Skip to content Skip to sidebar Skip to footer

Duos Technologies Could Become An AI Pure Play

Source: Seeking Alpha as of 26-04-2020

Summary

Duos Technologies has got a foothold in the rail industry where it is automating railcar inspections, and it is adding AI smarts to this process. The basic capabilities of the company can be applied to a whole range of situations and verticals; it is more or less a generic capability. While its rail revenue could significantly increase through regulatory change, the company is getting into additional verticals which gets us excited about the opportunity here.

Duos Technologies (NASDAQ:DUOT), a company that builds automated inspection sites for rail cars, recently moved up to the Nasdaq and got an infusion of funds that should allay fears about its cash situation at least for a year or so. Duos has automated the inspection of railroad cars through proprietary technology, and our investment thesis does not hinge on that market alone. The inspection technologies and the additional AI platform the company is building out have wider applications in other verticals. Indeed, a description of its technology in the 10-K sounds wonderfully abstract:

…focuses on the design, development and turnkey delivery of proprietary “intelligent technologies” that enable our customers to derive measurable increases in return on investment for their business.

The company has two divisions:

  • Duostech (turnkey data capture and management command and control interface)
  • TrueVue 360 (machine learning).

Growth

There has been a fair amount of revenue growth since H1 2018:

ChartData by YCharts

We see a number of growth vectors:

  • New customers.
  • Venturing of its data capture towards other verticals.
  • Adding algorithms to its rail solutions for existing (and new) customers.
  • Building a stand-alone AI division to pursue a host of verticals.
  • Moving towards a more recurring revenue driven business model.
 

Much of this comes from the company’s RIP, or Intelligent Railcar Inspection Portal, which enables customers to automate the inspection of their railcars (which by law have to be inspected on some 110 parameters, the so-called “Why Made” codes). Automating inspection brings potentially substantial cost savings (10-K):

Under current practice, inspections are conducted manually; a very labor intensive and inefficient process that only covers a select number of inspections points and can take up to 3 hours per train. It should be noted that approximately 50% of the rail industry’s operating costs are for maintenance, including 30% of the time trains spend in workshops resulting from manual failure diagnostics.

The company has automated this process by building technology into railtracks (typically installed between two rail yards), which captures images and data in real-time with trains going at up to 70mph. The images and data are then processed and presented in real-time by the company’s proprietary intelligent user interface, branded as centraco, and analysed on a host of mechanical anomalies and with the help of its AI program TrueVue 360. Once RIP is installed, it generates a stream of recurring revenues which are set to reach 20% of revenues (per the Gateway Investor Conference presentation, with thanks to Sergio Heiber). These systems have been verified by the likes of Johns Hopkins University Applied Physics Laboratory (JHU/APL), the Department of Homeland Security (DHS) and the Transportation Technology Center, Inc., a wholly-owned subsidiary of the Association of American Railroads, a transportation research and testing organization (TTCI) and have found commercial success.

There are a number of growth vectors and catalysts:

  • Regulatory change and innovation.
  • Completing the TrueVue 360 AI applications for rail safety.
  • Developing TrueVue 360 AI applications for other verticals.

Regulatory change involves approval of a completely automated railcar inspection process, which (10-K):

The Company is collaborating with certain industry professionals to pursue such regulatory rule changes and we believe that there will be broad acceptance of our technology as soon as a majority of required AI algorithm models are completed and tested.

Given the above description about the laborious, time-consuming and labor-intensive process of manual inspection, it comes as no surprise that there is broad support in the industry for such regulatory change. The innovation involves enabling automated inspections at up to 120mph with additional sensor technologies which the company currently has under development and is scheduled for H1 2020. There is extensive description in the 10-K of all the proprietary capture and sensor technologies (which are also sold as stand-alone systems) and there are also some new applications here (10-K):

We have completed a pilot (proof of concept) of our Platform Analytics tunnel and track intrusion technology concept deployed for the New York City Transit Authority (“NYCT”). The technology is designed to automatically detect objects fouling tracks adjacent to transit passenger platforms and to alert incoming rail traffic to that effect. Field installation of the prototype has been completed and field testing employing our truevue360 AI application has been conducted since mid-4th quarter of 2018 with near “0” false positive/negative episodes.

There are quite a few other innovations and new applications, also ones outside the rail industry described in the 10-K like:

  • Pantograph Inspection System apis – A system designed to inspect the structure connecting transit locomotive high voltage power lines.
  • Tunnel and Bridge Security.
  • Remote Bridge Operation.
  • Virtual Security Shield – Intrusion detection zone, Radio Frequency Identification (RFID) tracking and discriminating “Friend or Foe” modules.
  • Facility Safety and Security – A suite of intelligent technologies-based homeland security applications for the “hardening” or safety and resilience of facilities against natural or man originated threats for the protection of critical facilities (energy, water, chemical facilities).
  • Multi-Layered Enterprise Command and Control Interface centraco – The interface for all these applications.
  • Intelligent Analytics Suite Praesidium (data and video analytics).
  • Automated Logistics Information System (Alis).
 

On the latter (10-K):

We have completed the development and commercially deployed a proprietary intelligent system to automate security gate operations at nine (9) distribution centers owned and operated by a national retail chain. Leveraging our proprietary multi-layered Enterprise Command and Control Interface technology (centraco®), the automation of gatehouse operations provides substantial improvements to the efficiency of distribution center traffic flow, resulting in the potential for significant return on investment to the customer. The Company initiated marketing this new technology to enterprise-level owners of distribution centers throughout the United States and beyond and expects to scale sales of this product line starting in early 2020.

It’s interesting to see how these technologies have a wide range of applicability and feed on one another. Take for instance Praesidium, which is an integrated suite of analytics applications which processes and analyzes data streams from a virtually unlimited number of conventional or specialized sensors and/or data points. From the 10-K:

This application suite also includes a broad range of conventional operational system components and sub-systems, including an embedded feature-rich video management engine and a proprietary Alarm Management Service (“AMS”). The AMS provides continuous monitoring of all connected devices, processes, equipment and sub-systems, and automatically communicates to centraco®, the Company’s enterprise information management suite

We had to quote somewhat extensively from the 10-K product descriptions to illustrate three important points:

  • The company’s technology isn’t just limited to rail; it has multiple applications and verticals where it can be useful and more are added frequently.
  • The technology platform works seamlessly from image and data capture to information management systems to analytics.
  • The analytics is significantly enhanced by machine learning, and this too has near universal application.

Recurring revenue

Most of the revenue is from projects. From the 10-K:

However, the company does generate recurring revenue in the form of maintenance and technical support from the newer projects and management expects this to grow as older projects convert to this format (10-K):

The Company continues to replace the declining revenues from one customer with new, long term recurring revenue from new customers which will be coming on-line in the next several months. The maintenance and technical support revenues are driven by successful completion on projects and represent services and support for those installations. The expectation is that revenues from this area will continue to grow based on the success of multiple installations in 2019.

We also surmise that there is a recurring revenue component in TrueVue 360 related to the quantity of datapoints. So we can expect some acceleration in recurring revenue this year, which also happens to boost gross margins.

There is also a line above that says IT asset management services, with revenues growing at 141% last year (albeit from a small base). From the 10-K:

This was the result of the ITAM division releasing a new version of its dcVue™ software which is anticipated to broaden market acceptance of its offerings. The software was beta tested at a financial institution with the objective of ultimately rolling out to additional locations and we anticipate a positive impact on revenues in 2020. The division continues to execute consulting services engagements through its partners.

These license part doesn’t have recurring revenues though.

Machine learning

The analytics part is also rapidly developing with the company betting big on developing algorithms through machine learning. Take for instance this description of Praesidium (10-K):

Our algorithms compare analyzed data against user-defined criteria, rules in real time and automatically reports any exceptions, deviations and/or anomalies… Our core modules are tailored to specific industry applications and the analytics engine(s) process any type of conventional sensor outputs, also adding “intelligence” to any third-party sensor technology. A key benefit is that the customer may often retain existing systems and we would integrate these into an overall solution.

Data comes from a variety of sensors and image capture devices, and this almost amounts to a generic technology. While the company is in the early stages of its AI efforts, we can see it developing into a generic AI pure play. The company has put its machine learning in a separate division called TrueVue 360, which was only started in January last year, but it has already received a $1M order last year. It will not only develop algorithms for existing customers in the rail industry, but it also has a much broader and stand-alone purpose. At the end of last year, TrueVue 360 had already completed the development of 21 AI models or applications for the railcar inspection, consisting of over 30 algorithms, but due to duostech’s (its data-capture division) development of 30+ additional aspect modules, it will develop additional 40+ AI railcar inspection models and applications in H1 2020.

Q4 Results

Q4 results were really good:

  • Revenue increased 125% to $5.75M
  • Gross profit increased 176% to $3.15M with gross margin improving 1,000bps to 55%.
  • Operating expenses increased 28% to $2.52M.
  • Net income totaled $592K, the company’s first profit.

Not too much importance should be attached to that 125% revenue growth rate as quarterly figures can be quite lumpy and this was indeed partially a timing issue. For the year as a whole revenue rose by a much more modest 13% to $13.64M. During Q4 there were a number of new relevant projects and customers:

  • CSX (NASDAQ:CSX), implementing the first full-scale RIP center in the US (CN (NYSE:CNI) has four in operation in Canada already around Winnipeg, although these seem to supplement rather than completely replace manual inspections). In the next phase (this year), TrueVue 360 AI solutions are going to be added to this.
  • Another RIP center was installed in Mexico and was ready at the end of March this year.
  • There is a beta test of apis at a transit rail location in Chicago.
  • The company completed installation of an industrial portal for tank car inspection in Michigan.
  • Installed the centraco platform designed to provide additional security and logistics for a banking group.
  • And just on Thursday April 23, the company announced a new $1.4M contract for a RIP inspection portal.

So there is already evidence of the company venturing beyond railcar inspection.

Guidance and Covid-19 impact

The company guided for revenue of $20M for 2020 (+47%) based on contracts in backlog and near-term pending orders that are already performing or scheduled to be executed throughout the course of 2020. However, the fallout from the Covid-19 pandemic makes this a bit less sure, even if there isn’t really a tangible reason to reduce revenue guidance for the year at this point. However, revenues in the first half of the year could be substantially below those of H1 2019 due to potential delays in project execution resulting from the restrictive travel environment currently in place.

Margins

ChartData by YCharts

There was a 1,000bp gross margin improvement, and operational margin returned to positive territory in Q4, but these margins remain rather volatile mostly on the lumpiness of revenue.

Cash

ChartData by YCharts

With all the investments, it isn’t a surprise cash flow is negative and the company only had $56K on the books at the end of Q4. But after uplisting to the Nasdaq, a 1-14 reverse stock split (to keep within Nasdaq listing rules) and a public offering of 1.54M shares at $6 in February, it has another $9.25M added to that which should easily see it out for 6-8 quarters.

ChartData by YCharts
 

Valuation

ChartData by YCharts

We don’t really see any valuation problems; in fact quite the opposite. Analysts expect earnings to turn around from a loss of -$0.56 this year to a profit of $0.22 in 2021.

Risks

Needless to say, this is a small company ($15M market cap) with just a few customers. While we do think CSX validates its technology and we are inclined to say these RIP centers look like a no-brainer for other rail companies compared to labor-intensive manual inspections, there isn’t a deluge in orders from these companies, and no guarantee that these will be coming.

There is also competition:

Conclusion

This company can be a real winner. There is likely to be more demand in the rail industry, given the savings its solutions bring, and especially if the sought-after regulatory changes to allow completely automatic inspections become implemented. But there are several other potential drivers:

  • The first signs of success of the venturing of its data capture towards other verticals.
  • Adding algorithms to its rail solutions for existing (and new) customers.
  • Building a stand-alone AI division to pursue a host of verticals.
  • Moving towards a more recurring revenue driven business model.
  • International expansion.

Much of this is rather tentative and hence speculative; it’s like having faith in the maxim, “build and they will come.” The company has built solutions, and some of these have gathered traction, most notably in the rail industry, but the real scope of what they can do and how big their TAM really is remains to be seen. We are inclined to be bullish on the company’s opportunities in the rail industry, with its data capture as well as its algorithms, much of which are still under development. It’s here where the advantages for customers seem most defined, and given the moderate size of the company, this market alone should offer it a decent living. The company’s capabilities and ambitions are clearly much wider than just the rail industry, so its TAM could be quite astounding relative to its size. And we see the logic here. The so-called resource-based view of the company, its capability to amass data, put it in an interface system and build algorithms to analyse them, seems a generic capability to us that has applicability in a wide array of sectors. It’s early days, and while there are reasons to assume that if it can be successful in the rail industry, there aren’t any compelling reasons to assume it can’t be successful in numerous other verticals.

Disclosure: I/we have no positions in any stocks mentioned, but may initiate a long position in DUOT over the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Show CommentsClose Comments

Leave a comment

News ORS © 2020. All Rights Reserved.