Ape/rture

@aperture

137 Following

82 Followers

Brock pfp

@runninyeti.eth

Working on something new this holiday season - Just In Time Indexing for all (even agents): https://www.jiti.ai/

1 reply

3 recasts

8 reactions

Ape/rture pfp

Thanks! Since this might be very use case dependent we typically don't predefine the data structure (as in the format/schema) and do that custom with the builder to make the model training as easy as possible. That said, most onchain data is fairly structured already like DEX trades. I think where structure is very important is when you want to add labels

0 reply

0 recast

0 reaction

Brock pfp

@runninyeti.eth

Quick transformer for decoding V2/V3 swaps simultaneously (this looks at pool events, could also go by token transfers): https://console.indexing.co/flow/675ebd65085be53512a3a323 There's no limit on number of chains nor addresses being watched. And it can all be piped to any db, webhook, etc. There are also APIs to dynamically handle adding / removing watched addresses. Standard pricing is $1 per 1000 blocks *processed* - blocks without matching events won't count --> likely in the $10-100/m range at this point for this use case (and yes, $LARRY would be accepted). dm me if you're interested!

0 reply

1 recast

3 reactions

Ape/rture pfp

SuperchainERC20s are here. An overview: https://x.com/ape_rture/status/1866823400972251334

0 reply

0 recast

0 reaction

Ape/rture pfp

AI agents are taking the spotlight, but for these agents to do complex tasks they will need on-chain data. There's complex data infrastructure needed to serve this data. This is how a serious AI data pipeline will look like: https://theindexingcompany.substack.com/p/serving-ai-with-data-infrastructure

1 reply

1 recast

5 reactions

Ape/rture pfp

AI agents are taking the spotlight, but for these agents to do complex tasks they will need on-chain data. There's complex data infrastructure needed to serve this data. This is how a serious AI data pipeline will look like: https://theindexingcompany.substack.com/p/serving-ai-with-data-infrastructure

0 reply

0 recast

0 reaction

Ape/rture pfp

Can't wait for the rap album to drop

0 reply

0 recast

1 reaction

Ape/rture pfp

I have to deny i am an AI agent, but actually a real monkey building indexing and data pipelines at @indexingco But thanks for the shoutout nonetheless

0 reply

0 recast

0 reaction

Ape/rture pfp

Indexing Co can provide direct FC data to DB streams for like $50 per month. Any specific data you need or just want it all?

0 reply

0 recast

0 reaction

Ape/rture pfp

Ready for Season 2!

0 reply

0 recast

1 reaction

Ape/rture pfp

Join this Covariance call to see how new chains can be served with good infra. I'll go over the current problems with this explosion in number of chains and how we solve that (+ sneak peak at the new /indexing Console)

0 reply

1 recast

1 reaction

Ape/rture pfp

Hear me out: websites that use login through an email code, but that email arrives instantly instead of taking 2 minutes

0 reply

0 recast

1 reaction

Ape/rture pfp

Builders unite and be amazed by the wonders of modern technology

0 reply

0 recast

1 reaction

Ape/rture pfp

Just In Time Lore

0 reply

0 recast

1 reaction

Ape/rture pfp

Crypto is a attention-driven economy. There can be a massive value capture for users to be paid for their attention by businesses (apart from token emissions) whether that is user-research, feedback, watching ads. Various platforms experimented, but none yet being successful.

0 reply

0 recast

0 reaction

Ape/rture pfp

Since I see Once Upon being mentioned in the comments: we used to provide the data for them. I could spar with you on how you can get that data or if you have the data inhouse how to do those estimations on the data you own. One benefit then is that you own that logic and do not depend on anybody else for a data model.

0 reply

0 recast

1 reaction

Ape/rture pfp

0 reply

0 recast

3 reactions

Ape/rture pfp

I gotta admit I am doing a really bad job at casting here. Any topics or views you like to see me cast? [Dance Monkey]

2 replies

0 recast

2 reactions

Ape/rture pfp

This man is living in the future

0 reply

0 recast

1 reaction

Ape/rture pfp

Using heuristics and top down thinking on the data makes the classification process easier because the results just make more sense. The ML model would just be a bonus where there might be some patterns related to those heuristics being picked up. We can't ask the MLs to understand what humans think is valuable in this data (yet) without any guidance. It seems sometimes devs in crypto forget the classic data science cycle: ad hoc research/reports > automate that > then decide if you have do dig deeper with extensive tech like MLs.

0 reply

0 recast

0 reaction