Answered from live records · 2026-09-15
How is Sato Swap different from routing directly through an aggregator?
It does not replace an aggregator: it asks several and adds four dated verdicts, a server-side policy, a simulation and a signed record. The cost is our fee and one more hop, as of 2026-09-15.
Ten dimensions, stated as facts on both sides. NO WINNER IS DECLARED and none can be: which column is right depends on how many venues you touch and on whether your limits have to survive an agent's own reasoning. The last row is the honest one — for a single venue you already run against, with limits in your own code, routing directly is fewer moving parts, and this page says so.
| Dimension | Calling an aggregator directly | Through Sato Swap |
|---|---|---|
| Who quotes the trade | The aggregator. It is the one with the liquidity graph. | The same aggregators. Sato Swap is not a liquidity source and does not compete with one — it asks several and reports which answered. |
| Venue choice | Yours, made before you knew whether it would answer. | Ranked inside one lane on liveness, the observed record of our own daily checks, verification state and only then quoted output, with chosen_by naming every field read. |
| Pre-trade checks | Whatever you wrote. Many agent integrations write none. | Four verdicts — both tokens, the venue endpoint, the recipient — each with the rule that fired. Unknown refuses by default. |
| Limits | In your own code, or in the prompt, where a message can argue with them. | A server-side policy evaluated after routing and before building: caps, allowlists, slippage and deadline ceilings, mode restrictions. |
| Simulation | Yours to run. | Run before the transaction is handed back; a reverting route is withheld with the decoded reason. |
| Custody | None — an aggregator API of this kind returns calldata and signs nothing. | None, for the same reason. Adding a layer here does not add a custodian: what comes back is still an unsigned object. |
| Cost | The venue's own fee, plus gas. | The same, plus 3 bps stable-to-stable or 15 bps otherwise as a parameter on that venue's quote, disclosed before anything is signed. An unsigned trade pays us nothing. |
| Failure modes | One venue's outage is your outage. | A venue that does not answer is ranked below one that does — but the failure surface is larger, not smaller: this call depends on our availability as well as the venue's. |
| Dependencies | One API you can pin and read end to end. | One more hop, run by us. That is a real cost and it belongs in this table rather than in a footnote. |
| When direct is the right answer | One chain, one venue you already run against operationally, limits already enforced in your own code, and no need for a signed record of the decision. | Several venues or chains, a policy that has to hold against an agent's own reasoning, or an audit trail of what was checked before signing. |
How this was answered
A dimension-by-dimension comparison of a method against a CLASS of API — no aggregator is named as the opponent, because Sato Swap routes to several of them. Every claim on the Sato Swap side is a property of chooseVenue(), evaluatePolicy(), simulateSwap() or quoteBinding(); every claim on the other side is a property of the pattern, not of any one product.
Every answer on this page is produced by one named function over Sato Hub's live records on the date shown. Nothing is typed by hand, no figure is carried over from a previous day, and an answer with nothing behind it says unknown rather than zero. The Sato Score measures how open, active and verifiable a project is — it is not a safety, quality or returns grade.
Go deeper
More questions
Sources — the projects' own repositories and docs
Cite this page
Sato Hub. "How is Sato Swap different from routing directly through an aggregator?." Sato Hub, updated 2026-09-15, accessed 2026-09-16. https://satohub.ai/answers/how-is-sato-swap-different-from-routing-directly-through-an-aggregatorData last refreshed 2026-09-15; this page is rebuilt daily. Citations carry the date so a reader can tell which snapshot a claim came from. Catalog data is licensed CC-BY-4.0 and the machine-readable copy is linked in the page's Dataset metadata.