Appian Corporation (APPN) Earnings Call Transcript & Summary

Hey, welcome to our next session.

Thank you, [ Raimo ].

Actually, this is an easy one, Malcolm. Maybe just talk a little bit about yourselves and your role at Appian and then we can take it from there, actually.

Yes. So, Malcolm Ross. I'm the Senior Vice President of Product Strategy for Appian. I lead our go-to-market, working on a product roadmap with our engineering and product management teams, and overall the direction.

And Sri, you're kind of protecting Malcolm on all the financial questions?

Exactly.

Malcolm, like you've been at Appian for a while?

19 years.

19 years? Yes, that's quite a period. Can you talk a little bit about the evolution of the company in terms of like what you set out to solve 19 years ago when you joined versus like the much, much broader offering that you have like today?

Yes. I mean, I was there at a really instantiation of the company, when we're starting to become a software company, and we set out to really become -- at the time, almost 2 decades ago, business process management. We had a vision of helping organizations unify their business operations or in workflows. At the time, it was very much workflow driven, but we recognize our customers need software that will be easier to use, faster to get value out of. So, that evolved in Appian very much focusing on low-code in the 2010s. And we saw that manifest in the low-code market around 2015. Appian is very well regarded in the low-code market even today. And as we saw, the process automation market overall is kind of nebulous piece of things. So, business process management is in there, low-code is in there. We saw the RPA days in the 2015 to 2020s or so, where RPA became very hot. Appian made moves in that market as well. So about 3 years ago, I worked it on myself, the acquisition of the RPA product that's now part of the Appian stack and now we're in AI. But at the end of the day, what we're trying to focus on is helping our customers achieve end-to-end process automation. How do I help major bank, government agency, ingest the case -- ingest the customer service requests, take that all the way to resolution. And when it comes to orchestrating human interactions of how customer service [ operator handlers ] work on that to automating the real legacy systems and unifying that so it can manage that process end-to-end and drive efficiency over time as well.

And where are customers on that understanding of like a broader offering and that it's kind of probably better done out of one starting point?

Yes. I mean we're, I think, past the RPA phase, I would say. And customers are definitely more focused on holistic automation now. RPA is a technology that's here to stay. It does something very unique. It allows you to automate interactions with client desktop applications and legacy systems, and that's still a new [ component ]. But everyone's dipped their toe into that. It's now part of the core tool set for automation and customer focus is now on what Gartner would call hyper automation, but it's a holistic automation and how do I achieve those outcomes in a holistic fashion, using RPA, using AI, using rules to achieve my overall business outcomes. So, Appian I think is very well positioned. We made those investments in RPA and focus on that end-to-end process within the company.

Yes, okay. And then, so we should think about it, over the years, we saw a lot of like scope creep in your space where it used to be like you guys and Pega were kind of the local guys and then you had like the UiPath and Blue Prism et cetera, you know RPAs. Now over the last few years, you have like everyone is trying to do everything a little bit and then you have like the Microsoft's coming in with Power Automate et cetera. Like how do you see this market playing out in the long run. And then I have a follow-up? Sorry.

Sure. Yes. I mean, consolidation is what we see. Even UiPath is positioning in the in-process automation, although they're mostly regarded as RPA still. The Gartner reports, Forrester reports that focus on this the most. You saw recently the [ fourth ] Digital Process Automation Wave report that came out, so they really cover that in that perspective. Appian is well regarded there. [indiscernible] covers it as an aspect of low-code of business workflow automation use cases. But what we see is ServiceNow has moved into RPA, UiPath is trying to move out of RPA into in process automation. Celonis is trying to have a greater value proposition in their process automation as well. Microsoft has the Power platform, Power Automate, that's kind of other tools. I think Appian has always been in this space of business process management of holistic views, made those investments in AI, RPA, to have automation technology in place to really capture that market share in the -- here as well.

And then there's like -- there's competition, there's [ peak ] competition, there's like label competition, but it's like then real competition. Like how do you see that out in real life? Because I can -- if I'm talking to practitioners, it's like, yes, look, if I do something around my ServiceNow platform, I probably use their platform. But like if I do solve some proper real life examples [indiscernible] Is that kind of how -- kind of how you see it playing out or like how would you think about that going forward?

Yes, competitively we see a lot of established vendors, so Microsoft, Salesforce, ServiceNow customers already have those investments and they say, "Well, you want to do case service management automation" sounds like maybe I can do that with ServiceNow. So they try to. But, with Appian, our value proposition is quite different, I would say. With Salesforce, ServiceNow, for example, typically the value proposition is, you take your data, you take your systems and the logic, you put it into ServiceNow, you build applications around it. Appian's value proposition is really around, you have Workday, you have NetSuite, you have ServiceNow and you have these technologies, you need to unify those; stitch those together, a new composite and digital experience with customer service center for contracts management, whatever it is. So, the core value proposition is actually quite different as far as a unification platform that says, keep your assets, but unify them in complete digital experiences. That said, as well, like ServiceNow, Salesforce, we compete against them a lot as their established vendors. We never see them as, say, there's a greenfield opportunity for both vendors. Like they don't have ServiceNow and they don't have Appian, they're considering both tools, that never happens. It's usually Appian is well regarded in that holistic business process management area and they regard us more against, I'd say, [indiscernible] as the primary competition in that space.

Yes. And then like briefly, you mentioned AI a little bit. Like can you just talk a little bit about how that plays into your space?

Yes. AI is, well, aside from being on -- the top of [indiscernible] of '23, it's the legitimate investment we have automation that every organization is making right now. And we're very much on top of investing in that ourselves. So in the early this 2 year, we announced our release of our AI skills Designer, which is really a low-code design experience, which allows our customers to not have a data scientist, but train machine learning models for AI skills that drive automation inside their business, specifically ingestion of e-mails and automatic routing, in document classification, document extraction and we have a roadmap of building this out further and further. But it's all about low-code relying AI to make it approachable and easy to start applying that inside your business applications. Furthermore...

Sorry to interrupt, is it a little bit like a copilot? Like Microsoft calls it Copilot, the next one calls it Autopilot, but is like [Technical Difficulty].

We bucket our AI skills in 3 [Technical Difficulty] specifically building new AI machine learning models in a low code design experience that a customer employs to drive AI automation in their enterprise. We also have AI copilot in Appian, which is typically based on LLM that is an assistive feature for building applications, guiding a user to a certain path. That's another pervasive feature throughout the Appian product. And then finally, it's all built on an architectural promise, what we call private AI, which is to ensure that when you're building these AI services, we respect the privacy, data retention and to make sure that [indiscernible] models also remain under control of our customers in that regard.

Like where are you on that AI journey? Like, obviously, like non-generative AI has been around for a while and people were doing anomaly detection, et cetera, et cetera. So, where are you on that journey and how has that accelerated in the last quarter?

Yes. We started working with OpenAI and the LLMs first half of 2023 as we all saw them come out, it's harder to offer an integration [Technical Difficulty] right away. Then we started embedding them and [indiscernible] Embedded features in August this year, where we have LLM specific -- initially OpenAI, that is an assistant feature that helps build application experiences. And just November 10 is where our last release was, just a couple of weeks ago. We released the integration of the LLM with our data fabric technology, which is really interesting use cases. So first, we're moving towards more of an embedded LLM strategy, which says instead of integrating the LLMs, we're just going to offer it as part of the Appian stack and that complete AI privacy. And then second, we're going to tightly integrate it with our data technology to allow you to have natural language conversations with their data set and it's really trying to solve the hallucination challenge with LLMs, which is, if I want to -- like one of the use cases we often focus on is imagine I build a composite application using Appian that unifies data from SAP on order history and Salesforce customer data and worker information in Workday. Then I say, well, write an e-mail to my customer about their order history over the past 3 months and then tell them what product opportunities they can take advantage of in the next quarter, common thing every account executive does. Well, by having the LLM understand the semantic question there, parse it out to the context of the data fabric, retrieve the data from the data fabric so we reliably report, what is the actual order history, not going to hallucinate and make it up. And then, compose that e-mail, so that it has confident information, but structured in that semantic e-mail that can be sent out. Really cool stuff coming out that we introduced already on our November 10 release that we're continuing to build on that combines LLMs with competent data from the data fabric at Appian.

And do you need to have vector database in there then as well to kind of do that?

Yes.

Okay, yes.

Yes. We definitely -- semantic vectored searching capabilities is a very important part of it as we overlay that on the data fabric to correlate natural language semantics to specific table names and specific kind of information inside the data fabric.

And do you have to develop it yourself or are you partnering with someone there?

We're partnering on the LLM side. So, specifically, we've been working with Microsoft Azure on the OpenAI in the past. We're now more aggressively working with BedRock. And we want to make sure we also have that portability of LLM so that we have -- we can swap out the LLM if necessary. BedRock is the one we're banking on right now as far as keeping that [Technical Difficulty]

Yes. Okay, makes sense. And then, last question, and then I don't know if Sri, he wants to kind of jump in. How do you think about that monetization or how you charge for that?

I think we're looking at multiple ways. There are different aspects. What we talked about is we're going to have a premium pricing for the [indiscernible] that would just involve having a separate SKU or having some element of usage-based pricing. But that is something that will come out later next year.

Yes, yes, yes. Okay. And then so if you think about it, one of the things that happened for you guys is you were more on a cloud journey, like you have more and more of your customers working in Appian cloud. Does that impact like in a way, like your choices there in terms of which hyperscalers to go with, et cetera, with all the AI coming up now? Does that kind of create like a new discussion, given like how strong Microsoft with OpenAI is at the moment?

We've always had the strongest relationship with AWS. So we host our Cloud in AWS primarily. So we do have a preference towards AWS, but we always want to integrate with AI services that are [ free ]. So we've historically had business partnerships with Google on document understanding. We've had integrations with OpenAI. So, at the end of the day, we want to be a platform that focuses on process orchestration and recognizing our customers. We'll always have a heterogeneous mix of cloud scalers or enterprise systems that we want to integrate in cloud.

Yes. But you're more an AWS person?

Yes, from an infrastructure perspective, primarily, yes, sure.

Yes, yes, yes. Okay, makes sense. And then last question -- no not the last question. Shifting gears a little bit. Like, I think historically you guys have been relatively heavy on the federal government side of the world. Does that kind of create like different challenges around AI, given like the greater need for security understanding stuff? Like how should I think about that?

Well, that's exactly why we're focused on bringing the LLM and the AI capabilities inside the firewall environment of customer's cloud so that we can give that confidence that -- it's interesting with AI, you don't need to just protect your data anymore. You also need to protect your machine learning clouds because they have an awareness of your data set. So we want to make sure our customers fully control both the data and the machine learning model and the scope of their Appian cloud environment. And then by embedding inside, when we introduce AI services, they can inherit our existing FedRAMP certification and then an IL-5 certifications that we have for our customers to have that data confidence.

Yes, yes. And then just will the AI -- like if you think about it, how big are those LLMs in terms of -- if you're in need, like can the customers still do it themselves in their data centers or do you really need to be in a hyperscale to do that? A follow-on question would be, does that drive more Appian Cloud actually in the future, because like you need to provide a lot of extra.

So, to answer the first question. It's just -- it's easier for adoption by embedding the LLMs in that because they could bring together different components themselves. But there's a lot of technical complexity in that. So overall, on the value proposition, Appian is simplifying that through our low-code experience and having all right there to rapidly gain value. It does introduce additional compute costs, things like this, especially as we do machine learning model training, GPU time, things like this. And we have premium pricing and other SKUs that we offer to capture that value as they upgrade their services using more the Appian services.

Yes, yes. So -- but could I do it myself? Could it be at home? Or does it need to be like with AWS or Azure if I do it?

Well, in spaces like machine learning?

Yes, yes.

Well you can do whatever you want with AWS, Azure and simply integrate it as a integration point to your Appian process. But, for the stuff that we rolled inside the Appian platform, it's all kind of white labeled wrapped inside the stack. So they don't even -- when they train a machine learning model for document extraction Appian, they're not aware that we might be using different services from Amazon behind the scenes, so it's all simple experience that user who's not a data scientist, companies are using and adding value to their business.

And then last question on AI, since you're a product person, I hope I'm not kind of going too far. Like, if you think about AI, more natural language, et cetera, what does it do to your AI [indiscernible] how you're thinking about your user interface and how [indiscernible] Appian solution going forward?

Yes, it's trending -- we, I would say, over the past decade, people have had to have AI chatbot for a while. And what the modern LLNs is really making those actual value we'll say. We've always had frustrating experiences, right, with dumb chatbots. Now they're actually very intelligent chatbots. So in the most recent release, we introduced something called our Data Fabric feature, which starts to introduce general business intelligence capabilities on the data that you bring into Appian for our customers to do data recovery and has the general BI capabilities to point and click configure a pie chart. But we also introduced what we call our AI copilot for it, which allows you to simply have a natural language conversation data set. So I can say, "Hey, from this report, can you give me a summary of expected orders I need to share with boss?" You can just have very free form conversations and it's able to give you very reliable responses back, which have context that you're looking at. So it does start to evolve the interface to be much more natural language based rather than more perspective UI designs, and we're leading into that in some of the latest features you introduced.

And then truly last question, where are customers on that AI journey? Like, I mean, obviously, there's a lot of like, "Oh we need to do AI", but like then doing it in practical terms is always like a different story. Like what are you seeing in customer conversations in terms of that AI understanding and AI adoption as well?

Well, it's about finding those use cases. And I think a lot of customers are still learning about what's possible and there's some really cool things possible now. For example, like we have a government acquisition management suite. Really, if you ever written a clause for government procurement, it's a boring job. People don't really want to do, writing 100-page government procurement docs. AI is actually really good at combining the data sets of government procurement contracts to do automatic clause writing and then automating that. So we're seeing lots of interest from customers like State of Texas investing in this area because it drives real productivity gains in just a procurement contract in government agencies. So, people are really exploring the boundaries of this, investing in kind of prototype and where that area is, while also simultaneously being very cognizant of the data privacy issues, especially things like in the European Union, the EU AI Act, which is proposed right now [indiscernible] very much aware of watching what's going on.

And then shifting gear a little bit. Historically, like I guess, with your heritage you were kind of very heavy on professional services and did a lot of the stuff yourself. The last [indiscernible] bigger partner [Technical Difficulty] partners. Like where are we on that journey of like what are you doing yourself? Are you kind of heading out to partners?

Do you want to take that?

I think we're still early in the journey. What we have talked about in the past few years, just look at this mix. At the time of the IPO, it was around 50% on services mix. In the most recent quarter, it's 25%. We expect the professional services mix to continue to decline as a percentage of total revenue. It's partly because; A, our subscription revenue growth is much faster than professional, but also we're leaning more on our partners to deliver some of those services. Having said that, we've always maintained that professional services will continue to be a strategic offering for Appian. What we have inherently found is when we look at our customer base, is that when that first implementation goes very well, the upsell that happens, it happens very quickly and at very large amounts. So there is an incentive just not only for Appian, but even for the partners for that first implementation to go very well. And so the instances, our professional consultants are embedded with their partners. And these are not even billable folks, but we have a common goal for that first implementation to go.

Yes, yes. And then, what are you seeing on the partner side? Like given the times we're in, like in terms of building Appian practice, kind of expanding capabilities and capacity there. Like where are we on that journey today.

I think we have a very strong interest not only from the global SIs, but also from some regional partners. We've talked about at the last Investor Day with a couple of partners have established some very aggressive hiring goals in terms of having Appian practitioners at the prospective front. And we still think [Technical Difficulty] and we've recently talked about building dedicated partnerships with some of the global SIs where they have specific incentives in terms of how many new logos they're going to add in any given year. So, we -- both of us have some common goals. But clearly, given our product expansion, platform expansion in the past few years, a lot of partners are more excited to work with us.

Yes, yes, yes. Okay. Yes, it makes sense. And then I'm staying with you Sri for a little bit longer. Think about guidance last quarter that you guys gave, there was an extra layer of conservatism because we didn't know what the government was going to do or what's going to happen there. Like, now it looks like that decision was pushed into next year. Does that give you an extra level of comfort or how do you think about that?

Yes, in general, like we've always maintained some elements of conservatism with respect to the guidance. And clearly, there were a number of factors that were going into 4Q, not only the macro, the geopolitical, but also we had the additional overhang of potential federal government shutdown. So we did take that into our guidance. But we generally feel good about where we are.

Yes, yes, yes. No, fair enough. And then Malcolm the -- it must be nice to have a boss like Matt there, because I remember like last year when everyone was thinking about like reducing headcount, et cetera, he kind of made the -- at the time, somewhat controversial decision to say like, look, there's so much good engineering talent coming on to the market, we kind of want to invest. Like, if you think about your product R&D organization, like what did you see there over the last year in terms of like funding capabilities, especially around [indiscernible].

We're expanding aggressively, had good growth in our Chennai office, which has more engineering capacity. But we believe very strongly in the value proposition for the market moving towards a hyper automation, end-to-end automation, making sure we have the most attractive product in the market to satisfy coming demand. I think it's been well represented also again by the Forrester and Gartner reports that just came out in the past month. Appian has been the most highly regarded vendor for business workflow automation and digital process automation for that market.

Yes. And then how tough is it at the moment like you're doing a lot of work around AI. Like what's the skill availability situation there?

Yes, AI skills, we've been investing in that specific area before 2023 -- so it's not new. We had established engineering capacity already building [indiscernible] the talent [ Azure ] of 2023 is actually more prompt engineering as everyone is worth learning how to interact with LLMs effectively, do fine-tuning of LLMs, prompt engineering of LLMs and we continue to hire on that. And again, Chennai has been a good focus of ours to build up additional engineering capacity for those specific skill sets as well.

Yes. If you're -- like if you think about -- as you mentioned, Chennai, is there -- do you -- like if you think about the classic AI, AI guys, they seem like crazy brainy guys that start working in Google or whatever is in the valley. Like if you do prompt, is that like from a skill level that is needed? Is that a different level than just the guys that [ formed ] the LLM?

Our Prompt Engineering is just one part of it. Fine tuning is certainly more the classical AI data scientists as well, engineering models with the extra context and awareness. But, it's like prompt engineering is more approachable to a non-data scientist to learn how to manage what they call tokens as far as the token capacity of LLMs. One thing like the -- it's more of an architectural conversation. For example, features that we worked on is, if you want to have a conversation with an AI system across a very, very broad [Technical Difficulty] LLMs can only really digest a few pages of a document at a time to give you contextual conversations about. They combine that with an architectural approach of traditional semantic AI search to refine the specific categories and then have that refinement in more context assessment area. So this is a design architecture called Rag Retrieval augmented generative capabilities. So, it's not as much around classic data scientists about architectural design about how we combine AI services together to get the outcome that customer want.

Yes, yes, yes. Okay. like look, from my end, that was my list of questions. So unless there's some from the audience, I think I'll give you back 2 minutes. But, hey, that was a great conversation. I really enjoyed it. Thank you.

Thank you.