Video: How Weaviate Is Revolutionizing Search With AI

hey Channel insiders I'm your host Katie boso and welcome to partner POV today we'll dive into the Innovative world of open-source AI native Vector databases with we8 if you're struggling to retrieve the right results through keyword searches in your business or you didn't know there was a smarter approach to information retrieval stay tuned we8 is redefining SE SE with AI powered semantic search making it possible to find information faster and more accurately for more on we8 and the real world outcomes its team is driving for customers I'd now like to welcome Byron vorbach field CTO of we8 hi Byron welcome hi ktie thank you for having me thank you so much for being here Byron tell me about we8 what does the company do and what challenge or challenges was it born out of so wv8 is an open source AI ative Vector database and what we do is we solve challenges for companies to search through data semantically rather than keywords and what this means I I'll touch upon in a little bit but it was originally started a little bit over five years ago uh back when Google released Vector search for the first time that was the first time that people actually were able to search with natural language rather than with keywords and our founder Bob F felt like this technology should be capable for for everyone to implement and that's where he started we8 we8 was originally built open source so that any company can start using it and the challenges it's solving is mainly being able to get more benefits from the data that you have already stored because we're removing the limitation of only searching through data keywords really interesting and to think that it came all the way back from when Google started doing this and you could build this interesting solution so going off of that where do you see traditional search techniques fall short you did start to touch on this but can you provide some specific examples to paint a picture imagine going to a website and you're looking for gifts for Mother's Day Mother's Day is coming up pretty soon the downside for for these types of search questions is that it's quite difficult from a technical perspective to map that user question to a set of products these products must have been labeled with information about Mother's Day or that they're specifically tied to a female gender and that kind of processing and data augmentation cost a lot of time my personal experience I've worked at similar type of companies where there were dedicated Personnel working on on tagging additional information to make sure that whenever a user searches with keywords that the responses are actually matching those individual keywords but if you look at a search term like Mother's Day it's kind of fake you know semantically if you walk through a store you can see these different aisles with different products and then you automatically think well that's that's perfect for Mother's Day think jewelry perfume or anything else that you would get from your mother but this data typically doesn't hold that information what semantic search does it has been trained by a machine learning model that's been trained on data of the world so any type of information that's found online and what it's capable of doing is translating the input search term of Mother's Day gifts to something that can be matched semantically with things like perfume jewelry Etc so there's no need for people to maintain keywords there's no need for people to maintain different data objects and them with additional information because we now have these machine learning models that are able to map these user questions to actual products so what you're saving is we V8 is saving the mother child relationship one relationship at a time these days exactly exactly to forget Mother's Day no exact exactly exactly and this is just one example um but there are many others in all different types of industry and domains where semantic sech as an initial start already has shown a lot of impact and and and different well I'm excited to use it in the future if I can get access for for that Mother's Day gift but moving on from there Byron we hear the term AI almost every day it's very much a part of my daily life and many businesses claim they've built it into their Solutions what makes your business different and stand out when it comes to how people Implement AI with we8 that's a great question and we get that question a lot so I work a lot with all our different types of customers both with prospects and actual customers and everyone right now is not not just claiming that they're building with AI but it's the way to talk about these new type of applications our like Vision on AI applications or applications where the large language model so the model that can do the uh the prompting and the generation of new data sits at the heart of the application in order for these large language models to work they need a database that's able to scale that's able to grow that's able to give good performance and fast results and good quality results and we see ourselves as a core component of these new type of infrastructures the combination of having a large language model with a store that's purpose built for these applications is for us what we call the AI applications that people are building so it's a combination of multiple factors It's a combination of the large language model It's a combination of the vector database chosen but also the application that has both of these two applications at the core of their uh core of their architecture how does being open source give weeva customers an advantage that's uh that's a great question as well open source means that companies can try out software before they buy it I think that's one of the examples where open source shines and especially since we originally cared a lot about the developer experience a lot of developers could just pull our software try it out see the benefits before they decide to do an actual implementation or do a large scale purchase in order to to to actually implement the software so I think open source has a benefit of eat of use initially another one is the capability of of running things in your own environment keeping data close to you not sending it to third parties is something that we hear back from a lot of customers that they see benefits in especially with a lot of the data getting scraped and data being used by other companies keeping your data Clos is a big requirement for a lot of companies and being open source means that customers can run uh our software in their own environments of course at the same time we can still support them so being an open source company our main activities with customers are helping them Consulting to them but also being able to help them run it in their own environment uh as part of the open source deck well let's keep getting to know who your customers are what types of businesses and verticals overall have benefited from leveraging wv8 the most that's a good question I think it's pretty much everywhere especially before the rise of CH GPT we didn't see too many customers in too many different areas mainly because it was pretty unknown not a lot of people saw directly the benefit of being able to ask natural language questions to your data but as you probably also used jgpt yourself there are definitely benefits to be able to ask questions and get a good constructed response back and in combination with vv8 companies can do that on their own Enterprise data we initially saw a lot of use cases just doing the VOR search part the second type of use cases which especially came around last year is what we call Rack use cases or retrieval augmented generation what that means is a lot of the chat Bots and the Q&A Bots that you see online so people are able to ask question questions on on websites and and get directly answers back that's sort of the second wave of application and those are across almost all vertical because that's mostly used for internal use cases so being able to ask questions to an ahr chatbot and large Enterprises all the way to like I mentioned before in e-commerce where we're able to describe what we want to buy instead of what we actually exactly know what we're looking for also into finance and legal being able to ask questions and and generate answers based on legal documents so there's almost no industry that's that we haven't seen or that that is not being yeah doesn't have any impact from from all of these advantages that we've seen lately so in addition to we8 helping me find a Mother's Day present for my mom can you share a customer success story with me in which a real life customer of yours managed to change their business operations for the better thanks to we8 one example is a is a large scale emo provider company that we're working with they have the capability now to search and ask questions on top of all your email so they store an x amount of email for a couple couple times back specifically so that you can ask questions like when is my next flight when did I stay at the Hilton and location y or how many PDFs have we shared with customer X those type of questions you yeah you typically have when you go through your email and you try to search who was in that email thread again being able to offer these kind of uh capabilities now especially in combination with some of the features that we8 has to to run this cost efficiently has changed the way they are able to onboard new users and able to add more data and add more emails for these users if you would compare this with other Technologies out there in order to run a similar type of use case we noticed that BV could run this a lot cheaper was able to leverage different types of functionalities one example there I want to highlight typically companies have to pay for the data that they have running for uh for a use case but in the case of email I only look through my email maybe 8 maybe 12 maybe sometimes 15 hours a day depending on the workday but the other times I'm hopefully sleeping during that time it would be a waste for the company to to be paying for for all that data but I'm not actually online able to leverage it and this is one of the features that we've deployed and and developed and deployed for this customer specifically about half a year to a year ago and we're now seeing massive impact for them to do cost savings as well as for other customers that have similar types of use cases so really like that success story the company is growing very quickly now because of some of the things that we build in the background so that's that's definitely one that sticks out that's so interesting to think about how you're able to scale to the the budgetary needs of your clients so I love to hear it in hearing all of this about we V8 I can't help but think about how heavy it might be so to speak how easy is it for customers to integrate we8 Solutions into their environments and workflows the easy ansers is quite straightforward uh being open source We Care like I mentioned before we care a lot about the developer experience so wv8 is able to integrate with any type of framework that companies use for for integrating databases into their workflows we're able to run on any Cloud uh we're able to run on customers own infrastructure and some of those benefits really help with being able to set weev it up get it up and running next to that we also offer free uh sandbox accounts so if you just want to play around and you want to try things out you can just spin up a closer in our in our online environment which allows companies to very quickly get a sense of like hey is this how do we compare is it easy to run is it easy to use and also it's easy to set up a proof of concept I mentioned before that because we're open source it's also easy to run locally uh so that means that it can run on a developer's laptop or on the infrastructure of of the customer itself and we provide all the tools and and tutorials needed for companies in order to to do so so from a deployment perspective is very easy from an Integrations perspective is easy and then the third part to sort of complete this this section is that from the beginning of we8 we decided to build we as a modular system so you can plug and play different components into we8 to connect to certain Integrations to certain what we call model providers so companies like openi and coh here that have built these great models that allow companies to quickly develop and integrate and we had integrates with all of these so we really try to offer the developer but also the different businesses an easy way to get started easy way to get deployed but also an easy way to to maintain the entire setup let's talk a little bit about security because it also goes hand inand with this conversation how do you help your customers stay secure when using we v8's open source database or what do you advise them or tell them they should know before using it that's a very relevant question I think security is one of the topics that does not always get discussed uh but is a very important topic and one of the things that I mentioned before was companies want to keep data closed they want to keep close to heart because it might be used for other purposes there might be confidential information when that that they do not want to share I think in the beginning of chat GPT when they just started a lot of companies or people working at companies were pasting copy pasting confidential information into a large language model which you shouldn't do because the model get trained on this information but it poses a real threat for certain companies especially depending on what type of data they handle as I mentioned before with we being open source companies can run this in their own infrastructure which means it's the most secure way of doing the deployment data never has to leave you never have to send data to a third party so that is already one level of security we do have companies that do not want to run themselves uh because of the operational maintenance Etc and and keeping keeping the service running then they rather have the company that built the software do that which for us is great because that means that that we can take over if companies run in our Cloud as we have our own our own S servers around the database we are fully suck to compliant use encryption both address and in flight uh so we adhere to all the security standards needed for companies to integrate so basically companies can choose whether they want to run themselves they want to run it in their Cloud but they want to have it managed by us which we can do securely or they run in our cloud and then there we have all the proper security measures in place in regards to sending data to the large language model like to ftbt as I mentioned before we8 also has the capability to integrate directly with some of these large language models which you can run also within your own environment so we can do full skill deployment within a sort of isolated architecture that allows for full security and and and close off of the internet for our users Byron we've talked about a customer success story already but I'm curious to know what's your favorite application you've seen we8 applied to that's a that's a that's a good question I think I would go for the first time where I actually saw the true benefits of vector search and this was this was a while ago this is a company that I work with fairly early on and they were doing media Asset Management especially for producers they have to select certain scenes in order to build the end results of the video and a lot of these companies have tons and tons of assets they do retakes of shoots of move for movies or for series and this company we were talking to uh they mentioned that they were trying to build a tool for these producers to be able to find exact scenes shots Depot certain lines that were spoken of dialogue throughout their entire Asset Management library and and this was very early stage this was before Chad GPT this was before we actually knew about all these capability or everyone knows about these capabilities right now and they built a um a piece of software that was able to if you search for I'm looking for the scene where it was raining dark in the background where a woman says I forgot my shoes and they were able to go and search through the entire media AET library and find the exact moment of a snippet where exactly this moment was shared and that for me was the first time where I was like wow we can do scene analysis we can not just find it based on the text but we can we know it's raining we know we know a lot about cont contextual information in the background and for me that was the first time I was like wow this is this is really cool especially with what we know today this this was really eye opening for me that is a wild story I really loved hearing that and thank you so much for explaining weeva to me and our audience today Byron if we're interested in learning more and getting in touch with you and exploring getting started with we8 where can we go there's a bunch of places to go we do a whole lot of Road shows through the US and through Europe uh we try to do as many events in person as we can we have different concepts called Road shows and hack nights that are free to get to they're hosted in all the different big cities in the US we also have our Academy and tutorials online where people can go and learn things at their own pace we have recipes for examples of certain use cas cases so if people know I have a used case in e-commerce we already have a couple pre-made examples on our getup and last as I mentioned before we have a free sendbox sign up page uh so you can sign up get started and also a place where you can reach out to us is our forum and our public slack so there's there's a whole bunch of different places where where you can find us thank you so much for your time today Byron it's been so exciting getting to talk to you and learn about we8 and of course thanks to our audience for joining us today for more episodes of partner POV check out channel insider.com once again I'm Katie boso and I'll see you next time

This transcript was generated automatically from the video's captions and may contain errors.