When it comes to zero-knowledge systems, most designs eventually face the same choice.

zk-SNARKs or zk-STARKs.

Both are powerful. Both have tradeoffs.

For an infinite compute layer that needs to push massive historical data off-chain, @brevis_zk made a very deliberate decision.

They built around zk-SNARKs.

Why this matters:

→ Extremely small proofs keep on-chain verification cheap
→ Fast verification enables near real-time dApp logic
→ Tooling is mature and battle-tested
→ Ideal for summarizing large historical datasets efficiently

This lets a dApp prove something like
a user executed 100 swaps in the last 30 days
without revealing which swaps or any sensitive details

All compressed into a proof only a few hundred bytes long.

Simple design choices like this are what make large-scale verifiable compute actually work.