Veritone Inc (VERI) 2018 Q2 法說會逐字稿

Good afternoon. Welcome to Veritone's Second Quarter 2018 Earnings Conference Call. After the market close, Veritone issued a press release announcing its results for the second quarter ended June 30, 2018. The press release is available in the Investor Relations section of Veritone's website. Joining us for today's call are Veritone's Chairman and CEO, Chad Steelberg; and the company's Chief Financial Officer, Pete Collins. Following their remarks, we will open up the call for questions.

Please note that certain information discussed on the call today will include forward-looking statements about future events and Veritone's business strategy and future financial and operating performance, including its expected operating performance for the third quarter of 2018. These forward-looking statements are subject to risks, uncertainties and assumptions that may cause the actual results to differ materially from those stated or implied by those statements. Certain of those -- of these risks and assumptions are discussed in Veritone's SEC filings, including its annual report on Form 10-K. These forward-looking statements are based on assumptions as of today, August 13, 2018, and Veritone undertakes no obligation to revise or update them.

In addition to the company's GAAP financial results, during this call, management will be presenting and discussing the company's earnings before interest expense, depreciation, amortization and stock-based compensation, or EBITDAS, which is a non-GAAP financial measure. A reconciliation of the company's EBITDAS to its net loss is included in the company's press release issued today. Finally, I would like to remind everyone that this call is being recorded and will be made available for replay via a link available in the Investor Relations section of the company's website at www.veritone.com.

Now I'd like to turn the call over to Veritone's Chairman and CEO, Chad Steelberg. Sir, please proceed.

Welcome, everyone, and thank you for joining us today. Q2 was a solid quarter with several highlights. Triple digit year-on-year growth in SaaS revenues, total customers, total accounts, third-party cognitive engines added and hours of video and audio files processed within our operating system.

Our SaaS business trailing 12-month year-on-year net revenues growth exceeded 200%. Earlier today we issued 2 press releases announcing the signing of business acquisitions, Wazee Digital and Performance Bridge Media. These companies had 2017 net revenues of $19 million and $3.7 million, respectively. The whole team here at Veritone is charged by these important developments. Adding the Performance Bridge and Wazee team members, customers and technology are important steps in Veritone's strategy. In this call, we will focus on Veritone results. Next Monday, August 20, at 4:30 p.m. Eastern Time, we will have an investor call to review the Wazee and Performance Bridge businesses and describe how they fit into our strategic plan.

For those investors new to Veritone, I'll quickly run through core aspects of our business and offerings. Veritone is comprised of 2 business units. First, a high-growth SaaS business, providing artificial intelligence and solutions to analyze information in all its primary forms, including audio, video and text at scale and with accuracy and speed. We address 3 vertical markets today with our Saas solutions: Media & Entertainment; Legal & Compliance; and Government.

Second, a media agency known as Veritone One. The agency is a high-margin business, which provides us with valuable information about the needs of media businesses and demonstrates the use of artificial intelligence to improve the effectiveness of media buys. The media space is a natural and huge user of artificial intelligence, thanks to its content-rich data trove. Veritone One has brought us into major radio broadcasters, which have in turn made Veritone One one of the largest suppliers of AI-enabled SaaS solutions in this market.

Now turning to our AI SaaS business, I'd like to take a few minutes and provide a small primer on AI technology and our offering. First, artificial intelligence-based solutions analyze video and audio content, which were previously indecipherable by computational tools that required structured content. AI has evolved to the point where large and complex bodies of video and audio content can be analyzed by machines much more quickly and effectively than human capabilities. The AI analysis of this content is performed by cognitive engines developed by companies throughout the world. Cognitive engines can be deployed within the Veritone operating system, aiWARE. Cognitive engines are specialized machine-learning algorithms trained to perform specific functions such as transcribe spoken language or recognize faces or objects within videos. The more specialized the engine, the higher the level of accuracy. The challenge is that most tasks are multidimensional and oftentimes no single engine is capable of producing the required analysis at a sufficient level of accuracy. For example, a law enforcement officer may need to search surveillance footage for specific faces and tattoos to identify suspects. That requires 2 different cognitive classes, face recognition and object recognition, to operate in tandem. Within each of these 2 cognitive classes, multiple cognitive engines must be needed to enhance the accuracy of the search. To solve this very complex technical problem, we developed our AI operating system, aiWARE, which is similar in concept to a computer's operating system. A PC operating system enables third-party components such as CPUs, RAM, peripherals and applications to work together. Our aiWARE allows a multitude of cognitive engines and applications to work together and delivers an orchestrated result for analysis of unstructured and structured content. At the heart of aiWARE is our unique orchestration technology, which we call Conductor. Today, Conductor uses our own proprietary AI to learn the capabilities of each engine within the NLP class and selects the right engine, or engines, to achieve the best results based on the content being processed and the desired information the customer's trying to discover.

In addition to Conductor and third-party cognitive engines, aiWARE has application that allows users to organize the output and take actions, much like how applications work with a computer's operating system and processors. We have developed a suite of 6 standard applications. Third parties are able to deploy applications on aiWARE, and we are developing specialized applications for customers. Customers deploy aiWARE because it allows them to analyze information in all its primary forms including audio, video and text at scale and use the results to meet their objectives with accuracy, speed and ease.

Our customers can be up and running on aiWARE in under an hour for a monthly fees as low as $500. They generally begin their AI journey on a single project and expand as the business value becomes evident. They can also develop custom applications to extend the usefulness of aiWARE for their organization's specific use cases.

We have multiple partner communities in our collaborative ecosystem: developers of cognitive engines, applications and data products as well as system integrators. The cognitive engines on our platform come from a wide range of companies, from small tech shops to large public companies, including Google, Microsoft, IBM Watson and Nuance. We pay the engine developers based on the amount of media we process through them. These companies integrate with aiWARE operating system, providing greater reach for their engines and greater accuracy of results due to Veritone's Conductor AI layer.

Next, I'll move on to the current business climate and then turn it over to Pete for financial details. The AI industry, as we know, is still in its early days. The largest and perhaps the earliest vertical to deploy AI for business value is the Media & Entertainment industry. Their business model, of course, is based on the creation and monetization of unstructured data. This space is where we have been the longest and, as such, it has shown the greatest and most consistent net revenue growth. It is up over 250% on a trailing 12-month basis. If I were to plot this vertical on the product adoption curve, it has crossed the chasm with their first and second evolution in deployments, the 1x and 2x wins, as I like to call them. Today, this industry segment is getting more sophisticated with how they use AI.

Well, at first, they took advantage of aiWARE's AI-driven processing speed and ability to quickly find and validate ads that ran for their advertising customers. Today, they are using aiWARE for applications such as analyzing which on-air talent connects best, which audience segments when discussing specific topics. They use this information to train on-air talents and select talent and content for enhanced ratings and ad revenues.

We see clearly how aiWARE capabilities will help every industry. Each quarter, we are adding more cognitive engines, and our operating system is literally getting smarter, enabling us to provide our customers with richer and more accurate analysis than any single-point solution provider.

We have now taken these capabilities and built go-to-market models addressing the legal, compliance and government markets. As the Media & Entertainment space has crossed the chasm in AI adoption, these other verticals are plotting their arrival and are in the education phase. To further our competitive advantage, we are building our arsenal of readiness for them. Specifically, we're expanding our tech abilities geared to their markets, sales channels and ecosystems of tech partners.

In the first quarter, we experienced a lift-off in the legal market with a meaningful revenue contribution. As we pointed out on the Q1 call, the bulk of our legal revenue came from a single project, which required transcription of over 1 million phone calls. Our experience with this project demonstrates what we said last quarter, our net revenues in the legal vertical will be lumpy until the number of projects we have increases to the point where it smooths out those lumps. We are continuing to execute our strategy to generate more sales in this market. This quarter, we brought to the legal market our capability of translation in addition to transcription, which is a meaningful expansion of the services we offer to this market. As with the Media & Entertainment space, growth will come, but will not always be consistent.

The Government vertical remains in the early days as well. We have multiple -- we have several proof of concepts underway, one of which has resulted in investigators matching multiple facial images recorded at crime scenes to a known individual in a library of over 40,000 individuals. While this is a great example of our early efforts in the Government vertical, the key revenue-contributing verticals for us this year will continue to be Media & Entertainment and Legal & Compliance. And we will expect Government to ramp up its net revenues in 2019.

In the SaaS base at large, our "land and expand" strategy continues to work. Again, during the quarter, our number of new customers expanded and the number of new accounts expanded further. In Q2, we had 126% year-over-year growth in customers and a 270% year-over-year growth in total accounts. It is clear to us that conviction in Veritone continues to grow. Conviction from customers, as seen in a 62% year-over-year growth in monthly recurring revenue under agreements, and conviction from AI engine developers, as seen in the 210% year-over-year growth in Q2 in active third-party cognitive engines integrated within aiWARE.

We are encouraged about the overall progress we are making in executing our strategy, and we see deep pools of opportunity in the market. To go after these pools with speed, scale and impact, we will increasingly be partnering with tech companies who are already processing petabytes of data for these customers. As an example, one such tech partner reached out to Veritone and brought us into Bloomberg, CBS News and the Masters. But as I have said many times, we are in the very early innings of a long game.

Now I will turn the call over to our CFO, Pete Collins, to walk us through our financial results and key performance indicators for the second quarter of 2018. Pete?

Thank you, Chad, and good afternoon, everybody. Our net revenues in the second quarter of 2018 increased 2% to $4.2 million from $4.1 million in the second quarter of 2017. The year-over-year increase in net revenues was mainly due to a 147% increase in SaaS net revenues from our AI platform, which increased from $0.3 million in Q2 last year to $0.9 million in Q2 this year.

Our total net revenues increased, even though our Media Agency's net revenues decreased 12% from $3.7 million in Q2 last year to $3.3 million in Q2 this year.

Looking closer at our AI operating system business, we generated net revenues primarily from one vertical in Q2 2018: Media & Entertainment. The Media & Entertainment vertical delivered the majority of the AI platform's net revenue in the quarter, as we continued to build on the business development work we started in 2016.

Our "land and expand" approach delivered the media vertical's revenue growth, particularly with iHeartMedia and ESPN. Our monthly recurring revenue, or MRR, under agreements in effect at the end of Q2, increased 62% to $214,000 from $139,000 -- $132,000 in the second quarter last year. This year-over-year decrease of $0.4 million or 12% in net revenues for our Media Agency's business was due primarily to the large initial campaign for a significant client in the first half of 2017 that did not recur in 2018.

Varying client situations and challenges, independent of Veritone, can impact the Agency business and this is what took place in the second quarter. Our Media Agency's net revenues have already recovered in bookings during the current quarter, and such dynamics reflect the nature of the agency business. We had 74 active media clients in the second quarter of 2018 compared with 45 in the second quarter of 2017, an increase of 64%.

Our gross profit in this year's second quarter decreased 11% to $3.3 million or 80.3% of net revenues from $3.8 million or 91.8% of net revenues in Q2 last year. This shift was the result of the decline for the Media Agency revenues, which carry a higher gross margin. The gross margin of our AI platform business is lower than in our Media Agency due to the difference in their business models. Our AI platform grows -- as our AI platform grows to become a larger portion of our total net revenues, we expect our gross margin will decrease. Our SaaS business incurs cost for the cloud platform and third party cognitive engines, which results in a 65% to 70% gross margin rate for the aiWARE net revenues. Turning to our operating expenses, our total operating expenses in the second quarter of 2018 were $17.8 million, an increase of 53% from Q2 last year. The year-over-year increase in total operating revenues was due primarily to our 44% increase in headcount, particularly in software development, data science and sales and marketing, which we expect will further enhance our aiWARE platform and drive higher AI revenue in the future. Included in our total operating expense this quarter was $2.7 million of stock-based compensation compared with $1.8 million in Q2 last year. Loss from operations in Q2 this year totaled $14.5 million. This compares with a loss from operations in Q2 last year of $7.8 million.

Net loss attributable to common stockholders in Q2 totaled $14.3 million or $0.88 per share. This compares with a net loss of $24.5 million or $2.94 per share in Q2 last year. These EPS figures are based on 16.3 million weighted average shares outstanding for Q2 of 2018 compared with 8.5 million weighted average shares outstanding in Q2 of 2017. It's important to point out that our net loss for Q2 included the $2.7 million of stock-based compensation as well as $0.5 million of depreciation and amortization expense related primarily to the amortization of a software technology we purchased in December of 2017.

Now turning to our non-GAAP metrics. Earnings before interest expense, depreciation, amortization and stock-based compensation, or EBITDAS, for Q2 was a loss of $11 million compared with a loss of $6 million in the second quarter of 2017 and a loss of $10.2 million in Q1 of 2018. The higher EBITDAS loss compared to the year ago period was due primarily to the addition of software development, data science and sales and marketing resources, as I mentioned before. While our EBITDAS loss was higher on a year-over-year basis, it was similar to the prior quarter loss, demonstrating a leverage in the model where we were able to increase our AI net revenues by 147% and hold costs relatively flat.

Now turning to our balance sheet, which remains strong. At the end of the quarter, we had cash, cash equivalents and marketable securities of $78.2 million and no long-term debt. In June, we completed our second follow-on offering of common stock. We issued 1,955,000 shares of common stock and received net proceeds of $32.9 million. We will use this additional cash to further develop aiWARE and to sustain our sales and marketing efforts. We expect that our operating system development and sales and marketing initiatives will increase our future net revenues. Also, the additional cash provides flexibility for potential future M&A, including the 2 acquisitions we announced this morning.

Next, I'd like to shift gears to our key performance indicators, or KPIs, for both our AI platform business and our Media Agency business. Our AI platform business exceeded our KPI guidance for the second quarter of 2018 across most metrics. We had a record 86 customers on the platform at the end of the quarter compared with 70 at the end of last quarter and 38 at the end of Q2 last year. In terms of total accounts on the platform, we had 625 at the end of the quarter compared with 591 at the end of last quarter and 169 at the end of Q2 last year. We had 214 third-party active cognitive engines on the platform at the end of the quarter compared with 184 at the end of last quarter and 69 at the end of Q2 2017. And finally, during the quarter, we processed 2.7 million total hours of video and audio files compared with 0.4 million hours in Q2 last year. The total hours processed this quarter nearly matched the amount we processed over all of 2017. As we grow and expand our AI platform, we will continue to provide goalposts to make it easier to track our success. Along that line, we expect to end the third quarter 2018 with 9 new customers, approximately 30 new accounts and approximately 20 new active third-party cognitive engines on our platform and having processed approximately 2.7 million hours of video and audio files during the quarter. For our Media Agency business, we evaluate 3 key performance indicators: first, the number of new clients added under master service agreements; second, the total number of active clients; and third, the average media spend per client.

During Q2 this year, we added 14 net new clients compared with 16 net new clients in Q2 last year. In terms of active clients, we had a total of 74 as of the end of Q2 of 2018 compared with 45 at the end of Q2 2017, a 64% increase. During Q2, our average media spend per client was $425,000 compared with $695,000 in the same period last year, which is a 39% decrease. It's important to keep in mind that while this business is mature and provides a solid foundation for Veritone, it can also experience volatility in net revenues from quarter-to-quarter. That completes my financial summary.

Now I'll turn the call back over to Chad. Chad?

Thank you, Pete. As a first-mover company with many shoots of opportunity, we see some are growing faster than others. Some have been planted longer and yet the soil is fertile for all. The industry is new. We know we are early and variability is inevitable quarter-to-quarter. The quarter-to-quarter trajectory of our business will be both unpredictable and lumpy, as a result of the industry dynamics. We do, however, expect to deliver triple-digit growth for our SaaS business year-over-year. We will continue to experience sequential variations in revenue growth rates in individual markets due to our current size.

Some of our business comes from projects such as legal cases that are varying in revenue size and duration. When a case ends, the associated revenue stream from the project also ends. As I have said, we are very early in innings of a long game. And over time, as we continue to execute our strategy and build our business across all of our verticals, we expect to deliver significant, sustained revenue growth.

Veritone entered its fifth year of business last month. My brother, Ryan, and I founded Veritone in July of 2014 when I became convinced that the updated AI models would lead to a massive paradigm shift and competitive advantage, or disadvantage, for organizations from the ability to best analyze and gain insights from the tsunami of information the world is producing.

Mainframes, open-system computers and associated software, together with chief analytic officers, were able to tackle the first waves of terabytes, petabytes and zettabytes crashing upon us. But those same computer systems and software could not unlock value from unstructured data. 80% of the information created today is in an unstructured form, and the information being created is doubling every 24 months. I thought, clearly, that a new solution would be needed to unlock the insights and competitive advantage from both information formats, structured and unstructured, together, and the massive volume of information that is now within our daily lives.

Artificial intelligence, paired with human ability to ask the right questions, is how the world is answering this giant change. Veritone has created the operating system for this new IT world order. Some of the most valuable technology companies in the world are working with Veritone and within our operating system. Why? Because Veritone makes their offerings better. With every byte of information that Veritone OS ingests, it becomes smarter and offers more value to our ecosystem of users, system indicators and tech developers.

I am excited about the future and even more eager to bring this vision to its fullest potential. It is my goal to help the market continue to see how Veritone is good for AI and good for the world. Thank you for joining us on this journey. Again, thank you for your time today. We look forward to updating you on our progress on future calls.

We're now ready to open the line for questions. Operator?

(Operator Instructions) And our first question comes from the line of Rob Stone with Cowen and Company.

I wanted to ask first for a little more color on the trend in hours of media processed. It was roughly flat sequentially, maybe down 3% from Q1, and you're guiding to essentially the same number for Q3. When the Q1 results came out, you were actually guiding for about 3.5 million hours. So I'm curious if there was a large project or something that didn't happen? And after many quarters of rising hours, it's flattening out. So help us understand what's going on there.

Sure, Rob. This is Chad. We have been looking at our business and it really comes down to 2 revenue streams. One is a very predictable, classic SaaS business model with a licensing operating system and applications; and then a variable component which is tied to, in many cases, an unpredictable stream different than, let's say, a radio station or a television station group, which puts out a fairly consistent amount of content on a month-to-month basis that's easy to forecast. What we've seen in the legal sector as well as in some of the new Government work, because of the project nature of the analysis, it's become very difficult for us given the law of small numbers for us to accurately forecast what that's going to look like even in a very few short months ahead of us. So rather than continuing to forecast to what we expect sort of a best case or a target case, we're falling back on a number which we know we can deliver on based upon what our more consistent media and other forms of consistent processing data is telling us on that trend, and we'll let the variability drift to the upside versus to the mean.

Okay. That's a great segue into my next question, which is AI revenue was down about 29% sequentially, although it was up triple digits year-over-year. And we know that you had a significant chunk from the Legal & Compliance segment project driven in Q1. Was the decline quarter-on-quarter -- how much of that was from legal? Were there other things that declined? Just a little bit more on the sequential trend in reported revenue.

So Rob, the underlying media business was up very nicely, like, 25% sequentially. The -- what we saw -- when you refer to the sequential decline, it was all related to the legal vertical, which is what we had spoken about last quarter and Chad kind of covered in his section of the prepared remarks that there was a large project, the project that had 1 million phone calls that we transcribed and we did not see a similar level of project-type work in the legal vertical this quarter. So that explains the overall sequential decline, 2 factors: Media & Entertainment being up 25% on a sequential basis, but legal being down due to that large project that we had called out in Q1.

And then finally, just a question for you, Pete, which is -- and I know this may be superseded by the impact of acquisitions, which, I guess, you're not ready to talk about tonight. But any color you can provide on what we might look for in the sequential trend in operating expenses?

I think overall, operating expenses are in pretty good shape. I think that we do have an increase in our stock-based compensation this quarter. That's in the neighborhood of $1.9 million to $2.2 million. But beyond that, I think that the -- there aren't any other significant items, apart from the M&A vertical, where we will have additional headcount for the quarter, and then we'll also have obviously transaction costs associated with those acquisitions.

And our next question comes from the line of Mike Latimore with Northland Capital Markets.

I guess on the media AI business, how are you looking at that going forward? Are there some further upsell opportunities? Are there new contracts? It's been the sort of strongest business segment to date. I mean, can you give a little color on kind of what to expect next in that vertical?

Yes, we continue to see strong growth both in terms of new customer adoption both domestically and abroad as well as based upon our metrics, more and more accounts. So we're seeing that "land and expand" model of not only new use cases, but new channels in the media space, whether it's additional television, competitive analytics, et cetera, driving usage up. So probably from my perspective, the thing that I'm most bullish about is some of the applications that are now starting to roll out that are actually not just using the same physical application on aiWARE, but new applications that they are adopting, whether that's, again, in the redaction phase or other forms of business use cases that companies like ESPN and Bloomberg are deploying. So as AI itself performs better and the accuracy rates of specific engines, for example, logo detection or even transcription, reach and approximate human levels, we're going to see more and more use cases open up in the Media & Entertainment space, things like closed captioning that still today are done by humans, we see in the not-so-distant future opportunities to replace and augment those capabilities. So again, as aiWARE improves, as the number of engines continues to increase and the accuracy in those also increases, we continue to expect extremely strong, positive growth on the "land and expand" strategy in Media & Entertainment.

Got it. And then on the bookings in the quarter, they doubled, I guess -- a little more than doubled sequentially and year-over-year. Was that all from sort of the media category? Or was there other things in that?

Yes, it's -- the vast majority of it, Mike, was in the media vertical.

Okay. And then just last on the legal. I guess, the large case you had in the first quarter, is there sort of more opportunity in that specific case? And then maybe just a little more color on kind of the pipeline of cases that might be applicable throughout the rest of this year?

Right. So just to give a little more color on that. When we're dealing with our Media & Entertainment, we're usually almost exclusively dealing directly with the end customer. In the Legal & Compliance space, specifically in the legal e-discovery sector of Legal & Compliance, we're effectively 4 layers removed from the paying customer. You have the client, you have the law firm, you have the MSP that's processing the e-discovery files and then you have Veritone. And what that does, it really just creates a lot of variability for us. We had a number of very large cases in the pipeline this quarter, in Q2. We expected at least 1 or 2 of those to hit in addition to the smaller ones that were running through. But due to circumstances outside of our control, whether it was a lawyer quitting and going to another firm or something else, those cases did not originate and land on Veritone from an analytic standpoint. So they still exist. We see that pipeline continuing to build. But predicting which quarter they're going to land in when we're dealing with such small numbers is obviously challenging. So maybe one final comment on that, Mike. Legal & Compliance but focusing purely on e-discovery, we are very confident that we are the dominant solution in the space. We have fantastic technology. And if you talk to our customers, the MSPs that have used us, it's second to none. So this is an early-stage problem for us. We're doubling down the space, and we will see this through and transform the legal and compliance industries.

And our next question comes from the line of Darren Aftahi with Roth Capital Partners.

Can you just talk on the media and entertainment vertical? Like, what was the mix in AI between existing and new clients and how did that change from Q1?

So it's -- a lot of the work, Darren, is on existing clients and it's that land-and-expand type of an approach. We are -- when you look at the customer count growth that we present, the bulk of that is in Media & Entertainment. But as Chad's talked about before, we typically start out with relatively modest levels of revenue on many of the new clients. And then as they get more familiar with the platform, they develop new business cases, start ingesting more content and our revenues build. So it's really a function of more the kind of internal growth rate versus bringing on large new customers.

Got it. And nice you're going to set up another call for the acquisitions. I'm just curious if you might talk high level just about the leveragability of the Wazee Digital asset management platform [option] verticals beyond Media & Entertainment. I think you called out Government in the PR. But I'm just kind of curious how leverageable that is across other sectors perhaps?

Yes, we believe Wazee's core product, if you split their business in 2 from core to then their more media monetization platform, but their core product, we believe, has extensibility into other verticals in addition to Government alone. So we're not going to talk more about that, but it is not a single point solution that's hardwired to the media and entertainment industry, and we see that as a broader opportunity for the company.

That's helpful. And just 2 more, if I may. If you were to back out sort of the onetime kind of campaign within the Media Agency business as well as the onetime Government, did those businesses -- can you, I guess, one, on the Media Agency business, talk about how fast that grew ex that campaign, if you have it? And then two, how fast sequentially legal grew if you back out that onetime revenue? And then just my last one was just any update on FedRAMP would be helpful.

So just kind of high-level here, Darren. You take out that one -- kind of initial campaign a year ago in the Media Agency and the overall revenue, I think, off the top of my head would have been kind of in the low double digits, probably 10% to 15% on a year-over-year basis. So very solid growth there. On the AI business, you're talking about growth of probably closer on a sequential basis. I already said that the M&E business was up 25% sequentially. That's probably about 25% to 30% would have been about what we would've delivered without that -- excluding the impact of that one large legal project from a year ago. Then you asked a question about FedRAMP.

So I can address FedRAMP. So we are now through FedRAMP certification. We have an Authorization to Operate -- Authorization to Test, I think it is, ATT, and we continue to work with a number of federal agencies with that program. So great progress, got our certification. Now we're simply working through the actual specific clients that are going to be deploying the platform.

Yes, the ATO has 2 more phases to it, and we think that we'll be able to get both of those phases done within the next 6 months.

Correct.

And our next question comes from the line of Chad Bennett with Craig-Hallum.

So Pete, can you maybe kind of speak to since we kind of have these onetime deals on the legal side that are tough to predict and I think it's obviously the way to think about it is kind of haircutting those in terms of your guys' expectations? But if we were just to look at, at the second half revenue both for the agency business and for the aiWARE SaaS business, kind of how do we think about it sequentially in the September quarter and the December quarter excluding obviously the 2 acquisitions that you guys just announced?

So on the Media Agency, Chad, if you look back at the quarterly revenue a year ago, it was higher in the first half of the year than it was in the second half of the year, and that's because of this initial campaign that a large customer did last year. That extended just through the second quarter and then it dropped off and didn't -- wasn't at that same level over the back half of the year. So the revenue growth that would be more sustainable in the second half is very similar to the figure I just shared with Darren in kind of that 10% to 15% level on the Media Agency on a year-over-year basis.

Okay. How about the AI business?

Well, I think -- where we are in the second quarter is the vast majority of that $900,000, $860,000 revenue is from the Media & Entertainment vertical, which we've talked about has a trailing 12 growth rate of 200-plus percent. So I think that -- I wouldn't necessarily sign up for the 200% growth rate continuing each quarter going out the back half of the year, but I think we're well into the 100% to 150% type growth rate for that vertical. And then to the extent we've got -- and there wasn't significant meaningful legal revenues a year ago in the second half of the year. So I think if we -- as we build up that legal vertical -- but again, that's kind of the variable aspect which Chad talked about how it's not baked into the number of hours we're forecasting for Q3, that would be really upside for us.

Okay. And then, again, I know you have the call next Monday for the acquisitions, but can you just comment at least from a financial profile standpoint if the combined revenue stream would be a positive operating income EBITDA business for you guys?

Yes, they're both -- one is -- on a combined basis, their profit -- they have EBITDA profits.

And our next question comes from the line of Tom Diffely with D.A. Davidson.

First question for Pete on the margins. You talked about how the AI margins are kind of in the 65% to 70% range. Wondering, though, if that changes when you go through the legal space where there are several layers of distribution you have to work through?

So actually this is Chad. I'll address that. Despite the layers in the legal, the value proposition, which is how we charge, not based upon the commodity nature of the service, is -- has higher margins in the Legal & Compliance space than we do in the Media & Entertainment primarily due to the value of the platform and what they're paying for in terms of the increased accuracy that really only Veritone can deliver around tough cases.

Okay. Great. And then Chad, maybe a product question for you too then. About a quarter ago you released the real-time framework for the aiWARE program. Has that opened any new opportunities for you? Or has it driven any recent wins? Maybe a quick update there.

Yes, the real-time framework is becoming mission-critical to us in almost all of our newer opportunities of customers. Part of that "land and expand" that we're seeing that growth rate with our existing media customers is due to that real-time nature. It's -- again, as accuracy improves and the real-time platform now that's been deployed, it really opens up the closed captioning marketplace and other forms of use cases for the media and entertainment. In the Government sector, the specific use cases I talked about in the prepared remarks around identifying suspects, et cetera, also is going through that real-time platform. As you can imagine, when you're looking for a potential suspect, time is of the essence. And so real-time is definitely critical not only to the local agencies, but also the federal ones that we're in discussions with.

Okay. Great. And then finally, when you look at the KPI, the third quarter projections, I assume all of that's just on the base business and has nothing to do with the acquisitions at this point?

Correct.

And our final question comes from the line of Sameet Sinha with B. Riley FBR.

This is Lee Krowl on for Sameet. A couple of questions. Just wanted to kind of understand channel mix with legal kind of being difficult to see when cases actually flow through, but it seems like channel partners are a very important part of the growth strategy. So can you maybe talk about the sales process and maybe the mix of revenue that's coming in through reseller channels and maybe the revenue that's coming in via your own sales force?

So on media -- Lee, you're referring to Media & Entertainment?

I mean, just general -- generally speaking.

So -- yes. Well, in Legal & Compliance, on the e-discovery side, that's all going to be through kind of channel partners or MSPs, right, because they're the ones that provide the foundation for the law firms who provide the service to the client, the ultimate clients that are paying. On compliance, which is a category that we're pushing into, it's a combination of us working directly with large financial institutions as well as system integrators or tech platform providers that have -- providing an outsourced solution to a lot of the financial services industry that we're addressing. On Media & Entertainment, a lot of the work we've done to this point has primarily been through our direct sales force, although the pipeline now is shifting more towards a lot of channel partner opportunities coming through. So as we've ramped up our resources there and expanded our capabilities there and done some technical integrations, that part of the sales channel, that pipeline is moving more towards the channel category versus the direct sales, although our direct sales team is still out there and working hard on pursuing leads and opportunities that they've been going after. Government is probably more towards the channel side, although our dedicated people there are really focused on enabling and working with those channel partners. But as you can imagine, getting into a lot of, especially federal agencies, the path there is working with channel partners and system integrators who already have an established business with those agencies.

Got it. That makes sense. And then just a quick question on SaaS. To the extent you can maybe quantify it, but just how much of the revenue in SaaS is purely licensing? And then how much is contingent upon like a volume element of it, i.e., hours of processing?

So the majority of the revenue that we're generating today is in the licensing category. The one part of the business that historically has had a large amount of processing is on that legal vertical, which is why we called that out in the first quarter. But on the Media & Entertainment side, most of that is licensing revenue. There are opportunities for us going forward to do processing kind of projects, say, analysis of large archives in Media & Entertainment, but that's really more kind of a future opportunity. We have not generated a lot of revenue in that phase yet.

Got it. And then just the last one on the legal space. You guys mentioned you had a pretty solid case pipeline. Would that imply that you expect to see kind of a bounce-back in the legal vertical revenue in Q3?

We would -- we can't commit to that. A quarter ago, we thought we would have had a good -- part of that work coming through revenue from that vertical in Q2. So the factors Chad talked about, about cases that, frankly, were in 7-figure contracts, that were in closed 1 state here internally in the middle of the quarter that didn't happen, gives us the reason to step back and say we're not going to forecast or count those until they're completed and everything's done on them.

At this time, this concludes our question-and-answer session. If your question was not taken, you may contact Veritone's Investor Relations team at VERI@liolios.com. I'd now like to turn the call back over to Mr. Steelberg for his closing remarks.

Great. Thank you for joining us on today's call. We want to thank our employees, partners and investors for supporting us as we pursue our mission of building the AI operating system of the future. We look forward to updating you on our progress on our next call. Operator?

Thank you for joining us today for Veritone's Second Quarter 2018 Earnings Call. You may now disconnect. Everyone, have a great day.