Redis

@Redisinc

See how fast feels

Redis
@Redisinc
19h
Add a filter to a vector search, and results can get worse, not better. It sounds backwards, but a tight filter can make the search skip right past items that would've matched, with nothing to flag it. 🧵
1
Redis
@Redisinc
Sep 2
97% of enterprise leaders say context is critical to AI strategy, but only 4% have built for it. Memory resets. Retrieval lags. State goes stale before your agent reads it. And bills nobody budgeted for. Redis Iris fixes the layer underneath the model. 🧵
3
Redis
@Redisinc
Sep 1
Most LLM APIs have no memory. Every request resends the full conversation, so costs pile up and quality can degrade as it grows. That's what five-line quick-start guides skip. 🧵
2
Redis
@Redisinc
Aug 31
94% say compounding context is essential to production-grade AI. Only 4% have actually reached it. Join us for a live office hours session on September 9, digging into why, plus why stale data still trips up even the most mature teams. Save your spot here: redis.io/resources/vide…
2
Redis
@Redisinc
Aug 27
Merging ranked lists with reciprocal rank fusion is nearly free. The expensive part is getting the candidates in the first place, and agents often retrieve several times per task, so every millisecond compounds. 🧵
9