Gearing Up for the ‘Intelligent Enterprise’
What will happen when all data becomes intelligent and computers have cognition?
By Paula Klein
The intelligent enterprise may finally be in sight. Fueled by a confluence of machine learning, big data analytics, and APIs, advances previously available only to consumers are now advancing into business environments.
This “new era of intelligent computing” and how enterprises can benefit, was the subject of Mark Gorenberg’s recent MIT IDE seminar. By combining huge volumes of data with learning algorithms, he said, “the software used in every existing enterprise line of business is being disrupted, and new industries that were previously data-starved are now open to optimization.”
Gorenberg, who has 26 years of venture capital experience and has funded and served on the boards of numerous successful start-ups, is now betting that the timing and technologies are ripe for AI to move into enterprise environments. He is currently a founder and Managing Director of Zetta Venture Partners, an early-stage fund focused on the intelligent enterprise. Zetta, founded last year, has invested in 21 companies so far including Kaggle, which was acquired by Google. It currently has $185 million under management. The company believes that in the current, fourth era of computation, huge volumes, or zettabytes, of software can “learn” from data. In other words, all data will become intelligent and computers will have cognition.
The year 2005 was a “tipping point” in terms of recommendation engines that allowed Amazon to take advantage of network effects to move beyond consumers and into the enterprise market.
While cloud services marked the most recent generation of computing, “now, it’s about the data,” he said. Businesses must have an AI playbook, and hire data scientists and machine learning experts to get in the game.
“For enterprises to compete, they will need to re-architect to include new cloud platforms, micro and data services, collaborative hubs, real-time business optimization dashboards, and new intelligent applications,” he said. And the best new applications, primarily being developed by startups, he said, will include a ‘virtuous loop’ — software that continuously transforms anonymous customer data, and public data “into machine learning algorithms to generate both cleaner data and insights.”
Gorenberg sees many new opportunities arising for vendors and enterprises alike. Enterprises will have to open their software services to allow for non-proprietary apps and new development tools.
The reward? “Everyone in the organization is a data analyst; even the CEO can check KPIs in real time on their phone.”
For investors, he said, “the playbook for startups is changing. It’s not just about apps anymore.” Startups are developing products that engage data collection and crowdsourcing of public and private data that can be applied in all industry sectors. For example, Marketing Evolution very specifically recommends to clients the best mix of media spending based on AI analysis of real-time customer patterns and behaviors. Another startup augments insurance claims assessors with deep learning images of auto parts damaged in an accident.
Zetta’s presentation points are detailed through a series of posts on https://medium.com/@Zetta . Highlighting a recent thesis being codified by Zetta Associate, Ivy Nguyen, Gorenberg summed up some key insights for both enterprises and entrepreneurs seeking gains from machine learning and data analytics, as follows:
1. Define your minimal algorithm performance to make it viable.
2. Race to get critical mass. Make sure you have the right data and analyze it quickly.
3. Choose the right business entry point.
4. Monitor for diminishing returns and keep updates flowing. Kill the old models of software development bottlenecks.
5. Maintain data rights to secure leadership. Data rights are the new intellectual property; contracts will be written around data ownership moving forward.
Watch a video of the MIT IDE presentation here.