Editor’s note: This interview is part of a Q&A series with winners of the ORIGIN Innovation Awards 2020. Tookitaki is a winner in the category Startup Awards – Artificial Intelligence.
FinTech is one of the biggest drivers of growth in the region. According to the e-Conomy SEA 2020 by Google, Temasek, and Bain & Company, investments “surged to a record high of $1.7B in 2019, a 40 percent jump from 2018.” The frequency of e-wallet transactions also rose from an average of 18 percent pre-COVID-19 to 25 percent post-COVID-19. Meanwhile, cash transactions saw a decline from 48 percent pre-pandemic to 37 percent post-pandemic, according to the same report.
However, along with this sharp increase in digital transactions comes the potential for misuse–particularly the risk of money-laundering.
Tookitaki, a winner at the ORIGIN Innovation Awards 2020, is a global regulatory technology company providing an ecosystem of AI-based smart solutions that create sustainable compliance programs for the financial services industry. It provides end-to-end AML/CFT analytics, the Anti-Money Laundering Suite (AMLS), to detect, investigate and report financial crimes.
In this Q&A with TechNode Global, Abhishek Chatterjee, Founder & CEO of Tookitaki, sheds light on the risks of increased FinTech activity and how intelligent and inclusive RegTech will help ensure a safer environment for FinTech players, financial services providers, and end-users.
What are the three key industry challenges that Tookitaki is addressing?
Today’s transaction monitoring solutions fail to provide financial institutions with a comprehensive Anti Money-Laundering (AML) risk coverage. The solutions are fragmented and mostly rely on rules or are augmented by traditional machine learning approaches that are not enough to keep pace with changing customer behaviour due to increased digitalization and detect complex money laundering activities, especially via new payment methods such as peer to peer lending, cryptocurrencies and e-wallets.
The COVID-19 pandemic has elevated the money laundering risk as criminals are adapting their pitches to suit the situation and come up with many fraud schemes and money mule scams. These money laundering schemes are relatively new and existing AML solutions may not be able to identify them. The “Fear of the Unknown” prevails largely. At the same time, the current solutions generate ultra-high false positives, making the AML programs ineffective and inefficient with increasing cost.
The challenges across current transaction monitoring solutions are elaborated under three broad points:
- Static and granular rules-based approach
- They are oblivious of the holistic trend and network of money laundering activities as the focus is narrow and uni-dimensional. Even money launderers are aware of the rules and adjust their transactions accordingly to stay under the radar.
- They are not self-sustainable, i.e., require manual tuning which is expensive and considerably time-consuming. By the time a new rule reaches production, it becomes obsolete.
- Siloed AML programs with no or limited knowledge from peer banks
- Absence of a mechanism of sharing insights and patterns across banks, geographies (even in different business units within the same bank) leads to insufficient and ineffective coverage of AML risks globally.
- Even with relaxed regulations on data sharing, the bank’s adoption rate on sharing and management of AML policies and dynamics is limited.
- Traditional machine learning approaches built with inspiration from rules
- Traditional machine learning approaches do not capture real-time money laundering data and are working off information that might be outdated.
Thus, the resulting machine learning-based models become quickly obsolete when the rules change or rule-based systems change.
- Model development life cycle is costly from both time and resource perspective.
- The rate of shift in data is usually faster than the rate of change of existing machine learning-based models being deployed in production, thus models being ineffective in capturing suspicious behavior.
These challenges result in:
- Spiraling labor costs
- Huge alert backlogs and unreviewed alerts
- Hefty AML fines
How is Tookitaki addressing these challenges?
Incorporated in November 2014, Tookitaki is on a mission to fight the evils of money laundering by creating a technology ecosystem that gives equal footing to all financial institutions and stops bad actors together, creating safe and sustainable societies. To achieve this mission, Tookitaki has developed an advanced machine learning-powered Anti-Money Laundering/Combating the Financing of Terrorism (AML/CFT) analytics solution, titled Anti-Money Laundering Suite (AMLS), to detect, investigate and report money laundering activities.
Further, Tookitaki AMLS allows the money laundering patterns to be automatically shared across banks globally (irrespective of their size) to facilitate collective intelligence and outsmart the criminals by identifying the suspicious money trails buried deep inside the mountain of legitimate transactions. The approach is unique and a paradigm shift from today’s rules systems and AI applications, as it looks beyond the siloed design where AML programs are created with limited data to cater to specific products, customer types, location without any knowledge on suspicious patterns observed in peer banks.
Our solution offers the following benefits to the industry.
- Improved risk coverage by detecting complex money laundering cases, including new-age methods such as money mule accounts, cryptocurrencies and e-wallets.
- Enhanced process efficiency with accurate triaging of alerts into three different buckets–L1, L2 and L3–with L3 being the high-risk bucket.
- Faster alerts disposition with explainable, defensible, and transparent machine learning models.
- Faster business decision with around 70 percent reduction in manual work.
- Financial institutions across the globe have the ability to share their locally detected money laundering patterns to the central repository for the benefit of their peers across the globe.
- Pre-packaged money laundering indicators to handle complex consumer behaviour and so that users can kickstart machine learning-based AML engine in no time.
- Auto creation of machine learning models and evolving them through Champion Challenger, a process that allows different approaches to test operational decisions in production, thereby minimizing building and maintenance process.
What makes your solution unique?
Other machine learning approaches in the AML space are inspired by rules-based systems. They are heavily dependent on static rules logic which neither capture money laundering risk holistically nor remain valid for long. The resulting machine learning-based models become quickly obsolete when the rules change, or rule-based systems change.
Another problem with these AI/ML solutions is that model development life cycle is costly from both time and resource perspective. The rate of shift in data is usually faster than the rate of change of existing machine learning-based models being deployed in production which results in models being ineffective in capturing suspicious behavior.
In contrast, Tookitaki AMLS system has prepackaged money laundering indicators (aided by our vast typology repository) to handle complex consumer behavior and so that users can kickstart machine learning-based AML engine in no time. Our platform also allows auto-creation of machine learning models and evolving them through a Champion Challenger process thereby minimizing the building and maintenance process.
Announcing the winners for the ORIGIN Innovation Awards 2020
What are the emerging trends in your industry that will drive innovation?
According to the e-Conomy SEA 2020 report by Google, Temasek and Bain & Company, the frequency of e-Wallets transactions rose from an average of 18 percent pre-COVID-19 to 25 percent post-COVID-19. Meanwhile, cash transactions saw a decline from 48 percent pre-COVID-19 to 37 percent post-COVID-19. It is no surprise that criminals are attempting to capitalise on this COVID-19 induced spike in online transactions, with criminals being extremely resourceful and sophisticated in how they exploit security gaps online.
For regulatory technology (RegTech) companies, there has never been a greater need for us to work closely with banks and financial institutions to detect new and emerging financial crimes and mitigate compliance risk while at the same time, ensuring that banks and financial institutions are aligned to the regulatory compliance requirements that have been set to safeguard the clients.
Any newsworthy updates you can share about Tookitaki?
We recently announced our partnership with UOB to develop an AI solution after more than two years of rigorous validation and evaluation. The technology will help sieve through an average of more than 5,700 transaction alerts each month to flag cases that are more likely to be suspicious with an overall true positive prediction rate of 96 percent.
We are also working with global FinTech leader, Broadridge Financial Solutions, for our Reconciliation Suite. The module helps automate reconciliation, building and improving tasks with automatic matching scheme configuration. This process is done by using supervised money laundering models and continuously improving the process through artificial intelligence and machine learning, saving significant time and cost for firms rolling out and managing large volumes of reconciliations.
Besides our AMLS (Anti-money Laundering Suite) we offer a Reconciliation Suite to our partners, such as Broadridge, where we can generate 90 percent match rates for general matching cases on items with minimum human intervention through our proprietary supervised pattern detection approach. The solution is easy to integrate into the financial institution’s existing and future up/downstream systems.
Featured image credits: Unsplash