Niraj Patel, CIO of multifamily lender Greystone, has a knack for bringing digital transformation to the financial services and real estate industries.

Greystone's Patel has led teams at DMI, IBM. Selex and GMAC before coming to Greystone. Through all of those CXO roles, Patel brought innovation, value and data lens to projects.

I caught up with Patel to talk about generative AI, vendor management and creating value with data. Here's a recap of our conversation.

CIO challenges. "Most challenges of CIOs aren't about technology, but adoption, change management and value creation," said Patel. "How do we bring that technology in and get productivity for the customer?"

Examples of value. Patel said CIOs need to know their industry and what really matters. For instance, in financial services and real estate success is a leverage game and that means interest rates are everything. "Technology is nice, but if we can arm sales folks with better information and insights they can get a better rate," said Patel.

Generative AI and AI. Patel is a veteran of using AI and noted that the technology has been around for a while. In 2012, Patel oversaw a project to use AI to predict energy efficiency. Later, AI was used for vision and natural language processing. "Today all of those pieces can be working together for generative AI," said Patel, who noted that streamlined forms and wider data sets can drive productivity.

One example would be property inspections where the inspector on scene captures forms, pictures and videos. AI can analyze unstructured data which can then be combined with the company's financial data and third-party data on the structure of the building as well as its content. "AI brings granularity to the inspection," said Patel.

Chatbots vs. generative AI. Greystone has invested in chatbot technology and robotics process automation. Now generative AI and technologies like ChatGPT can advance the ball. When asked about chatbot sprawl, Patel said it helps to put guardrails around the technology and the data sets used. "We think generative AI can accelerate chatbots to make them more individualistic and bring interactions to you specifically," said Patel. "Today chatbots are more generic horizontal ones, but they can become more vertical with the enterprise data it's accessing."

Large language models. When it came to LLMs, Patel said there's an art to using foundational models and tuning them with enterprise data. Using off-the-shelf models is about acquiring data from the outside world. "Off the shelf is data about the building and what's around," said Patel. "Financial data like underwriting and rents are my domain."

Integration. The various data sets in a commercial real estate transaction will require integration. For instance, property managers have one software platform and lenders have another. Then there's the decision on where to put the data, which will be partly on premises and cloud. Patel said mobile devices can also process data. "We'll probably want more processing on the edge," said Patel. "You also need to decide whether you want just actionable data or everything."

Vendor management. Patel said he prefers to have a set of vendors that can provide innovation and be manageable. Greystone is a Microsoft shop so leveraging Azure is a natural fit. However, Patel said he works with Google Cloud and AWS too as well as smaller vendors.

"Having all big vendors is a problem so we have smaller ones too. I break vendors up into small, medium, and large, but we don't do too many and all of them need a strong integration later," said Patel. "Smaller vendors bring speed and more innovative thinking. Smaller players are also more functional in nature."