Episode 9: What is Talla? 

Rob May and Brooke Torres have an exciting announcement from Talla and background on a question we get a lot, "What is Talla?" 

Subscribe to AI at Work on iTunes or Google Play


Rob Circle Headshot

Rob May, CEO, Talla 

Brooke headshot circle

Brooke Torres, Director of Marketing, Talla 





Episode Transcription 

Rob May: Hello everybody and welcome to a special edition of AI at Work Podcast. I'm Rob May, the co-founder and CEO of Talla. We have a pretty big announcement here at Talla this week, the week of August 20.

Since we've been through these podcasts for a while, and we hope most of you know what Talla is, but for those of you who don't, since we have a big announcement this week, we thought we'd do a shortened podcast here and tell you a little bit about us as a company, and what's really unique about our product.

I'll start by talking a little bit about my background and the background of the team here, and the founding of Talla. We're a little over two years old as a company. We've raised $12 million in venture capital. The idea behind Talla when we started was to build digital workers.

The hypothesis is that there is a way to take any work stack and make a sort of complexity chart. For any job, what's easy about that job all the way up to what's difficult about that job. The things at the top, the things that are very difficult, are things that we're decades away from machines being able to do. The things that are at the very bottom, the things that are easier, are things that can be automated today.

We believe there's a whole class of things in the middle that you can automate away with technology to free up humans to focus on the more interesting work. But, you don't have a data set to train the AI models on. Talla is very focused on building workflows and tools that get those data sets, capture those data sets, train those models, and automate those workflows.

Now to build these digital workers, you need three parts. You need an interface, which is chat, text, voice, whatever. You need a knowledge base. The digital workers have to know things about your company-- who's there, what are they working on, what are the products, et cetera. Then they need skills, tasks that they can do, right? Update this record. You know, run this campaign. Perform this specific task, fill up this form, etc. 

We decided to attack this problem from the knowledge side. So what you're going to see today is a new kind of knowledge base, that's what we've launched, that covers part of this digital work revision that we have for the long term.

Like I said, the company is about two years old. I had a previous company before this that some of you that have followed Talla know about. It it's called Backupify. Sold that company in 2014, and had a very successful exit. I'm a hardware engineer by training. What interested me in Talla as a problem space is that I also did a partial master's degree in computer science focused on AI. I didn't finish because this was the early 2000s and AI was just so far away from where it is today.

I really followed the space for the last 15 years conceptually. When I had a chance to start a new company, I realized that that was one of the things that I wanted to work on, because I feel like it's the right time to go do this. So, I thought that would be a lot of fun. Brooke Torres, I'll let her introduce herself. Was one of the founding team members here and ran marketing for a while.

Brooke Torres: Yeah, hi, everyone. I'll also tell you a bit about other key team members here because since we've started, we've had a really strong executive team. So we've got Byron Galbraith. He is co-founder, and he's chief scientist, and he's got a PhD in neural and cognitive computing, super, super brilliant guy. And then we've got Paula Long who is a repeat entrepreneur. She was co-founder of EqualLogic. That was  over a billion dollar exit for then and just a really, really well-known person in the Boston ecosystem where we're based.

RM: Yeah, and after EqualLogic, Paula did some work for a robotics company, and then she also had her own big data software company DataGravity. She has a lot of relevant experience to this space.

BT: Yeah, and all that is really important for entering an early market too, so just a good group.

RM: What we're talking about today, what we've announced this week, is v2 of our product, which we're calling a new kind of knowledge base, and what's new about it? What's new is that we're really merging content and automation and machine learning together. We've been really fascinated by what the team over at Airtable has done. If you haven't seen that product, check it out. It's an amazing product. By merging spreadsheets and automation, we've tried to do that for content.

If you look at what part of the reason that Chatbots suck so much for sales and support is that they're point solutions that are tacked onto databases later or content repositories of some kind. You're really relying on the models to do a lot of the heavy lifting. Going back to my earlier statement about being able to capture data to train models to automate tasks, one of the things that the Talla user interface does is it really vertically integrates a lot of this stuff. When you create the content, you can tag it in a lot of cool ways that make it easy to train machine learning models down the road and make them better.

BT: One of the questions that we get a lot about the product based on where we've been as a company before is, "who really uses this?" It is customer facing teams. So we had a very similar product over the last year or so that we had a good amount of HR teams using. You do see HR teams, especially those who think about their employees as customers, who still use this. We've just seen much better traction in sales, support, success teams, because they rely so much on accurate information that they have instant access to and automating the workflows that are not related to selling or customer conversations.

RM: In particular, we've seen a lot of success with companies that have very complex product spaces that are changing a lot. You can deploy Talla either to your reps or to your end customers or both. Sometimes, even though you can use it just like a traditional knowledge base and people can look up things and find pages, those pages are automatically turned into bot format, either one bot, or multiple bots, whatever you want to do.

A good example is you have a complex product, your sales person is on the phone with a customer. The customer asks if your product supports Linux 2.6. They don't even know what that is. In a traditional knowledge-based solution, they would search for Linux. They would get 14 results. They would have to dig through those on the phone while trying to keep the customer busy to figure out which one was relevant. With Talla, they just get the answer, and that's whether it's your rep or your customers, and we can do that because we have real AI built into the product.

Obviously, with the background that Byron has and the background that I have in this space and the work that we've done, we don't want to tack AI on here as just a marketing thing. We have a true machine comprehension model that is deployed in production. We have a bunch of different types of machine learning models. We wrote our own natural language Understanding Stack for part of this. We have a large data science team for a company of our size, including people who have worked to places like the Allen AI Institute and people from Harvard and MIT, so it's one of the advantages of being in Boston. But the product really does learn with usage and get better over time, and it does so pretty quickly.

BT: I think one of the nice things about Talla, compared to other AI tools, is that there are a lot of AI products on the market, where in order to get started, there's a huge onboarding process. We've talked about this with other guests on the podcast. That sometimes what you do is you have a huge amount of data, and you have to go through it and annotate it to whether or not you can have some success or some insights with that product is just a long way down the road.

But the way that we've built Talla is such that while you can onboard with some of your existing content and a lot of people do go through that. You can also just start day one and work it into your natural workflow, so that you're creating content in the Talla knowledge base about your product, about your processes. You find that pretty quickly it's very helpful as you annotate and as it learns.

RM: So, a lot of people ask what is Talla, and maybe that's what we'll title this podcast episode, but in short, Talla is really a new kind of knowledge base. It merges machine learning and automation with your content. So that when you have sales or support or success reps that need information, they need to ask specific questions, they can get specific answers. Then we can also take it to the next step. We have a lot of automation built into Talla so that if the answer that is returned has an action associated with it, in many cases, we can integrate with the systems and take that action as well. So really save you a tremendous amount of time.

We've seen pretty amazing results with a lot of the early customers and the places we've been deployed. So our goal with this podcast is really to be a resource for listeners, for people that want to keep trends of AI at work top of mind and figure out what's hot and what's not and what they should be thinking about. And so we try not to be overly promotional about Talla. That's not the point of this podcast but decided that since we did have a big announcement this week, and we do get questions from podcast listeners and many people about Talla, we thought we'd take the chance to explain a little bit more about us, so that we don't have to do it on every podcast.

If it's something that you're interested in, if you have a sales team or a support team or a customer success team, and they're looking for a new knowledge base or they're generally looking for bots or ways to become much, much, much more productive and integrate AI into your company, then we hope you give us a chance and check out a demo. We'd love to show you something like that. And then also, if you're new listener, we hope you'll go back and listen to some of our old podcasts. We've had some great guests, and as always, send us questions or content ideas, people you'd like to see on or topics you'd like us to talk about. You can send those to podcast@Talla.com, and that's it for today, and we'll see you in our next episode

Subscribe to AI at Work on iTunes or Google Play and share with your network! If you have feedback or questions, we'd love to hear from you at podcast@talla.com or tweet at us @talllainc.