The CoW AMM integration into Tycho has taken longer than expected, mainly due to delays in the integration process. A lot of time went into debugging LP token swap logic and resolving tricky testing issues, which pushed this off the original schedule.
The good news is that the heavy lifting is done and we’re back on track. You can follow progress directly here:
Hi,
Thanks kindly for the update on this matter. The grants committee are happy to see work resume on the grant. Please however not that extensions are subject to the committee’s discretion and we would very much like to see concrete deliverables in the milestones within 30 days
If there is anyway we can help with reviews please be sure to ask!
@tanayj Hi, we’re just kindly following up as to the progress of this grant if you can please provide an update on the go-live period of time that you mentioned?
Thanks for the follow-up. Here’s a brief progress update on the integration:
Since the last update
The integration advanced through the main Tycho SDK PR (#191) and the cowamm-execution repository.
While nearing completion, the bounty hunter’s implementation still had several gaps. Our review identified:
• Non-standard crate setup and redundant local copies of CoW crates.
• Unnecessary orderbook posting logic within the encoder.
• Inconsistent project structure and undefined references affecting build reliability.
• Solver interface types that limited compatibility with Tycho’s runtime.
We’ve shared detailed feedback and recommendations, and the contributor is currently implementing the fixes.
The revised timeline is that we expect the integration to be live and complete by the end of November. We’ll share another update once the revised version is merged.
We appreciate the committee’s patience and understanding. Please let us know if you have feedback on the implementation. We’ll share another update once the revised version is merged.
Since the last post, the core structural and encoding work was in place, but the Tycho test suite surfaced issues that required deeper fixes. To keep the integration moving, we stepped in alongside the bounty hunter and took on a substantial portion of the remaining implementation.
What the tests revealed
• Bind() liquidity arrives before the pool is formally created, leading to balance mismatches in Tycho’s model.
• Duplicate balance deltas and ordinal collisions.
• Simulation not producing quotes until indexing was corrected.
• Indexer diverging slightly from RPC state when replaying deeper history.
What’s now fixed
• Bind deltas are re-anchored to pool creation so Tycho reconstructs state correctly.
• Store logic updated to prevent duplicates and ordinal clashes.
• Execution tests now pass end-to-end.
• All balance-changing events are now handled (binds, joins, exits, LP changes, trades).
What’s left
A small indexing drift still appears when replaying very deep history (~2.18 wstETH on one pool). It doesn’t affect quoting, but we want to close this out before calling the integration complete.
Timeline
We expect the integration to be finished by December 15 (latest).
We’ll share another update early next week. We’d like to thank to the committee for their continued support and understanding.
Thanks for the update and for keeping progress moving forward as of late. Please drop us another update next week re: status, given we’re close to the 12/15 target date.
Unlike most Tycho integrations, CowAMM requires some solver-specific encoding due to its design.
Combined with the current size of the CowAMM pool set (3 pools with meaningful TVL) and reduced solver bandwidth during the holiday period, this may delay the first routed transaction.
We expect to see a CowAMM transaction routed by a Tycho solver by January 16, and will continue pushing to accelerate this where possible.
We appreciate your continued support and patience - thank you.
Hey everyone! Following up here with another progress update -
Since the last update, we resolved an indexing issue uncovered during dev sync related to the CowAMM substreams implementation, and the fixes have now been incorporated. Indexing and simulation run successfully with the updated logic, and no further changes are expected for Milestone 1.
The remaining step is a clean dev sync and final review, after which the existing PRs will be merged and made available for solver consumption.
Due to the substreams bug encountered in the bounty hunter’s implementation, we now expect the first CowAMM transaction to be routed by a Tycho solver by the end of this month.
We’ll share another update once the dev sync completes. Thank you for your continued support and patience.
CoWAMM has been succesfully integrated into Tycho .
All issues with indexing that surfaced in the dev sync have been fixed and solvers can now use Tycho to index CoWAMM’s latest state and simulate swaps through it.
As noted in earlier updates, the delay was due to the bugs in the external bounty hunter’s implementation. Our team took over this engineering effort to actually push this through the finish line.
Any cowswap solver looking to route through CoWAMM can DM me on Telegram (id - tanay_j) for support on this.
As a next step - we are now working with Tycho users (who are cowswap solvers) to ensure they are supporting CoWAMM via Tycho. We are aiming to share transaction showing a Tycho user routing through CoWAMM in the next 10 days.
Thank you everyone for your support and patience on this.
Thanks @tanayj for the update on this. Please share your next update as soon as possible we’d like to get this wrapped up and the grant closed out ASAP.
.