🔷 [Epic] AI Agent Integrity

charleslavon commented 8 months ago

What ?

High Level Goal:
Build towards a vision of having trusted smart agents work on your behalf across chains.

Why ?

Background:

Similar to being the DA layer for ZK operations, how do we become the Data Integrity layer for AI? Build features for verification of agent operations.

How ?

Research in support of Eth Denver talk
Plan future alignment with the AI/Edge workstream
```
### Tasks
```
[x] Agent Workshop #1
[x] Detail exactly what data is written for a use case and example Tool wrapper
[x] Record Rough draft of talk
[x] Deliver talk at meetup on 2/21
[x] Internally share EthDenver version of talk on 2/27
[ ] Describe standard LangChain Tools for AI Trust
[ ] Build one of: FetchState, HumanApproval, RecordDecision, or TakeAction Tool wrappers

ilblackdragon commented 8 months ago

Hm, why does this live here?

charleslavon commented 8 months ago

@ilblackdragon @gabehamilton has pitched a talk at ETH Denver on this subject. This task is tracking his time on research in support of that. Do you see an opportunity to better align this with other AI work that's going on?

ilblackdragon commented 8 months ago

Two points:

this is near discovery tasks so seems strange to have an Epic like this.
and yes, there is a separate conversation on priorities around some of the short term product work that @gabehamilton is part of.

gabehamilton commented 8 months ago

Created a new repo and moved this issue there. Either or both can be cleaned up and rearranged as priorities get set.

near / ai-platform