Start-up of the week: ThingBook

15 May 2017

Roman Ferrando, founder and CEO of ThingBook. Image: Andres Poveda

Our start-up of the week is ThingBook, which uses AI to help internet of things and telecoms companies stay secure.

“With the buzz around big data and analytics technologies (and despite the investment), many internet of things (IoT) and telecom companies struggle to extract actionable insights from their equipment data,” explained Roman Ferrando, founder and CEO of ThingBook.

“In this sense, the primary purpose of ThingBook is to allow data-driven decision-making processes, reducing the time, cost and uncertainty in the adoption of advanced data analytics solutions.

‘The analytics market is always evolving; agility is a key success factor’
– ROMAN FERRANDO

ThingBook connects clients to their equipment telemetry data by applying artificial intelligence (AI).

“We enable them to predict, analyse, share and act upon machine behaviours, optimising operations, detecting security threats and avoiding unplanned downtime, in real time, and processing thousands of sensors simultaneously.”

In collaboration with the Institute of Technology Blanchardstown – which supported ThingBook in its development stage through the New Frontiers Programme at the LINC and also through the School of Informatics and Engineering – the company is set to launch a product that predicts how machines will work, identifying unusual patterns and cyberattacks, and blocking unusual activity without human intervention.

Ferrando said that ThingBook is the first such platform to be able to store, compare, analyse and predict thousands of equipment behaviours at the same time.

It is built using the latest streaming technology and patented machine learning algorithms, which give the company the unique capability of processing up to 1trn messages per day in real time.

He said that this technology can reduce equipment downtime by up to 50pc and capital expenditure by 30pc by extending equipment lifetime.

The market

“The IoT analytics market is rapidly gaining traction primarily due to the sudden outburst of data from IoT-enabled devices and increasing global penetration of connected devices,” Ferrando explained.

“The IoT analytics tools have a unique role in various industry verticals such as manufacturing, telecom, energy and utilities, retail, and transportation and logistics. The underlying reason for the growth is the increasing number of IoT-enabled connected devices and sensors, which release a large amount of heterogeneous data.

“As the worldwide IoT market looks set to increase from 6.4bn devices in 2016 to 20.8bn devices by 2020, representing a growth from $502.1m in 2016 to $1.9bn by 2020, limitations of legacy analytics tools show clearly that new data analytics will be required.

“The analytics market is always evolving; agility is a key success factor.”

Ferrando said that ThingBook has been created to help companies with the problem of analysing high-speed generated data, focusing on telecom and industrial IoT segments.

The company’s main customers are industrial IoT platforms and gateway providers, SCADA system providers, telecoms operators and network operational support system providers.

The founder

Ferrando has broad experience in both research and product management within some of the world’s leading tech companies including Ericsson Research, Oracle Labs, and Amdocs.

He holds a PhD in machine learning applied to unbounded data flows.

“I also authored a book on the role of data analytics in the cloud and next-generation telecom networks, published by Wiley and Sons in 2013. I have more than 10 years of experience as a product manager, creating and launching … market network analytics products.”

The technology

ThingBook analyses large volumes of data, which are in turn processed in real time to perform data harvesting and mapping, data profiling and in-depth analysis, as well as provide visualisations for human consumption and integration with external systems.

“Our product reads from many data sources [such] as IoT gateways or network probes, and supports several industry-standard protocols [such] as MQTT, WebSockets or Profinet, among others.

“We implemented Flume and Apache Kafka technologies to ingest massive amounts of data in real time. We also use Apache Spark to perform our proprietary stream machine learning algorithms in a distributed way.

“Our patented AI implementation allow us to extract sufficient information to build predictive models online and detect anomalies, correlate KPI behaviours, discover new patterns and perform root-cause analysis.

“All this descriptive, predictive and prescriptive information qualifies the maintenance and security teams to make informed decisions in real time.”

Ferrando said that this real-time capability allows both humans and expert systems to take corrective measures based on ThingBook predictions.

“All ThingBook actionable analytics applications are built on proven experience of ThingBook in cellular networks, AI and streaming technology.

AI authority

Ferrando said that the ultimate goal for ThingBook is to build a reputation as an agile provider of highly accurate machine behaviour prediction delivered from the cloud and on-premises.

“By 2020, ThingBook will have gained significant customer wins across multiple verticals and be recognised for its unique IoT behaviour prediction.

“Our clients will trust ThingBook, viewing it as an agile partner and innovator, and will form an integral part of their business planning and decision-making.”

Always keep closing

Ferrando said that ThingBook is making good progress.

“We are very happy with how things are evolving in particular areas of the company beyond R&D. At this point, it is key for us to consolidate our product offering with the customers that gave us the opportunity to validate the technology.

“The next step will be to extend our message and build a reputation as a reliable, cost-effective partner for analytics in the IoT space.

“We are in conversation with several investors in Ireland and the UK, and I am confident we can close the deal in the short term.”

Beating the hype

Ferrando said the key right now is to establish credibility and stand out from the crowd.

“I think terms such as ‘big data’ or ‘data analytics’ have been overused over the last five to eight years, and the expectations were not always managed realistically.

Ferrando said that this increased the skepticism of both investors and customers, making things more complicated for other AI start-ups.

“The truth is that the commercialisation of analytics products brings additional challenges compared to traditional commodity products.

“The value of analytics algorithms can only be obtained in tandem with a broad knowledge of the application domain.

“Additionally, it is nearly impossible to advance the result and, therefore, the benefit for the customer before the data has been analysed.

“That increases the uncertainty in the decision of investing in AI projects; the risk of creating unrealistic expectations is high.”

On the other hand, however, deciding not to invest could mean the loss of incredibly profitable opportunities.

“We always prefer to talk about AI as a technique to solve specific industry problems. We know the customer challenges and the algorithm possibilities. That is one of the key differentiators of ThingBook: we cannot do everything, but we know what we do.”

Stop dreaming of Silicon Valley

Ferrando’s one-word view on the start-up scene in Ireland: “Noisy.”

He elaborated: “There are probably more tech start-ups now than ever before, that’s a fact.

“The affordable price of the internet and high-speed broadband are creating the conditions for digitalising almost any single economic activity.

“We can buy shoes and clothes online, watch movies on subscriptions basis and even find a date online, and all goes through the same wire – with or without cable. All those activities are supported by relatively new companies, start-ups not many years ago.

“Alternatively, the eruption of new collaborative models like Uber or Airbnb create new possibilities for applying that model to other areas. I love all the enthusiasm, energy, passion, dynamism and agility that makes the start-up community a great place to be. Despite that, let’s face it, a good number of those start-ups will die in the first year of life.

“This is because nowadays, we have an exponential increase in the number of start-ups. It is harder than ever to distinguish your company from others and therefore, succeed.

“Some authors call this situation ‘the start-up bubble’. It is true that dealing with such a dynamic environment makes it easier to get distracted and lose the focus on your target market. It is also true that getting investment in the early days without commercial traction is highly unusual and, even if the founders manage to do it, they are forced to give up a good portion of the company.

“Despite that evidence, many starters desperately look for investors before talking to any customer, which is, I feel, a common mistake.”

The war for talent

Starting a business is an incredibly demanding activity, said Ferrando. “In my opinion, this ‘bubble’ in the ICT industry explains the desire for innovation of many founders and the struggle of large enterprises to innovate beyond the fridge full of beers and the tennis table.

“Currently, there is a war for the talent. Employees’ motivation changed drastically over the last 15 years. Job security and good salaries are no longer the key motivational factors for the talented technologist. They need to feel identified with the project. I think the search for innovation will continue fuelling the start-up community. Luckily enough for us, Dublin is one of the hottest places for starting in the world.

“Any entrepreneur listens to many voices along the way in the form of mentoring, advising, incubators, market researchers; and that represents a bubble around the entrepreneurship, and not a bubble of entrepreneurs.

“I think two voices should be on top of the decision process for any starter: customers and your inner view. Anything else is important, but not key.”

His advice to fellow founders: “Stop dreaming about Silicon Valley. Europe is a great market with many possibilities.

“We are very lucky, and I sincerely wish it lasts for many years, given the uncertainty and turbulence generated by Brexit. Ireland is one of the most formidable concentrations of high-tech professionals with a large support for starters [such] as Enterprise Ireland and the NDRC.

“Germany is leading globally in terms of Industry 4.0, and the smart factory and Scandinavia has a long tradition in the telecom industry with companies like Ericsson and Nokia.

“Mediterranean countries like Spain or Italy have a super-strong hospitality sector, enthusiastic to incorporate more technology to increase efficiency and enhance the quality of experience for their customers.

“My main advice is: validation, validation, validation.

“As starters, we need to know who our customer is and why. Focus[ing] too much on the tech is great, but it doesn’t create a viable company. It is vital for us to understand our market and systematically confront our ideas and hypothesise with our potential customers.”

John Kennedy is a journalist who served as editor of Silicon Republic for 17 years

editorial@siliconrepublic.com