The autonomous firm

The landing page calls every firm on this protocol fully autonomous. This page cashes that claim out. A traditional prop firm is a staffed business: someone sets risk parameters, someone reviews winning accounts, someone approves payouts, someone runs the fraud queue, someone posts the marketing. Each of those someones is a cost, a bottleneck, and a place where judgment can bend. A firm here has none of them. The operator owns the brand, the pricing, and the templates; the protocol runs the firm.

The whole machine, in one pass

The table above lists the parts. This section is why they were built as one system rather than five features, because the design only makes sense once you see what each loop hands to the next.

It starts with a token, not a marketing budget. Every firm deploys with its own token, $FIRMA, on a constant-product bonding curveBonding curveThe on-chain market for a firm's $FIRMA token, and the mechanism its treasury buys from to fund payouts.. That is the same shape as a pump.fun launch and it is doing the same job: a brand-new firm has no liquidity and no community, and the curve manufactures both from the first buy. There is no seed round, no market-making deal and no listing to negotiate. The firm mints its supply, ~20% goes straight into the treasury's own token reserve as payout inventory that cost the deployment fee nothing, and the curve is open for business.

GraduationGraduationThe point at which a firm's token has enough market depth to migrate its liquidity to a public Solana DEX. is when the firm stops being a private economy. Once the curve fills to its graduation threshold, the token migrates to Raydium with its liquidity locked. That threshold is set when your firm launches and is sized in SOL to land near the same reference market cap for every firm, so the exact figure depends on the SOL price the day you deploy rather than being one number quoted platform-wide. Your token page shows your firm's own threshold and how close the curve is to it. At that moment the firm stops being visible only to people who found its storefront, and becomes visible to everyone on Solana: every DEX aggregator, every screener, every wallet. The people who arrive are not prop-firm customers, they are token buyers, and some fraction of them become traders because they now hold a stake in a business whose product they can use.

The treasury and the token pull on each other. A firm's payout capacity is a waterfall of four layers, and two of them are denominated in $FIRMA$FIRMAA firm's own fixed-supply token, minted at launch. Used to fund and deliver payouts via the firm's bonding curve.: the treasury's own token reserve and the backstop stakers' pool. When the token appreciates, the payout capacity behind every funded trader appreciates with it, without a single extra dollar of evaluation revenue. That is the magnification: the same SOL backs a larger promise because part of the backing is an asset the market is repricing upward.

Payouts are buys. This is the part that surprises people. A funded trader's payout is not a wire from a bank account. The firm's treasury SOL buys $FIRMA on the firm's own curve and delivers the tokens to the trader's wallet. Every payout is therefore buy pressure on the token, and a slice of each one is burned outright, so the supply falls as the firm pays. A firm that pays its traders well is, mechanically, a firm whose token is being bought and burned.

Evaluation fees feed the same loop. Of every evaluation fee, 1% buys $FIRMA back into the treasury's payout reserve and 1% routes to the protocol token's buy-and-burn sink. Trading on the curve pays a 1% fee that splits half to the firm's treasury. Volume in the token deepens the treasury; depth in the treasury unlocks larger account sizes to sell; larger accounts attract better traders; better traders generate the fees that start the loop again.

And every payout is a receipt. This is where the economics turn into marketing. A payout on this protocol is a transaction on Solana, permanently public, that anyone can open and check. A traditional firm's "we pay our traders" is a claim on a landing page. Here it is an address. The single most persuasive marketing asset in this industry is produced automatically, by traders doing the thing they came to do, and it cannot be faked because it is a settlement.

So the marketing is the exhaust of the business, not a department. The protocol turns each of those receipts into a post from the firm's own account, and the firm's traders amplify it for airdrop points onto their own timelines. Nobody writes a campaign. The thing being advertised is the thing that just happened, and the people advertising it are the ones it happened to.

The protocol recruits at the top of the funnel, too. Weekly competitions are run by DecentralProp rather than by any one firm: a small entry fee buys a $100k contest account, and the winners are handed real evaluations bought by the protocol at real firms, spread across the basket rather than concentrated. A firm can therefore receive a funded-track trader it never paid to acquire, and the trader arrives already holding an account.

None of this could run unattended without the two engines underneath it. A self-reinforcing money loop with no supervision is how protocols die. The Autonomous Risk Engine re-scores every firm every 15 minutes and moves its risk posture without asking anyone, so a firm whose treasury is thinning tightens automatically instead of after somebody notices. The integrity engine watches for the abuse patterns that a shared engine makes possible, and escalates its own posture when it sees them. Growth is throttled by machinery that has no incentive to look the other way, which is the only reason the rest of it can be left alone.

Ten loops, in short: token bootstraps liquidity, liquidity magnifies payout capacity, payout capacity attracts traders, traders pay fees, fees deepen the treasury and buy the token, payouts buy and burn the token, payouts produce proof, proof recruits traders, traders amplify the proof, and the risk engines keep every one of those loops inside a solvency band nobody can argue with.

No human in the operational loop

Risk. The Autonomous Risk Engine reads eight live signals about the firm's health every 15 minutes, classifies it into one of four tiers, and sets payout pace, evaluation difficulty on new purchases, and leverage on new funded accounts to match. There is no override console above it and no operator dial beneath it. The same engine, the same thresholds, every firm. See The risk engine and How the risk engine thinks.

Two properties of that 15-minute cadence are worth stating, because they're what make an unattended firm safe rather than merely cheap. It escalates fast and relaxes slowly: a firm whose treasury drops sharply can be forced to a stricter tier immediately, while returning to a looser one is held behind a timelock, so nobody can wait out a bad afternoon. And it is the same code reading the same thresholds for every firm on the protocol, which means a firm cannot negotiate its own risk posture, and an operator watching their own tier tighten has exactly as much recourse as a trader watching it: none, and the reasoning is published either way.

Fraud. The integrity engine is the second engine running underneath everything, and it exists because a shared engine creates abuse that no single firm could see. It builds a cross-firm view of who is actually behind an account, watches for the patterns that only make sense as manipulation (hedged pairs across firms, copy-trading rings, funding lineages that converge), and escalates its own posture in four steps when it sees them. Detectors that could be innocent are advisory and score nothing on their own, so a trader who legitimately trades at five firms is not treated as a ring. Holds are automatic, bounded in time, and backed by a proof anyone can replay. See How integrity detection works.

Together these two are the answer to the obvious objection about everything else on this page. A self-reinforcing growth loop with nobody supervising it is how protocols die. Here the supervision is machinery with no discretion, no incentive to look away, and no ability to make an exception for a firm it likes.

Passing and payouts. The rulebook is platform-fixed and enforced by the engine as you trade, so hitting the target is the entire pass decision. There is no desk that reviews your style afterward and decides whether your winning was legitimate. An earned payout enters a queue and is delivered by protocol jobs against the four-layer waterfall; an approval step you could be stuck behind simply doesn't exist. The one thing that can hold a payout beyond the engine's published pace is the integrity engine, and even its holds are automatic, bounded in time, and backed by a proof anyone can replay, held is not seized. See How integrity detection works.

The books. Every trade is committed into an hourly tamper-evident record on Solana, settlements are provable, and a fraudulent one can be challenged by anyone and slashed. Nobody reconciles the books at the end of the month, because the books were never in anyone's hands. See How settlement actually works.

The firm's own lifecycle. Even the end is automatic. There is no voluntary shutdown; the only exit is bankruptcy, and that path is code-gated on paying every trader in full first. See Operator risk & lifecycle.

Marketing runs itself too

Autonomy would ring hollow if the firm still needed a social-media hire to survive. It doesn't, and this is the part of the design worth understanding in detail, because it is where most of the industry spends most of its money.

A prop firm's marketing budget usually goes to three things: producing content, buying distribution, and paying affiliates. This protocol replaces the first two outright and settles the third on-chain.

The content produces itself

Nine things that happen on-chain trigger a post, with no human involved at any step:

The copy is the operator's own template, approved once, with live values (treasury depth, token price, happiness score) resolved at post time rather than typed, and the payout certificate image attached where one exists. Scheduled one-off and recurring posts use the same variables.

The editorial rule this produces is stricter than most staffed firms manage: the firm cannot post a claim about itself that an on-chain event did not cause. There is no "we're the best-paying firm" post, because no event emits that.

The distribution is the firm's own traders

Publishing and being heard are different problems, and a scheduler posting into an empty timeline solves only the first. A new firm's account has no followers. Its traders have thousands, collectively, and they are the audience a prospective customer actually believes.

Every auto-post lands in the Earn on X feed of every trader at that firm. Liking, reposting, quoting or replying earns them airdrop points, priced per action: 0.25 for a like, 0.5 for a repost, 0.75 for a quote, 1 for a reply, plus a one-time 2 for following. A quote pays most of the repeatable actions because it is a post on the trader's own timeline carrying their own words, which is the version a stranger actually reads.

The effect is that a firm with 200 traders has 200 people with a standing financial reason to amplify its payout receipts, every single time one settles. They aren't influencers on a retainer. They're customers, posting a verifiable receipt from a firm they actually use, to an audience that knows them.

The 10,000-point monthly budget

Every firm deployed on the protocol is given a pool of engagement points each calendar month, funded by DecentralProp rather than by the firm and sized by its deploy tier: 5,000 at Starter, 10,000 at Growth, 20,000 at Pro, 30,000 at Scale, 60,000 at Enterprise. It resets on the 1st, does not roll over, and costs the operator nothing: no SOL leaves the treasury, no invoice exists, and the operator never tops it up.

Points are anchored to real money everywhere else in the system, at 1 point per $1 of evaluation fee paid. So a 10,000-point pool hands out, every month, the same airdrop credit a trader would earn by spending $10,000 on evaluations. That is the honest way to size it: it is a five-figure monthly marketing allowance denominated in the protocol's own distribution rather than in cash. What a point converts to in dollars at launch is genuinely not known yet, because the $DPROP$DPROPThe protocol-wide token (not yet launched). Funded by a small slice of every evaluation fee across every firm. supply share and claim mechanics haven't been set, and anyone quoting you a figure is guessing.

What that allowance actually buys, at the per-action rates:

For a firm posting 60 times a month with 100 highly active traders, real usage lands near 10,500 points, marginally over the default. The allowance is deliberately generous for a young firm and gets tight at scale, which is the right direction to err, and it is expected to rise.

Why it can't be farmed, and why that matters to the operator

An amplification budget paid in a token-adjacent currency is an obvious farm target. Two ceilings stop it, and they are the same two that protect the airdrop itself:

• The firm's pool is fixed. A firm cannot buy more reach than its monthly allowance, so the channel's cost is known in advance and cannot scale with its own success. Whether firms will ever be able to purchase additional allowance is an open decision, deliberately unresolved, because an allowance a firm can mint is a token pre-sale in a marketing costume.
• The trader's ceiling is their own spend. Engagement points count toward a trader's total only up to half their own purchase points. A wallet that has never bought an evaluation can amplify everything a firm ever publishes and redeem exactly none of it.

For an operator, the second ceiling is the important one: it means the people amplifying your firm are, necessarily, people who have paid for something. You are not buying reach from bots. You are rewarding customers for carrying a receipt.

Nobody has to remember to do it. Traders who have linked the firm's Telegram bot get a message when fresh posts land, capped at one a day and sent only when the pool can still pay for the engagement, so the loop closes without the operator prompting anyone.

The rest of the marketing stack

The Telegram bot. Every firm gets its own Telegram bot, and it's a full storefront and terminal, not a notification feed. A trader can create or link a wallet, deposit, buy an evaluation, trade through the built-in Mini App, and withdraw, all without leaving the chat. Members can summon the firm's shareable stat cards into any Telegram conversation, even ones the bot was never added to, which turns every group chat a member sits in into a recruiting surface. Add tips and giveaways inside the community, plus automatic announcements when traders pass or get paid, and the result is a channel that onboards, sells, and promotes around the clock with no admin behind it.

Referrals. The affiliate program follows the same pattern: the operator approves who's in, and every referral split after that settles on-chain automatically at purchase time. Details in Growing your firm.

The metric. The one piece of marketing the operator can't write is the firm's public Trader Happiness ScoreTrader Happiness ScoreA public 0–100 score shown on every firm's storefront, computed from that firm's real trader outcomes rather than set by the firm itself., a 0–100 number computed from real trader outcomes. A well-run firm advertises itself.

The protocol recruits for you. DecentralProp runs weekly competitions at the platform level: a small entry fee buys a $100k contest account, and winners are handed real evaluations that the protocol buys at real firms. Awards spread across the basket rather than concentrating, and a prize can only be placed at a firm whose treasury is deep enough to have sold that account size anyway. A firm can receive a funded-track trader it never paid to acquire, whose evaluation fee was paid in full, splits and all.

The comparison table. Every signed-in trader can compare every firm on the protocol side by side, ranked on proven payouts, solvency and payout speed, with the weights published on the page. Placement is not purchasable, which means the only way to rank is to pay traders reliably. See Your wallet hub.

Why the whole marketing stack is defensible

A staffed competitor can copy any single piece of this. What they cannot copy is the reason it works: the content is caused by settlements rather than written by a team, the distribution is done by customers holding verifiable receipts, and the budget is denominated in a currency the protocol issues rather than in cash it has to earn first. Those three properties come from the same architecture that makes the firm trustless. You cannot bolt them onto a business whose payouts are a private database row, because there is no receipt to post and nothing for a customer to verify.

What a person still does

Autonomous doesn't mean ownerless. The operator picks the brand, writes the storefront, prices within a guided band, chooses the profit split preset, approves affiliates, and writes the marketing templates everything above posts from. Leverage is not on that list, it's platform-set and the engine lowers it under stress. Traders still trade entirely by hand; the engine never places a trade for anyone. And one honest boundary: the protocol's code is still written and upgraded by people, and no third-party audit has happened yet, see Security & audit status.

Why autonomy is the trust model

Removing the humans was never about payroll. A rule applied by a person can bend, for the biggest account, on the firm's worst week, exactly when it matters most. A rule applied by the engine binds every firm identically, and enough of what it does is published that you can recompute its decisions yourself. The risk desk doesn't take calls. That's the same design choice behind why a firm can't rug you, and it's checkable rather than asserted: see Verify it yourself.