The BigPanda ScaleUp Journey: Human/AI Collaboration, Predictive Accuracy, and Scale Power in AIOps

The BigPanda ScaleUp Journey: Human/AI Collaboration, Predictive Accuracy, and Scale Power in AIOps

At the beginning of the COVID-19 pandemic, we anticipated a slow-down in IT-related spending. In reality, the opposite occurred. Companies massively expanded their digital offerings using the same IT staff they’d had pre-pandemic, even as the teams lost access to many of their existing tools while working from home. This acceleration put immense pressure on IT teams everywhere, resulting in messy incident management, outages, and a huge shortage of talent.

BigPanda, the leader in AIOps, jumped into action, using its powerful AI software to help businesses and IT professionals find the root causes of IT incidents and resolve them. BigPanda is changing the way we view AI/human collaboration in AIOps with explainability, delivers differentiated machine learning software that continues to improve as more customers use it, and has a unique scale position in the market that benefits customers and partners.

Human/AI Collaboration and Explainability

For as long as AI has been in our lexicon, there has been a concern about AI taking jobs away from humans— although today, with a labor shortage especially in the technology sector, we may actually need AI to come to the rescue. We may need to look to companies like BigPanda to demonstrate the power of AI to augment human talent and alleviate shortages and related inflationary pressures. AI, working together with humans to increase productivity, could help avoid the need to resort to the “Plan B” of intentionally slowing down the economy via macroeconomic policies.

As discussed in “Prediction Machines: The Simple Economies of Artificial Intelligence, some human skills are economic complements to machine learning systems and can increase in value as AIOps systems like BigPanda are widely adopted. AIOps is ultimately the most powerful when its ability to make accurate predictions is paired with high-quality human judgment about what to do with those predictions. Also, AI is often not able to understand causality as well as humans. While AI can handle some human tasks with extraordinary speed and accuracy, humans can in turn complement AI—creating a flywheel of productivity and improvement.

One key ingredient in human/AI collaboration is explainability. Many popular AI systems today function a bit like black boxes. They can make predictions but can’t explain why they’re making those predictions. It’s much harder for humans to work with predictions coming from an AI system if they don’t understand the “why” behind the prediction.  BigPanda introduced a capability called Open Box Machine Learning, which makes its AIOps solution more explainable—a huge differentiator. This capability engenders a new level of trust in the AI solution from all involved—customers, investors, and employees alike.

Increasing Predictive Accuracy

Both customers and investors want to see that AI businesses and their machine-learning systems continue to improve with more training and feedback data. As the AI system scales, it goes through a sequence of stages. First it just makes predictions and displays the information to humans. At some point, the system’s predictive accuracy gets good enough that it starts to make some decisions and take some actions on its own, and only shows exceptions or edge cases to humans. Finally, it hits a tipping point where it becomes so accurate that human involvement isn’t needed at all (think: an elevator determining where to stop before opening the door).

Marginal increases in prediction accuracy can yield significant increases in utility as machine learning systems pass certain precision thresholds. For example, going from 98% accuracy to 99.9% accuracy improves the error rate 20x—a vital, valuable improvement in contexts such as self-driving cars and automating IT incident resolution for a mission-critical application, where error tolerance is extremely low.

BigPanda is uniquely positioned to achieve these predictive accuracy improvements as a native cloud company. Cloud platforms enable the spin-up of huge amounts of storage and computing power, removing a significant resource constraint in scaling machine learning applications. Special purpose hardware and software accelerate computationally intensive tasks that are common in machine learning applications. And with the right privacy protections in place, training and feedback can fuel learning centrally, benefitting all customers.

Scale Power

BigPanda enjoys a few different scale advantages:

  1. The business has natural network effects, whereby the product improves as more customers and partners use it. More data means better predictions, and more usage means more user feedback, which also improves predictive accuracy (as discussed in the last section). As a platform, it attracts other vendors in its space—such as upstream IT monitoring vendors and downstream automation vendors—who want to build applications that work with it. These applications lend power to BigPanda while contributing to a community of continuous innovation.
  2. As one of the largest pure-play AIOps businesses, it enjoys tremendous economies of scale on the supply side, for example it can invest more in R&D than smaller players with less revenue and scale.
  3. BigPanda’s investors, such as Insight Partners, are patient, have very long-term time horizons, and are willing to invest today in achieving scale to generate returns many years into the future.
  4. With its rapid growth and fun, mission-oriented culture, BigPanda can attract and retain very high caliber talent – critical to build and market technology in a fast growing, large market like AIOps.

Improving productivity through more accurate and explainable machine learning, better human/AI collaboration, and scale will be essential for the future of AIOps.  We couldn’t be more excited to be part of the BigPanda ScaleUp journey.

– Lonne Jaffe, Managing Director at Insight Partners