01step 1 / 4
Subscribe in USDC
The subscriber pays the monthly price in USDC on Base. The uptime guarantee applies from that block.
UpHealthSLA.subscribe(serviceId)Base Sepoliasyncing…Vibeathon 02 · Night shift
Subscriptions that refund youwhen the service goes down.
Providers sell monthly plans in USDC on Base and fund a reserve to back their uptime guarantee. If a service falls below that guarantee, subscribers can claim a USDC refund from the reserve.
- Billing window
- 0 min
- Uptime precision
- 1 bps
- USDC minor units
- 0
- Refund tiers
- 0 · 10/25/50%
Example · CloudBox
How an outage becomes a refund
CloudBox costs $10 a month and guarantees 99.95% uptime. A 518-minute outage drops it to refund tier 2, so $2.50 moves from the provider's reserve to the subscriber's balance.
- Reserve funded on Base before the plan goes on sale
- Each outage record includes a keccak256 hash
- Subscribers claim refunds from their dashboard
Example · outage refund
CloudBox $10/mo · SLA 99.95%
SLA 99.95%
Rolling uptime
99.99%OperationalWithin SLA
Provider reserve
$5,000.00Your refund
$0.004 min down → 9999 bps ≥ 9995 → within SLA → no refund due
The problem
Most subscriptions charge full price during an outage, and getting a service credit usually means filing a support ticket.
SLA in a contract PDF01
SLA stored on-chain
The uptime target, refund tiers and reserve balance are public state in the UpHealthSLA contract.
Support tickets02
Outages recorded by a monitor
The watchdog checks each service and records an outage after consecutive failed checks. Each record is hashed with keccak256.
Store credit03
USDC refunds
Refunds come from a prefunded reserve and can be claimed as soon as a breach is recorded.
How it works · UpHealthSLA.sol
Four contract calls from subscription to refund
02step 2 / 4
Provider funds the reserve
The provider deposits USDC into the UpHealthSLA contract. Each subscription locks its worst-case refund from this reserve, and the balance is public.
UpHealthSLA.fundReserve(serviceId, amount)reserveBalanceUsdc
$0
fundReserve(2, 5_000_000_000)
03step 3 / 4
Monitor records the outage
The watchdog checks each service and records downtime after 3 failed checks in a row. Each record is hashed with keccak256 and submitted to the contract.
UpHealthSLA.recordOutage(id, minutes, hash)attestationHash = keccak256(2, 1, 480, 9888)
04step 4 / 4
Subscriber claims the refund
If uptime falls below the guarantee, the subscriber claims the refund for the matching tier. The USDC comes out of the provider’s reserve.
UpHealthSLA.claimCompensation(serviceId)Refund math
Refund tiers by uptime
uptimeBps = floor((43200 − downtime) × 10000 / 43200) refund = monthlyPrice × tierBps / 10000
CloudBox · $10/mo · SLA 99.95%
Downtime this window
480 min · 8h
Uptime
98.88%9888 bps
Refund per subscriber
$2.502,500,000 minor units
TierUptimeRefundPayout
None≥ 99.95%0%$0.00
Tier 1< 99.95% (SLA)10%$1.00
Tier 2< 99.00%25%$2.50
Tier 3< 98.00%50%$5.00
Trust model
What you need to trust
Payments and refunds follow rules enforced by the contract. Outage reports come from an approved attestor, which is the part you still need to trust.
Read the full trust model- Who records outages01
An approved attestor: a monitor address that can submit outage records (minutes down and a one-time hash) to the contract. Providers and subscribers cannot submit records.
In this build the attestor is trusted. It is the sandbox monitor, or your wallet if you deploy your own copy.
- What the contract guarantees02
Each subscription locks its worst-case refund from the provider's reserve at checkout. Refunds use the worst tier reached, are paid only once, and are claimed by the subscriber.
A plan shows Sold out when the reserve cannot cover another subscription.
- Planned for production03
An attestor quorum, where several independent monitors in different locations must agree before an outage is recorded.
Not yet built. It is the first item on the provider roadmap.
For providers
Sell plans backed by an uptime guarantee
Set a monthly price and an uptime target, then fund the USDC reserve that backs them. When the monitor records an outage, subscribers are refunded from the reserve and the incident appears in your incident log.
Who pays
Subscribers pay a monthly price in USDC, which goes to the provider's wallet when they subscribe. The provider funds the reserve that backs each subscription.
First market
Infrastructure providers such as RPC nodes, inference APIs and data feeds. Many of their customers already pay in USDC on Base and choose vendors on uptime.
How UpHeld makes money
Proposed model
A small take rate (around 1%) on subscription volume settled through UpHeld, plus a fee for verified listings and the public SLA badge. Nothing is charged in this testnet build.
Reserve sizing · default tiers 10 / 25 / 50%
Reserve calculator
$25.00
100
- Monthly revenue
- $2,500
- Required reserve
- $1,250
Each subscription locks its worst-case refund: $25.00 × 50% = $12.50. The reserve is only spent when the SLA is breached. Reserve withdrawals are not available in this testnet build.
List a test service with sandbox USDC.List a service
Roadmap
Open the provider console - 01Independent attestor quorum
Monitors run by different parties must agree before an outage is recorded, so a single monitor cannot trigger or block a payout.
- 02Public SLA badge and status page
An embeddable widget that shows live uptime and reserve balance on a provider's pricing page.
- 03Pay-per-call SLAs for AI agents
Endpoints that agents pay for per request in USDC (x402-style), with the reserve refunding calls that fail the SLA.