Jay B retweeted
Your agent got approval for one change and shipped a different one. Our 16-case lab shows we have solved that problem. SYNCStaff binds approval to the exact change a person reviewed, so your team ships more and grows faster. syncstaff.ai/blog/ai-guardra…
Replying to @fuzz_loop @AlexGDimakis
That’s a lot of work for you, no? We did something a bit different.
10
Replying to @mishraPriyan97 @AlexGDimakis
Right, most people aren’t considering that and they are like look!!! I can spin up 100000 agents and get it done in less than a min lol
13
Replying to @mishraPriyan97 @AlexGDimakis
I’ve noticed that a lot of benchmarks just go off completion but don’t penalize overwrites/conflcits. ParallACT benchmark kinda nailed that one.
1
10
Replying to @fuzz_loop @AlexGDimakis
This is the real point. But certain work will drift or could drift unexpectedly. How do you know there isn’t overlap?
1
15
Replying to @EternitiesAI @AlexGDimakis
That’s due to lack of persistence of context though… right?
22
Replying to @AlexGDimakis
This is great! We have a product for multi agent coordination and it rocks. The bench mark we are currently running against is parallACT because it focuses on the conflict or collisions. Do you guys do that?
1
47
Nice, does this help manage the work as well? Meaning feature requests etc?
We have something that addresses that if of interest.
6
Replying to @carlomakes @myalya
Do you run multiple agents on multiple repos with others? Like a team?
1
1
4
Replying to @pablomanjarress @myalya
Humans role in the future… for now.
Jk, we already solved that w AI lol
1
Replying to @swordBluesy @AravSrinivas
That’s what I’m saying. What is the use case for this tool? Like yes. Cool.
1
1
15

