Iris News September 2024
The Gen AI Promise & Challenges and Emerging Technology Trends
In this issue:
> Gen AI - potential, productionizing, and use cases
> Low-code – a viable path for critical applications
> Perspectives on data intelligence, asset tokenization
> Performance milestones and awards drive enthusiasm
Iris News September 2024: The Gen AI Promise & Challenges and Emerging Technology Trends
Thank you for reading
Iris Insider September 2024 The Promise & Challenges of Gen AI and New Applications
There has been a lot of enthusiasm around AI coding assistants and how they can help deliver better quality and productivity. Our teams are using coding assistants, such as GitHub CoPilot, in client contexts and believe that Gen AI’s promise in software engineering goes well beyond them.
Taking inspiration from CoPilot, we developed a VS Code IDE extension (called IDlite) to apply Gen AI to other parts of the software engineering lifecycle. As an exploration of the art of the possible, we have utilized standardized prompts, Large Language Models (LLMs), and orchestration to review, rate and summarize requirements, identify test scenarios, generate Gherkin/Playwright scripts, and review API specifications for OA Spec conformance, etc.
We have been sharing our experiences with clients and are seeing enthusiasm build around the potential.
Please refer to our latest perspective paper, How Generative AI Transforms Software Engineering, for more details on this topic.
Over the past several quarters, we have partnered with many of our clients on Gen AI Proofs of Concept (PoCs) and pilots around business use cases. With many cloud-based offerings and easily-deployed on-prem models and libraries, PoCs are relatively easy to achieve.
From our recent experience taking some of these PoCs to production, we realized that there is more to this iceberg below the waterline; productionizing these PoCs is much harder. In addition to coping with significantly increased volumes of documents, we also had to deal with security (e.g. SSO integration), scalability (LLM connectivity-sharing across user sessions while preserving and using user-specific context), cost controls (monitoring token usage and optimizing where feasible), capturing user feedback on results and utilizing it for fine-tuning, and many other such concerns.
LLMs and Generative AI can help with a variety of business and technology use-cases
In many early experiences in Gen AI, we applied its summarization capabilities to enable end-users to gain easier access to policy and compliance information codified in large documents. These initial successes have been across a range of business processes such as credit adjudication and product assurance support in the banking sector, compliance workflows in underwriting insurance policies, and API interfaces for other industries. Here are a few case studies of such successful Gen AI implementations.
Gen AI-powered conversational assistant strengthening AML product assurance
Summarizing complex commercial lending policies using Gen AI
Reduce crop insurance underwriting cycle time using Gen AI
How Gen AI is helping create user-friendly API chat interfaces
We have shared our experiences around key productionizing considerations and an illustrative architecture in our new perspective paper, Productionizing Generative AI Pilots
We see many business-managed applications (BMAs) developed and utilized across front-office, marketing, finance, risk and other functional groups at numerous client organizations. While they are much faster to develop or change, and often bring together internal and external data for critical analytics, reporting and decision-making, BMAs pose a variety of control, scalability and other challenges. Attempts to industrialize BMAs using pro-code approaches have been costly and lose agility in ongoing business-driven evolution.
Low-code platforms, especially Microsoft Power Platform, in our experience, offer a viable alternative in this regard. While low-code platforms were initially applied to developing simple workflows and very small applications, we see capabilities in platforms, such as the Microsoft Power ecosystem, that can be applied to industrialize critical BMAs cost-effectively and preserve agility of ongoing evolution.
We are working with a number of clients on this front and have outlined some of our learnings and approaches in our recent perspective paper, Industrializing Business-critical End-user Compute-based Applications using Low-code Platforms.
Organizations have collated large troves of information through digital initiatives over the past decade. They can now tap into those with data science engineering and Data & ML Ops to enable scaling of the intelligence part of the data monetization lifecycle. It has become easier to support multiple modeling and data science teams working on multiple problems and opportunities concurrently by using cloud technologies and services to effectively scale intelligence.
Perspective paper: Extracting Intelligence from Data at Scale and Speed
Asset tokenization, an innovative application of Distributed Ledger Technology, enables digitization of physical assets. This makes it easier to issue and trade in them, either in whole units or fractions. It also helps preserve the asset’s value and integrity, while ensuring secure, transparent transactions. As adoption grows, and a better integrated market landscape develops, asset tokenization can potentially transform the investment landscape.
Perspective paper: Asset Tokenization Can Transform Financial Markets
We are proud to share some recent customer feedback and positive accomplishments.
Top ratings from clients - In our last biannual customer survey, conducted in early 2024, we achieved an overall satisfaction rating of 6.5 out of 7 and net promoter score of 95% out of 100%, continuing our positive trend.
Key award won - Iris was named a Top 100 Best Place to Work (and scored higher than the IT industry on the Trust Index) by the Great Place to Work Institute (GPTWI) of India, our 3rd year in a row.
Iris leader honored - Anil Apte, Co-founder and Managing Director, received the prestigious Most Trusted Leader award from GPTWI, which cited our transparent and innovative culture.
Three Irisians were recognized among the Top 100 Great People Managers of India, our 4th straight year and 2nd in a row with three managers on the list, a major feat.
