Contract
Participant wallets (test USDC)
Why Solana
Actions
Recorded on-chain
| Nothing yet |
Solana · stablecoin rails for construction
A client keeps 5–10% of every payment as a warranty for one or two years. Holdback locks that money in a Solana vault with no key, freezes only what a defect costs, and lets anyone release the rest the day the warranty ends.
On a $200,000 contract the client holds back $10,000 in its own account, mixed with everything else it owes.
The warranty ends and the answer is "next month". The subcontractor chases the money for months.
For the client it is free financing for the next project, paid for by the smallest firms on the site.
If the client fails, retention joins the pot for creditors. This happened to subcontractors of Carillion in 2018.
New Zealand forces trust accounts (2023). The UK plans to ban cash retention. Holdback keeps the guarantee and removes the abuse.
Sources: Gowling WLG (2025), Fenwick Elliott, Siteline State of Subcontractor Billing 2026, NZ Construction Contracts (Retention Money) Amendment Act 2023, Kennedys (2025).
Each invoice splits 95% to the subcontractor and 5% to the vault. Both transfers succeed or neither does.
The vault is owned by the contract's program address. Only the program can move the money, and only by the rules.
The client freezes the cost of one defect with an evidence hash. An arbiter pays the repair or unfreezes it.
After the warranty date, with no open defect, any wallet or a Blink link sends the balance to its owner.
The subcontractor can sell the locked claim. The funder pays and becomes the beneficiary in the same transaction.
Two calculators with editable assumptions. The first is for one subcontractor, the second sizes the market. Defaults are our hypotheses; interviews will replace them.
Cost of waiting = stuck amount × cost of cash × months / 12. Selling at 90% costs a flat 10%. With a 14-month wait and 18% cost of cash, waiting costs about 21%, so selling is cheaper.
Output of 10.705 trillion ₸ is the 2025 figure from the Bureau of National Statistics. Every other input is an assumption. Kazakhstan is a pilot market; England alone holds £3.2–5.9 bn a year, so the pitch stays global. Details in docs/MARKET-MODEL.md.
| Client's own account | Bank trust account | Ban on retention | Holdback | |
|---|---|---|---|---|
| Protected from the client's bankruptcy | No | Yes | n/a | Yes* |
| Paid out without asking | No | No | n/a | Yes |
| Subcontractor sees the balance | No | On request | n/a | In real time |
| Client keeps the quality guarantee | Yes | Yes | No | Yes |
| Cash before the warranty ends | No | No | n/a | Sell the claim |
* The money sits outside the client's account. How a court treats the vault in an insolvency has not been tested. No analog among the 5,428 projects in the Colosseum database.
Anchor program with 8 instructions, 22 passing tests on a local validator, demo videos and this simulation.
7 subcontractors and 3 developers in Shymkent. Goal: confirmed retention clauses and 2–3 letters of intent. Kit: docs/INTERVIEW-KIT.md.
Deployed to Solana devnet. Press "Use Phantom on devnet" in the demo above to sign every step with your own wallet.
One real contract with test money, then New Zealand and Australia where trust accounts are required.
Not yet verified: how common and how late retention is in Kazakhstan, what a developer would pay, and how a court treats the vault.