E171 ‘The Responsible AI Revolution’ with Inspeq AI’s Apoorva Kumar

Today’s guest is Apoorva Kumar, Co-Founder at CEO at Inspeq AI, who are a finalist in the Best Application of AI in a Startup category for the 2024 AI Awards. for their work in tackling Generative AI challenges by offering a comprehensive framework for evaluation, guardrailing and governance. 

In the episode, Apoorva describes the journey to create Inspeq AI in late 2022 alongside co-founder Ram. With backgrounds from big tech companies like Microsoft, Meta, LinkedIn and Amazon, both Apoorva and Ram identified a pressing need: the responsible harnessing of AI. Apoorva highlights that the inception of Inspeq AI was driven by the challenges and potential they witnessed in the field of Large Language Models (LLMs). These models, while powerful, are notoriously nondeterministic, raising concerns about reliability and safety.

The duo engaged with over 200 developers and data scientists to better understand these challenges, which led them to develop a developer platform focused on evaluating the outputs of LLM applications. However, their mission didn’t stop there. In response to a growing demand for a robust AI governance platform, Inspeq evolved by defining a new category in the industry: Responsible AI Ops.

Inspeq has positioned itself as a quintessential responsible AI layer, particularly for generative AI solutions. Apoorva explains their innovative approach of allowing various SaaS applications, such as AI bots and content generation agents, to build on top of their platform. This focus on responsible AI governance is crucial, especially within regulated industries where the adoption of AI solutions can be fraught with compliance and security concerns.

One real-world application that Apoorva highlights during the podcast was in the financial services sector, where generative AI can revolutionise processes like chargeback scenarios, drastically reducing processing times and costs. However, without a protective layer ensuring regulatory compliance, such innovations could face significant adoption barriers. Inspeq’s platform offers this layer of protection, ensuring safe AI deployments and compliance with stringent regulations.

When asked about unexpected challenges and straightforward wins, Apoorva shared that while leveraging LLM outputs with another LLM seemed simple, the complexity of biases and latencies became apparent. Inspeq tackled this by avoiding reliance on LLMs within their protection layer, distinguishing their platform as faster, cheaper and more accurate. This approach has been benchmarked and validated by industry experts.

The podcast also delves into Inspeq’s remarkable achievements and recognition. The company is not just a finalist for the AI Awards; it has also gained accolades such as being listed in Wired Magazine’s top 100 hottest European startups and featured at Web Summit. Their partnerships with prominent consulting firms and global deployments highlight the scalability and trust they have built in a short span.

The future looks bright for Inspeq AI as they continue to gain traction and recognition in the AI community. The company, with its strategic locations in Ireland, the UK and India, promotes a strong and innovative team committed to making Inspeq the next big phenomenon from Ireland. This is just the beginning for Inspeq AI as they strive for further innovation and responsible AI deployment on a global scale.