How would AI (like in the movies) redefine it?
Google’s suite of AutoML tools, launched in January with AutoML Vision, is unleashing the power of machine learning to a large number of companies by allowing developers with little or no ML capability to use Google’s systems to automatically create machine learning models. The Vision product is already 10% better than any human created model and with tools planned for speech recognition and translation, it’s worth asking the question: “Could an AutoML for human office worker collaboration improve worker collaboration by 20%-40%? Having been a desk jockey for nearly 20 years, my guess is a resounding YES.
Is there really that much room for improvement?
Over the past decade, the output for the average software developer has grown by 10x due to services like AWS, Open Source components like Git and wide adoption of methodologies like Agile. In that same time, it is doubtful that the productivity of business workers in Sales, Marketing, Support, Legal and Finance has grown at the same rate. Office work is often much more qualitative than more binary “hello world” software development, but the productivity deltas between software and business workers is still significant. Software people think “I’ve had to do this twice, I’ll have to do it again, how can I automate it?” They are also keenly aware of the dangers of context switching and set up their time to avoid task shifting. Business work jobs aren’t structured the same way and unnecessary emails, unstructured meetings, repetitive reports and bloated PowerPoint presentations. These current approaches to business work clogs up companies productivity. If AutoML had the powers of AI in the movies to free humans from our generation’s version of Elevator Operator work, what sort of products would it build?
What would AWS do?
The first place to start is to examine what Amazon’s Amazon Web Services (AWS) group has done to automate and accelerate the way software is built. AWS has eliminated massive amounts of work for developers by turning nearly every component of a software app into a service that can be turned on/off like a utility. The concept has proliferated to them offering 123 distinct products that can be turned on with a checkbox and the same credit card. Then AWS connects data from each of these services with a consistent set of Application Programming Interfaces (APIs). This image from the AWS re:Invent conference shows how to architect these types of systems for developers.
*AWS CTO on stage AWS re:Invent
The aim is to make complex architecture diagrams like the one in this image remedial for any developer to put together. So when the Services and APIs are this easy to string together, “All the code you ever write is business logic” (Amazon CTO Werner Vogels.)
Defining the future of business jobs
Vogel’s future might not arise immediately for business workers, but the products that are built on this construct of Services, APIs and Business Logic generators are the ones that will be tomorrow’s most successful companies. Seeing this perspective, my guess is that AutoML would clone the AWS approach to development and spawn a whole new crop of companies to automate and accelerate repeatable work. To get the same end deliverables of reports, strategies and support, this approach can grind out the meetings, the busy work and the politics. A few smart business logicians will be able to eliminate the work that dozens of humans are needed to produce today. The before and after will look something like this:
Some, like Naval Ravikant claim that we will see a single founder $1Billion company. While that might not happen for some time, there are a whole range of companies that are already adopting these types of approaches to drive business teams to the potential 10x gains. Many are still nascent or show up as small incremental changes, but they are inevitable.
Applications become services
A new crop of services companies is evolving**.** On the surface, they may look like previous generation of applications, but their impact in the industry could be greatly different. They are designed in ways where they enable significant productivity gains. For example, Figma is a SaaS tool for designers. It benefits from Open Source methodologies for templating and versioning design projects. When design guidelines are populated into Figma, it enables designers to speed up their workflow and much faster collaborate with developers. MixMax is a tool for sales teams that automates appointment setting. Sales emails are treated like developers treat code snippets and test their accuracy like web developers test conversion rates. Lyrebird automates salespeople’s voices so that they can automate the intro part to every call. For paralegals, Casetext condenses weeks of human work into minutes by querying the relevant case law to the specific brief in minutes. For writers, GitBook gives Github workflow review process for writers in order to work through revisions. For general business work, Notion and Coda provide anyone the power of a developer to create and publish collaborative work. Finally, for all business users SuperHuman has re-envisioned the email client with more contextual knowledge and relevant suggestions to make each email session twice as productive. The list of companies that can turn each discrete work project into a service that can boost production is growing each day and enabling future work to get completed with less labor.
API’s deliver the context
All of the data being generated from these services can become even more impactful for businesses when it has an Application Programming Interface (API). As AutoML would look to accelerate office work further, it would look to a number of companies that connect this data together and begin to process the work further. One of the biggest killers in business team productivity is context switching some go as far as to call it harmful for the brain. Moving between apps breaks flow so much that APIs will be critical work in the future. For example, Segment is a company that connects together 200+ enterprise applications to perform tasks like personalized customer tracking. Slack is a collaboration product with a rich API that has enabled hundreds of Bots to update employee collaboration with the right info at the right time. Zapier is a business tool that has combined together integrations between over 1,000 applications. Its extended features of starting workflows in one app and completing that action in another app blurs over to the final key topic for AutoML’s automation algorithms: business logic companies.
Business logic governs what services deliver
While the services can boost productivity, they are just tools and while APIs can deliver amazing contextual knowledge, they still need rules to define what they do. Business Logic companies are the missing link that Werner Vogels alludes to. He envisions a world where services are so automated and APIs are so aligned that the last bit of software programming is writing Business Logic. Today’s current crop of Business Logic companies have often taken a long time to establish, but once they reach critical mass, they can become entrenched and dominant. A quick glance at public software markets shows that products with major workflow components have grown to become the biggest SaaS companies: Salesforce, Workday, ServiceNow and Atlassian. While these companies now have many products, their core products that have built their franchises are built around managing workflows.
In addition to these leaders, AutoML would likely lean heavily on a relatively new company called UIPath, who’s mission is “to eradicate repetitive tasks through intelligent use of software automation.” Whenever the human algorithms that are built up get so understandable that they can be designed in a business process, they can train robots to complete the job.
Will your company win in this paradgm?
It is my conviction that there is huge upside for companies who architect their products like the earlier examples. As I kick off my Entrepreneur-in-Residence at Kleiner Perkins, my aim is to find like minded entrepreneurs who see the world from this perspective and to work with them to build the world’s greatest new companies. If this is your passion too, reach out.
About the author: Daniel Freeman has held Product, Marketing and Sales leadership roles at leading enterprise software companies like, Atlassian, Cloudflare, Sumo Logic and Symantec.