Tammy Ven Dange of Roundbox Consulting chats with Rob Barnes, CEO of Betty Bot about their innovative use of AI to leverage and accelerate access to Association information and content.
In this interview, we learn about:
- 00:15 – About Betty Bot
- 02:08 – When was Betty Bot started?
- 02:50 – About Rob Barnes
- 05:33 – Betty Bot functionalities
- 10:05 – Overcoming challenges with AI
- 13:17 – Governance and version control with Betty Bot
- 15:35 – Betty Bot clients
- 16:44 – Betty Bot for smaller Associations
- 21:28 – Further information about Betty Bot
Links and Resources:
- More info about Betty Bot – https://bettybot.ai
Other Vendor Videos:
Our Vendor Video Library is growing every month. If you liked this video, you might find these other vendor videos useful too:
Tammy regularly helps Not for Profits make IT investment decisions. Let her know if you need some help.
Tammy Ven Dange is a former charity CEO, Association President, Not for Profit Board Member and IT Executive. Today she helps NFPs with strategic IT decisions as an independent consultant. She does not take commissions nor sign partnership arrangements with vendors.
Video Transcript with Betty Bot (Minor modifications have been made for clarity)
Tammy Ven Dange – Today I welcome Rob Barnes, CEO of Betty Bot. Rob, thanks for joining me today.
Rob Barnes – Tammy, great to see you, thanks for having me.
About Betty Bot
Tammy Ven Dange – Well, tell us more about Betty Bot.
Rob Barnes – You know, having worked in Associations for an inordinately long time now. One of the things that I think we all agree is that Associations are this treasure trove of information.
They sit on, in some cases, some of the largest knowledge bases in their profession or their industry in the entire world. They’ve invested incredible amounts of time, money, energy, in developing these resource bases, but they’re really challenging to get access to when you need them most.
And so whether it’s member-only or it’s a not-for-profit but a lot of information that they’re trying to get out to the public, it’s a real challenge to do that in an efficient way and meet the member or meet the person’s knowledge need at that point of knowledge need, right? Because you can never know when that’s going to be.
So Betty Bot is an AI tool that was developed specifically to solve that for Associations.
My co-founders and I, our backgrounds have been working with Associations and my co-founders are the technologists that have been working with GPT models for four or five years now, and almost stumbled upon this use of a particular piece of AI that they were working on, on the GPT platform and then realised its application in the real world.
So now we have this opportunity to use the development of AI to unlock all those latent knowledge assets that live in all these organisations around the world and make them accessible to create new knowledge, but also really to serve the knowledge needs of members all over the place.
When was Betty Bot started?
Tammy Ven Dange – Well, AI’s been around for a little bit, but ChatGPT made it famous probably this year, early 2023. How long has Betty Bot been around?
Rob Barnes – So, Betty Bot’s first iteration came online in about February of 2023. As I said, prior to that, Thomas Altman and Dray McFarlane, who are the architects of Betty Bot, had been working on language models, AI tools, and then the GPT platform before it was publicly available for about three years prior to that.
Then , when OpenAI decided to release GPT to the public, that gave us the opportunity to bring the Betty Bot product to life. And her first contact with a customer was in around May of this year.
About Rob Barnes
Tammy Ven Dange – And so how did you actually get involved in Betty Bot?
I guess it’s a really interesting story for some. I also feel like it’s the culmination of about 10 years worth of networking and relationships for myself.
I worked for Associations for a really long time and then in 2013 I met a guy named Amith Nagarajan, and currently Amith is the chairman of Blue Cypress. And Blue Cypress is this family of companies that’s developing solutions specifically for the Association space.
So previously, he had a company called Aptify, an AMS product. And he reached out to me back in the day and said would I be interested in opening Aptify in Australasia and running their offices.
Now, at that point I hadn’t worked outside of an actual Association itself, so it was a bit of a leap of faith, I guess, to do that. But we did that, we built up the Aptify team, brand, and customer base in Australia and New Zealand 2013 up until about 2018.
It was that relationship and that sort of grounding that brought me back into the fold. So when I finished working with Higher Logic in June, July of 2023, I reached out to Amith as one of my sort of trusted advisors and said, “Hey, I’m just about to go on vacation for a while, “but I’m in the market again. If you’ve got any great ideas, let me know.”
And unbeknownst to me at the time, Amith and Thomas and Dray had already been talking about me as a potential co-founder and CEO for Betty Bott prior to that call.
So when all of a sudden I had time available, they jumped on the opportunity, and I started doing some research into Betty Bott as a product. And it was one of those moments where you kind of realise what you’re sitting in front of and when I saw what she was capable of doing for some of the customers that were, I automatically went back and went, “I wish I had access to this when I worked for Fitness Australia,” my Association for a long, long time.
But I saw that real-world application and felt really confident that I could front that opportunity to the market. And so, yeah, came to the States.
So I’ve been here in the US for nearly three months now and we’ve been building out our customer base and building out the brand and building the company itself for that matter. And I’ve signed on there to be a CEO and co-founder for Betty Bot for the next few years at least.
Betty Bot functionalities
Tammy Ven Dange – Well, tell us more, a little bit more about Betty Bot in terms of its actual functionalities. I know you said what it was trying to solve, but as far as the actual functionalities, if you could share that, that would be great.
Rob Barnes – Yeah, when we first talked to an Association, it’s part of the exploration is, “So where does all of this valuable content live?” And the kinds of challenges that they share with us are, “We have a new website,” or, “We’re developing a new website because nobody can find anything on our website.”
That’s a pretty common challenge that Association staff face. Or, “We have a decade’s worth of videos that we’ve recorded. We have our annual conference sessions recorded and they’re just sitting in a repository. We have journals, we have three journals, we have a magazine.”
So this litany of resources sitting somewhere in the organisation, whether it’s accessible via the web or whether it’s not.
What Betty is designed to do is to ingest all of that content. And literally, I mean all of it. She can absorb every article of every journal that’s ever been written, every video transcript, every PDF document, every policy document. Everything that is either currently in a text form or can be converted to text really simply, she can ingest all of that content.
Then we give her a face and we give her a user experience. And by connecting with her, you can ask any natural language question that a member would typically ask to interrogate that body of knowledge.
What she’s capable of doing in having that natural language conversation with a user is stopping for a moment and just thinking about the context that we have given her, this constitution.
We’ve called her Betty, but our customers can call her whatever they want. And typically they’re using something that’s meaningful to them as an organisation or a profession.
But we can also create her constitution, her personality on their behalf. So she’s a true representation of their brand, their tone of voice, the type of users that they have.
Academic customers, like a little bit more of a buttoned-up Betty, someone that’s a little bit more formal in tone. Others would like her to be a little bit more colloquial, friendly, empathetic, whatever it might be.
That’s one of the wonderful things about the AI tools available now is you can create this constitution and you can create that context. And so given that, you can have a conversation and say, “I’m an Association executive. I’ve been in the membership space for a long time. “I’d love to be a director “of membership within three years.”
And if she was trained on content, like she is for instance with the American Society of Association Executives, you can ask Betty to use her knowledge about everything that she has been trained on through ASAE and she will have that conversation with me.
Now, not only can she recommend where the content is, so solving that problem about what, website managers don’t have to worry about where the content sits on a menu or a dropdown anymore because she already knows where it is.
But she’s also read it and she’s memorised it and she can have this conversation using all of that in real time.
So whether it’s a video transcript or it’s a podcast or it’s a session outline or an education course or a journal article, she can use all of that knowledge to have this conversation. And then she references exactly where she got her knowledge from in the conversation as well.
She can prove that what she’s talking to you about comes from fact and these are the things that are owned by the Association.
It’s pretty neat to see in practise. I don’t think I’ve seen so many people say on a demo, “Oh my, are you for real?” I haven’t heard that before and I’ve done a lot of demos in my time.
Overcoming challenges with AI
Tammy Ven Dange – I’m actually really fascinated by it because I know that there are a couple of questions and challenges with the currently open model-type AI tools.
One is that the information it’s trained on is generally proprietary or open source depending on what it is. It sounds like you’ve been able to turn the large language model to be the actual language is actually coming from the Association itself.
But the other part that has also been a major issue with AI tools is accuracy. They talk about hallucinations. So how have you guys managed to get past that? Because that has been a major issue for some tools so far.
Rob Barnes – Well the simple answer to that is that she’s not trained on anything that she shouldn’t know about. And she’s not trained on anything that isn’t content that’s already owned by the Association. So again, the context with which, the knowledge she’s using to have this conversation is only that content that’s coming.
Hallucinations are caused by the fact that the source material is the public internet and an amalgam of all of that. And so eventually it’s just gonna be using random facts and random pieces of information to construct. Because what it’s trying to do is simply construct the next series of words that it’s going to use.
But when it’s doing that only on content that all is relative to the particular topic, and has this constitution also trained, this context also given, which is about that profession or that industry, the hallucination goes away.
Governance and version control with Betty Bot
Tammy Ven Dange – And what about structured versus non-structured data? I know that’s another problem, even with Microsoft Copilot, that they’re not allowing that AI tool to be available for organisations that don’t have really well, strong data management practises.
How do you get past that too? Because that’s a common issue in most organisations is the data architecture is not strong enough for these types of tools.
Rob Barnes – Yeah, I think because we are primarily using content, right? We’re primarily using source documentation that is text-based. Actually having to interrogate a database as such is not typically, that we’ve seen so far, the primary use case. And it’s not why we’ve built Betty.
There’s actually another tool within the Blue Cypress family, if you like, called Skip. And Skip is similar technology built on it from a similar premise for the purposes of interrogating an actual database.
Again, starting to solve that same problem where data is very messy and using the AI to do that. So we’re focusing Betty primarily on that, creating this user interface to the huge knowledge base of content that an Association lives on.
And again, because anything can be converted to text, once it’s in text and we’re working with our customers to build out, I guess the structure at that point, that means that she is responding in a way that is meaningful to the end user.
That’s actually not that difficult to do. And actually, as AI-to-AI tools continue to grow, we’ve recently built a team of AI assistants for Betty Bot that is helping pass the data, pass the content down into the manageable chunks that she likes to use best. And so we have a team of AI tools working directly for Betty so that that process is a lot easier.
Tammy Ven Dange – What about version control?
So as we know, most professional organisations evolve over time and they might have a significant amount of data, but there could be legislation changes that impact how they would govern or provide advice to their own members.
How does Betty Bot deal with that?
Rob Barnes – Yeah, so there’s a couple of ways that she can be, I guess triggered, if you like, so that her training continues over time.
If she’s tracking, if part of her knowledge sources are webpages and the webpages get updated, then the next time we run the sort of the cycle of her tracking those particular ways, she’ll pick up the new content and she’ll automatically be updated and that will become part of the knowledge.
If it’s a document, and standards documents are the ones that are kind of top of mind for us. We’re working with a lot of heavily technical specifications documents in the medical and construction industries for instance. There’s a couple of ways that that will work.
So our customers can get to choose. In some cases, they don’t want the 2017 version of a specification document still to be part of the knowledge base when the 2023 version is published.
We remove the 2017 version, or in fact, our customer will be able to remove that from her knowledge base. She’ll no longer reference that. It will be the 2023 version. And that’s a pretty simple update, just via admin UI.
In other cases though, the historical documentation is equally important to the knowledge needs of the member because they might want to compare and contrast. “Hey, tell me what the difference is between the 2023 specification and the 2017 specification as it relates to concrete standards,” whatever it might be.
But someone being able to be knowledgeable about both of those and do the compare and contrast and then help dig deeper into that is exactly why Betty was built.
So from a knowledge base, she can have all of the knowledge to have those conversations where comparisons are required or some sort of reporting. “Tell me a little bit about this, compare it to that.” In other cases where it’s only the latest is what the Association wants their members to be able to access, then her knowledge base is just constantly updated.
So she’s only referencing that information where she’s having the chats.
Betty Bot clients
Tammy Ven Dange – Alright, so who are some of your customers that you have? I know it’s a new company, but it sounds like you’re already working with clients.
Rob Barnes – There’s about a dozen customers that are actively in place right now. I think one of the more exciting ones was one that just launched here in the US just the other day,
The Marine Retailers Association of the Americas. And we’d been working with them for a couple of weeks leading up to their major event of the year called Dealer Week. And so huge trade show, boat manufacturers, all setting themselves up, and in fact cleverly, the Association team there built a booth just for their version of Betty.
Their version of Betty is named AIMIE, which is AI for the Marine Industry Education. So AIMIE had her own booth. She was on screen and members and other participants at the Dealer Week trade show were able to come and sit and actually ask questions and have conversations.
She’s been trained on a whole range of marine industry education information and knowledge about the industry broadly, which is a lot of it has got to do with market data so that boat dealers understand where there is boats, where there isn’t boats, all these sorts of things that that particular Association has become expert at collecting that information and wanted to make it much more readily available.
MRAA has been exciting to watch the evolution and exciting to see the launch particularly so cleverly, I think, at a trade show.
Others are organisations where particularly the documentation is, as I said, it’s heavily technical or it’s in the legal profession or the medical profession.
So what’s common to both of those types of Associations is they are mindful and very focused on ensuring that any AI connection to that information yields an accurate result.
One of the reasons that they chose Betty Bot to do that is because she’s able to prove where she is getting access to this information, where she’s answering that question from.
They also like the fact that their version of Betty can be, as I said before, a little bit more buttoned up. So her personality is designed to be, like who you are talking mostly to, in the case of the American Health Law Organisation, you’re talking to lawyers in a health setting that is one of the most regulated industries in the world.
“So please be formal, be super buttoned up. There’s no room for interpretation here. Please just repeat what you know,” et cetera, et cetera.
So being able to guide her has given those organisations a lot more confidence about AI, I think more broadly, our AI tool specifically. Because when they do a beta test, with a small cohort of their members who come in to try to break her, ask her really, really heavily technical questions.
They are super impressed not only with what she knows, where she’s getting this information from, but the way she is articulating it.
And I think that that user experience is what I think everyone is broadly now going to start to accept as being the norm when it comes to interacting with AI.
But certainly for Associations, they want to be able to protect their intellectual property, they want to be able to deliver access to it in a meaningful way just for the people that should have access to it.
And they want to make sure that it’s right, it’s giving the right information at the right time when that person needs it. And so far, so good.
Betty Bot for smaller Associations?
Tammy Ven Dange – Well, it sounds like some really complex use cases. Do you see Betty Bot working in smaller Associations?
Rob Barnes – Yeah, we’ve certainly scaled the commercial model, if you like, to make her available to any organisation that’s sitting on content that they would love to make available.
So, the way the subscription works is it’s scaled based on the size of the organisation from their annual revenue perspective.
That way, and why did we do that? Because when I was with my Association, a couple of them and I would want to build a business case for my board or for my CEO and say, “Hey, I think we should make this investment.”
I would always want to go into that discussion with, “Here’s a model. If we spend this, I believe this can happen or it’s going to unlock this value. This might be how much revenue we earn. But based on how much revenue we already earn, this is how much it’s going to cost.”
These are pretty simple calculators to kind of build, but they’re very powerful in building those business cases for change that boards and CEOs want to see.
So, we’ve built our subscription model to suit so that anyone can look at the price that they would invest in Betty Bot and I guess articulate that against their annual revenue, articulate that against their membership fees, this sort of thing, and then say, “Well, if we’re a $2 million-a-year organisation, or a $1 million-a-year organisation. And Betty Bot costs us 2000 bucks a month, or 1,000 bucks a month, or whatever it is for them, it’s cents on the dollar when you boil it down to a per-member, per-month fee.
In some cases it’s like 22, 23 cents per member per month to make that investment.
S,o they would only need to generate a little bit of extra revenue to cover the cost of that investment. And I think that that for me is, I never want the solution to be out of reach to an organisation just because you’re trying to make a lot of money out of it, and that’s not the purpose of doing this kind of work for us.
For us, it’s that we know they’ve got all this information. The smaller organisations are often the ones that struggle more to grow and to build the opportunity, but they’ve got this great content that they’re sitting on that they could leverage. And that’s what we’re really excited about being able to do.
Further information about Betty Bot
Tammy Ven Dange – Well, is there anything else you wanted to talk about, Rob?
Rob Barnes – Yeah, come and visit us at bettybot.ai. I think the other thing I’d love to share about it is that… It’s exciting building a new company from scratch, right?
Particularly on the back of this kind of experience where we’re our Association-first. We’ve come from the Association sort of industry, we get it.
But we’re also building Betty for Betty. And what I mean is that a lot of people have lots of questions about Betty Bot. So we’ve trained a version of Betty to know about herself. So things like the story of, well, why Betty?
So Betty is named Betty because she’s named after a woman by the name of Betty Holberton. Now Betty Holberton was a computer scientist back in the thirties and forties, unheralded for the amount of foundational computer science work that she did.
She’s one of these kind of hidden figures. So building Betty and naming Betty Bot the way we have is our hat tip to her. That’s a story that we would love told more and more.
People ask us this question, people ask us questions, “Well, how much does Betty Bot cost?” “How long does it take to implement Betty Bot?”
So as we’ve been doing about a dozen or so demos a week for the last few months, obviously we have the transcripts for this information.
We anonymize it and we’ve trained our own version of Betty on all of that information that people are asking us, the answers that we’ve given.
And you can go to our website and you can actually launch a version of Betty and you can ask her all about working with Betty Bot.
So I feel like it’s our drinking our own champagne kind of moment that if we’re going to try and sell a tool to Associations and say she can do certain things, she needs to be able to do those things for us as well. She’s live on our website right now. You can test her out.
Tammy Ven Dange – Rob, I’m like totally excited.
I could see so many use cases for Associations and other not-for-profits even in Betty Bot. And I know you’re focused on Associations, but I can’t help but think beyond that for what I do.
It’s great to talk to you again. I know we met originally in Australia, but it’s great to see you in the US now and good luck with Betty Bot and thank you for sharing.
Rob Barnes – Thank you so much for having me. Yes, it’s great to be in contact with you again. It feels like we’ve swapped sides of the world somewhat. I’m looking forward to being home for a little while. But again, great chat. I appreciate all the work that you do. Looking forward to continuing to stay in contact. Thanks for having me.
Tammy Ven Dange – Thanks, Rob.