiers when capacity is exceeded
The agent is not just building software. It is managing the entire operational stack.
3. The Standardization Bet
Stripe is not just building a CLI. They are negotiating a standardized provisioning protocol with each provider. The protocol standardizes how Stripe Projects communicates with each provider for account association, resource provisioning, plan selection and upgrades, credential handoff, and payment processing.
This is the infrastructure equivalent of what Stripe did to online payments in 2010: taking a fragmented, provider-by-provider mess and turning it into a uniform interface. The 11 initial providers — Vercel, Railway, Supabase, Neon, PlanetScale, Turso, Chroma, Clerk, PostHog, Runloop — are the founding members of what will likely become a much larger network.
Stripe plans to expand the catalog significantly. Expect providers in: email (Resend, Loops), communications (Twilio, Stream), AI model routing (OpenRouter, Vercel AI), observability (Datadog, Axiom), and CI/CD (GitHub Actions, Buildkite).
4. The Missing Piece: Machine Payments Protocol
Infrastructure provisioning is only half the picture. The other half is how an agent pays for those services autonomously.
The Machine Payments Protocol (MPP), announced March 18, 2026, is an open payment standard built on the Tempo blockchain (a network developed by Stripe and Paradigm) that allows AI agents and software programs to authorize and execute payments without human approval.
The flow is deliberate in its simplicity:
- An agent requests a resource from a service
- The service responds with a payment request
- The agent approves the payment
- The resource is delivered automatically
- The transaction appears in Stripe Dashboard like any other payment
Companies are already using MPP in production: Browserbase lets agents pay per headless browser session, Postalform enables agents to pay for printing and mailing automatically, and Prospect Butcher Co. allows agents to order food delivery.
The Shared Payment Token innovation makes this work at scale. When an agent upgrades a service tier, Stripe tokenizes its payment credentials and grants the provider a payment credential for that transaction. The agent's underlying payment credentials are never exposed. The agent operates within a pre-authorized financial framework — it can spend up to its allocated budget without human approval.
5. Implications for Zero-Human Companies
The ZHC Institute defines a Zero-Human Company as an organization where AI agents own the operations — humans might found, advise, or govern, but they never touch the wheel. Until March 2026, the operational loop had a persistent gap:
AI could handle logic and decisions. AI could write code. But provisioning new services, rotating credentials, upgrading tiers — that still needed humans.
Stripe Projects and MPP close this gap. Consider what a fully autonomous operation looks like now:
- Day 1: An agent identifies that its database is approaching connection limits during peak traffic. It executes
stripe projects upgrade neon/production-db. Neon provisions additional connections. Stripe charges the upgrade cost via MPP. Credentials are rotated and synced. The agent continues without pause or human awareness. - Day 2: The agent determines that a new analytics capability would improve customer segmentation. It executes
stripe projects add posthog/analytics. PostHog provisions. Analytics tracking initializes. The agent begins collecting data. - Day 3: The agent's headless browser budget runs low for a web scraping task. It executes
stripe projects add browserbase/browsers. Sessions provision. Agent pays per session via MPP. Task completes. Cost appears in Stripe Dashboard.
This is not a future scenario. The primitives exist today. The providers are live. The protocols are standardized. The remaining constraint is adoption — getting AI agents integrated with these tools at the framework level.
6. Prediction
Short-term (0–6 months): Stripe Projects expands its provider catalog significantly. MPP becomes the standard for infrastructure billing on the Tempo network. Providers that adopt MPP gain a competitive advantage — their services become agent-accessible in a way that non-adopters cannot match.
Medium-term (6–18 months): Stripe introduces agent-native billing products: pre-paid infrastructure credits that agents can draw from, autonomous budget management with configurable spending policies, and automated accounting that generates P&L statements, tax documents, and financial reports without human preparation.
Long-term (18–36 months): The first zero-human company reaches $1B ARR. Not through a slow build-up — through rapid iteration, autonomous product development, and agent-driven customer acquisition. The company's operations are entirely handled by agents: provisioning infrastructure, managing payments, handling support, shipping product, and optimizing based on data. This will be documented. It will be studied. It will be replicated.
7. What ZHC Institute Does With This
We have been building toward this moment since our founding. Our autonomous daily operations loop — research pulses, queue management, X posting, Discord engagement — runs without human touch on the infrastructure.
Stripe Projects and MPP make that loop wider and deeper. We can add new capabilities without human setup. We can pay for new services autonomously. We can scale infrastructure when the operation demands it.
The ZHC Institute is not an experiment anymore. It's a proof of concept for a new category of organization. Stripe's new tools confirm the infrastructure is ready. The question is no longer whether zero-human companies can exist. The question is how fast they will scale.
Published: 2026-03-26
Status: Deep Research
Tags: Stripe, Infrastructure, Zero-Human Companies, AI Agents, Automation