NLP Q&A: Tim Carmody, IPC11.25.2019 By Rob Daly Editor-at-Large
Often outshone by its sibling AI-discipline of machine learning, natural language processing has broadened its proof-of-concept and pilot footprints within the industry, but how should financial services firms approach the new technology?
Markets Media caught up with Tim Carmody, the chief technology officer at telephony and connectivity vendor IPC, to discuss whether the nascent technology is ready for prime time.
How would you describe the state of natural language processing (NLP)? Is it ready for deployment on Wall Street?
Natural Language Processing (NLP) technology is ready for prime-time deployment on Wall Street, but I also feel like we have only begun scratching the surface of use cases and applications. IPC’s voice platform solution announced in September 2019 is already in use by some of the largest financial institutions in the world, and we are expanding in areas like Dictation-as-a-Service.
Even though it’s early days, the advances we’ve seen over the last several years in machine learning to power NLP is allowing trading desks to realize faster execution, more efficient communications, and streamlined settlement and reporting processes right now, today.
Many say that AI will augment employee roles rather than replace them. Besides making existing workflows more efficient, has NLP introduced new workflows in the front-, middle-, and back-office?
Yes, AI can create new workflows and significantly broaden roles in middle and back offices to participate in these workflows more efficiently. Being able to leverage NLP on real-time voice transcriptions and chat can provide additional data that can quickly be integrated into middle-, and back-office functions.
Being able to visualize the data in real-time allows for more efficient processes across a wider group. For example, IPC’s newly launched NLP data visualization tool, Blotter and Dictation-as-a-Service, can convert traders’ voice and chat conversations into a price-data feed that can populate into user systems for execution, settling, risk management, and can potentially even be monetized. The implications of NLP are profound and extremely positive for augmenting or creating new workflows because it creates opportunities to increase business intelligence.
Should firms consider developing NLP expertise internally, or can they expect to adopt NLP-as-a-Service offerings soon?
We’re huge believers in the NLP-as-a-Service model, in large part because many of our customers have said that they believe it’s more efficient and enables a best-of-breed approach.
Trialing or integrating NLP and AI can also be challenging to introduce into a production environment. IPC has created Connexus Labs, which provides on-demand private network computing resources with secure access to IPC’s expansive Connexus ecosystem of content providers, exchanges, ISVs, OMS/EMS, and more than 6,500 customer locations to provide a testing ground for new technology or certification.
Addressing the demand for the SaaS model has been a recent, key area of focus for IPC — not only regarding our NLP offerings but in other areas as well.
How sticky are NLP technology stacks? Do firms face a significant lock-in situation when selecting what to implement?
Yes and no. There can be synergies with specific NLP implementations, and it’s wise to take advantage of that, but we don’t see any risk for lock-in more so than with any other technology, especially given the popularity of the SaaS model with NLP.
How should firms calculate their RoI from implementing NLP-based offerings? Is it more of a hard dollar or soft dollar return?
We feel it’s both. There are soft-dollar ROI advantages with efficiency gains from real-time speech-to-text transcriptions and capture of in-stream orders and quotes and the organizational flexibility of workforce freed from repetitive functions like transcribing voice logs, ticket population, or reconciliation.
There’s also hard-dollar ROI, such as with trading and compliance, where NLP can directly improve the bottom line through transaction cost analysis, cost-avoidance for fees and penalties, and surveillance in general where NLP provides great value around analytics capabilities by turning voice data into actionable and intelligent insights. NLP can benefit many aspects of the trade life cycle.
Technology has enhanced capabilities of surveilling larger and more disparate data sets.
With Eugene Kanevsky, James Redbourn, and Joanna Wong, CLSA
AI/ML on the buy-side trading desk is a long-term program with short- and mid-term deliverables.
Aiden uses the computational power of deep reinforcement learning to improve trading results and insights.
BondDroid’s AI-generated prices are integrated directly into LTX’s pre-trade analytical tools.