Skip to main content
Ash Ashutosh

Ash Ashutosh

CEO of Actifio

Aug 11, 2026

2:46
And I'd like to, if you could talk about that a little bit, about how that actually works.
2:50
Yeah.
2:51
We started seeing late last year where a new kind of user started showing up and using a lot of our APIs.
3:00
If you recall, Pinecone provides a fully managed vector database service.
3:04
And most of our customers come in using the APIs.
3:07
The humans would query, get a response, query again, get a response, get their task completed.
3:14
And we saw a very different user called an agent.

10 MINS LATER

13:38
So Ash, you've organized Pinecone in a way that's really focused on initiatives, people running things end to end with AI filling the gaps, right? A few months on now, I mean, what have you learned from that so far? What's working? What's challenging? What do you see happening in the future?
11:39
Okay.
11:39
When you move the reasoning closer to where the, the data is, closer to where the, the curation of the data, where the actual processing of the data is happening, you can do a lot more things.
11:53
So for instance, you can get the right kind of data because now you know what context I'm addressing for.
11:59
More importantly, you can start citing and attributing, so you actually can say, "This is the citation of why and where this, this answer came from," as opposed to an LLM would not.
12:10
It just probably talks to some MCP server, gets some informationAnd brute force is a way to some answer, whether it's right or wrong.

11 MINS LATER

23:36
... what for Nexus, how does all that work, and what was the innovation there?
23:41
Yeah.
23:41
So once we built Nexus and you have an engine where you could have an agent define what its task was and what, what kind of a knowledge e-engine it needed, it just didn't have a way to specify that.
27:28
but
27:28
the AI algorithms required to process data in memory are different because they were designed for different things than on the disk, than on object storage.
27:37
So being able to not only just structure it so that the read paths and the write paths don't conflict, not only do you have to worry about making sure that the, we call them slabs, you tier how the data is in the right cost envelope in the right stage of its life, but also the algorithms required to process that data dynamically happen.
27:57
I mean, you could do this manually.
27:58
You can say, okay, now turn on this knob, turn on that knob.
28:01
That is genius work.
28:02
That is an absolute genius work to be able to pull off something as complicated as it.

8 MINS LATER

35:56
is around security.

We value your privacy

We use cookies to understand how you use our platform and to improve your experience. Click “Accept All” to consent, or “Decline non-essential” to opt out of non-essential cookies. Read our Privacy Policy.