“Kushaal Devanahalli shares how his AI consultancy is revolutionizing industries with tailored strategies, predictive solutions, and cutting-edge innovation.”
In a recent interview, Kushaal Devanahalli, founder of Proximity Works, shared insights into how his company is helping businesses leverage artificial intelligence (AI) to create more efficient, innovative operations. Headquartered in San Francisco, Proximity Works focuses on AI consulting and Software as a Service (SaaS), working with some of the world’s largest companies like Disney+, ESPN, and NASDAQ. Their primary goal? To help companies develop and execute AI strategies in a cost-effective and predictable way.
See Video Interview: https://youtu.be/XfZP08oOn7Q
“AI is a very fast-moving technology landscape,” Devanahalli explained. “Companies are not well-equipped to be aware of all the changes happening in the AI space.” Proximity Works steps in by staying up-to-date with the latest in AI and machine learning, using that knowledge to craft tailored AI roadmaps for their clients. This process starts with a comprehensive overview of a company’s data landscape—what data they have, if they are already using machine learning (ML) solutions, and which business areas they aim to impact. By bringing this information together, Proximity Works can recommend the most relevant AI use cases and outline the resources needed to implement and maintain those solutions.
Devanahalli emphasized the importance of predictability when working with AI: “We bring a lot of predictability to companies and help them adopt AI faster.” This involves not only identifying areas for improvement but also setting clear timelines and costs for implementing AI solutions. For example, Proximity Works might advise a company to focus on three key use cases, estimating that it will take six months and a small team to achieve results. This approach, according to Devanahalli, reduces uncertainty and makes AI adoption more manageable.
One particularly innovative project Proximity Works is currently involved in is developing an AI-powered commentator for a major sports company. “We are helping build an AI commentator…which will help scale the talent of their best commentators for all kinds of matches,” Devanahalli shared. The AI system would allow top commentators, whose time and energy are limited, to be virtually present even at smaller, state-level matches. This is just one example of how AI can unlock new opportunities that might not have been previously considered by businesses.
When it comes to building AI solutions, ensuring accuracy and reliability is paramount. “Implementing an AI solution has a lot of stages,” Devanahalli noted. These stages include testing models, fine-tuning them to improve accuracy, and addressing any hallucinations or errors. Proximity Works also employs “AI guardrails” to protect the input and output of AI systems, ensuring that they remain reliable even after deployment.
However, as AI learns from the data it’s fed, there’s always a risk of bias. Devanahalli acknowledged this challenge: “If there is bias in, there’s going to be bias out.” He explained that controlling for bias is easier when companies build their own AI models, as they have more control over the data used for training. But when using open-source models, it’s essential to understand what data was used to train them. “It’s necessary to go under the hood and assess if there is an inherent bias already,” he said. Adding guardrails to these models can further mitigate risks of bias, especially in cost-effective AI implementations that rely on open-source technology.
Another issue that can arise in AI systems is what’s known as “model drift”—when an AI model’s performance deteriorates over time, moving away from its original accuracy. “You really can’t avoid model drift,” Devanahalli admitted, giving the example of GPT-4, an OpenAI model that users claimed had become less accurate by the end of 2023 due to unexpected behavior introduced by feedback mechanisms. To address this, Proximity Works uses techniques such as swapping out models with earlier, more reliable versions or retraining the model with clean data. However, retraining can be expensive, so companies must weigh the costs of maintaining AI models against their performance needs.
Looking to the future, Devanahalli believes that AI has the potential to transform industries, but it must be implemented thoughtfully and with robust oversight. “We help companies identify opportunities they haven’t thought about before,” he said. From automating processes to discovering new insights from data, AI can drive innovation in ways that were previously unimaginable.
As AI continues to evolve, Proximity Works is committed to staying at the forefront of this exciting field, helping businesses not only adopt AI but also maximize its potential. “AI is so fascinating. You guys are in a good line of business,” Henze Healy, the interviewer, remarked at the end of their conversation. And with companies like Proximity Works leading the way, it’s clear that AI will play an increasingly vital role in shaping the future of business.
