yeah this mostly just says there aren't many good app-layer ai products yet, which is a skill issue on app layer cos' parts + timing issue of the nascency of the app layer. product intuition to figure out how to work *with* and not *against* the non-determinism of ai to create delightful yet reliable products is a harder problem than the 100s of apps that flooded the app store after ios launched.
it's very simple really - just whether the agent is proactive vs reactive
proactive agents should be ebullient, high EQ, heat-seeking missiles for pain
reactive agents should be meek, salt-of-the-earth ground-dwellers that speak only when spoken to
I strongly believe the opposite is true: @-mentioning (or explicitly invoking) an agent in Slack is the superior experience because it's the clear, and exact, handoff between a human conversation and work for the agent. If/when your ambient listening goes wrong in the middle of an incident, you're now trying to course-correct both humans and agents at the same time.
When people switch to Inspect from other tools, they think they miss ambient agents as the main UX, but they quickly realize that tagging is the better experience.
granola getting into voice dictation...wispr getting into notetaking...willow getting into both plus rewriting... oh it's about to be knives out in here
@staysaasy just like hundreds of saas cos in the 2010s iterating on some kind of auth scheme, billing, SSO, and an all-in-one observability platform
Just a million people doing the same thing.
founder/VC trivia night:
- every trivia question relies on alpha you could've only milked from first party sources
- no answers are validated aloud
most terminally gossiping team wins
You are a struggling second tier competitor that is more unethical and lying just because you have no decency or sense of proper behavior and shows your desperation. Straight out lying about if Chris being fired I thought would be below even you.
gtm world tour
tomorrow
speakers:
founder appdirect (1.5B) + landbase(42.5M raised)
founder smartly (100M ARR) + zerogtm (10.3M raised)
for revenue leaders interested in the future of ai & gtm
RSVP while tickets are available
@VirioAI@danielsaks@AppDirect @LandbaseGTM @triekki@smartlyio@zero_gtm
3. jev guessing the most likely reactions that a linkedin post will get
- the most likely reaction on any post is Like
- jev + heuristic-based prompt = can guess the second-highest reaction with 90% accuracy!
all the best things that my team has built with jev in the past week:
1. real-time AI writing detection
as soon as text is written, jev highlights the phrases that are most likely to signal AI writing
2. jev for clustering/knowledge graph construction (vs LLM)
- jev is bad at clustering. it keeps deciding to make new groups instead of adding to existing groups
- no deterioration in decision making ability with higher cardinality of choices, the way that early LLM agents did with tools (it's just bad throughout lmao)