Key Takeaways
- Microsoft is investing big in the low code space and has put together a collection of products that is hard for other companies to match, capped recently by the announcement of PowerFX.
- The target in their sights is the Robotic Process Automation (RPA) companies such as UIPath, Automation Anywhere and Blue Prism who are closing big deals with big enterprises.
- The moat protecting the market share of the big RPA companies is created by the mature deployment systems that enable large enterprises to run hundreds or thousands of automated processes
- Microsoft’s recent low code announcements indicate that, instead of building a bespoke deployment system, Microsoft’s approach will be to incorporate low code software into an enterprise’s existing CI/CD workflows
- This approach befriends rather than alienates enterprise CIOs which will shorten sales cycles and improve deal flow.
- Further bad news for RPA companies is that AWS and Google have products equivalent to Microsoft’s collection in almost every category and can close the gap quickly.
- With billions of dollars in revenue on the line this will be one of the most exciting tech races to watch over the next few years.
Why does low code matter?
Low code refers to software development platforms that enable non-professional developers to build useful business applications.
The rationale behind low-code development is that business users who deeply understand a business process can contribute to building apps. And that this will enable companies to innovate faster than their competitors by leveraging the business expertise of business users and the development expertise of the IT teams.
Gartner predicts that by 2025 most enterprise apps will be developed in part using low code software. (You can view a sortable list of low code apps at this link.)
Low code software components
An enterprise low code software suite typically has four core components.
- Data storage
- Integration and transformation pipelines
- App interface
- Reporting
And two optional components:
- RPA (Robotic Process Automation) for interacting with legacy applications
- Data extraction (Extracting semi-structured data from PDFs)
In addition to these components, a low code software suite needs two additional capabilities:
- A way to deploy applications, and
- A way to monitor them after deployment.
Over the past 6 years, Microsoft has been building out the first six components and now it is tackling the deployment and monitoring capabilities.
Here is the tweet that terrifies Microsoft’s RPA and low-code competitors.
Hat tip to Gus Bekdash for linking this tweet to Microsoft’s push into low code and RPA.
A brief history of RPA and enterprise low code software
There are plenty of ways of characterising the history of low code software in the enterprise. If you wanted to, for example, you could go back to 4GL software in the 1990s. But for the purposes of this article, we’ll start our history from the early 2010s with the rise of Robotic Process Automation (RPA software).
In its early days, RPA software was only slightly more advanced than the screen-scraping systems from the days of green-screen terminals. But RPA software very quickly started to build out the other components of today’s low-code suite (data storage, integration / transformation, app interfaces and reporting). For example, many processes require human input or decisions at certain stages in the process (such as approving a payment) and so the RPA suites began building out the ability to create simple apps that allowed users to interact with the automated workflows.
The big innovation in RPA suites was their orchestration capability. This allows administrators to deploy and monitor hundreds or thousands of processes. When you hear someone state that an automation platform such as Microsoft’s Power Automate platform isn’t as mature as one of the big RPA tools such as UIPath, Automation Anywhere or Blue Prism, they are saying that Microsoft Power Automate’s monitoring and deployment are not as mature as the big three RPA tools.
It is this area that Microsoft is aimed at improving.
Microsoft’s journey toward becoming a low-code powerhouse
Microsoft has assembled all of the pieces required by an enterprise to deliver low code solutions. If they execute well on this strategy they are poised to become unassailable in the low-code world.
When Microsoft talks about low code, they have a pretty expansive view. The language they use when describing low code encompasses everything from an accountant writing a formula in Excel, to a software engineer using a pre-built connector to pull data from an API, to a consulting firm building a bespoke end-to-end claims management solution for a customer.
Microsoft realises that the real challenge with scaling low code is not writing low code applications - it’s deploying and monitoring low code applications. And it is firmly on a trajectory to solving this challenge.
Over the past 6 years, Microsoft has built or acquired the following components:
- Data storage solution (Dataverse)
- Integration and transformation pipelines (Logic apps)
- App maker (Power Apps),
- Reporting (Power BI)
They also have:
- RPA solution (Power Automate)
- Data extraction (Azure Cognitive Services Form Recognizer)
The missing pieces to the puzzle have been deployment and monitoring.
The big RPA suites solve these challenges by requiring customers to adopt their orchestrator systems.
But MS is taking a different approach. Microsoft is saying to development teams: "Don't set up a new deployment and governance practice. Just use your existing CI/CD practices." With the recent announcement of PowerFX, Microsoft is attempting to link all of these components using a common programming language that allows enterprises to deploy low-code apps using their existing CI/CD processes and governance framework.
This will give Microsoft a significant advantage over their low-code competitors because it will make Microsoft’s low code solution the safe choice for CIOs. When an enterprise chooses which low-code platforms they will allow to interact with their systems they’ll have a choice between using the Power Platform that fits with their current governance framework or using something else that does not. Many enterprises will just go with the Power Platform.
Microsoft has put together a pretty impressive strategy. I don’t know how much is by design and how much by tactical zigging and zagging but, judging by the dates that the company released each of the pieces in this strategy, it looks like sometime in 2019 someone at Microsoft had a lightbulb moment about how all this should fit together, and they’ve been executing against that strategy ever since.
Here's what the journey looked like:
- 2015: Releases VS Code
- 2016: Releases PowerApps and CDS (now Dataverse)
- 2018: Acquires GitHub, and GitHub Actions released
- 2020: Acquires Softomotive RPA
- 2021: Announces PowerFX programming language
The journey combines continuous integration / continuous delivery (CI/CD) components used in professional software development with low-code and RPA products, and ties them together with a common programming language (PowerFX) that facilitates good governance of data and applications. A killer strategy!
What is the PowerFX coding language and why is code so important in the low-code movement?
PowerFX is the programming language in Power Apps. In April 2021, Microsoft’s Greg Lindhorst introduced PowerFX with the following animated GIF showing the difference between PowerFX and JavaScript.
PowerFX is designed to offer an Excel-like experience to its users.
The reason it is important for Microsoft to get its low-code developers writing code is so that low code apps fit within the enterprise’s existing workflows and governance framework. Below is what I see as the most important part of the Microsoft announcement. It describes how an enterprise’s low-code apps can be incorporated into their existing development workflow.
Microsoft is investing heavily in enabling business developers to get comfortable with code. An early indication of the direction they are heading is an AI-powered Power FX formula writer built on GPT-3.
Once business developers are using Power FX to build business applications, these applications can be enhanced by the professional developers in the enterprise and managed by IT in the same way as their other software projects. By combining business developers building low-code apps with professional developers enhancing and deploying these apps, an enterprise can quickly build and deploy sophisticated, robust bespoke applications that support the enterprise’s unique competitive advantages.
As an example, imagine an insurance company that specialises in insuring trucks and other heavy equipment for the mining industry. Using PowerFX and the Power Platform, business users can build an app designed specifically for their line of business; professional developers in the company can build a bespoke premium rating engine in a language such as Java or C++; and all the pieces can be maintained using the company’s standard CI/CD processes.
What are AWS and Google doing?
Microsoft has assembled all the pieces they need to succeed, and they have an effective distribution capability. This is bad news for the big RPA companies such as UIPath, Automation Anywhere and Blue Prism.
Worse news though is that AWS and Google could quickly also become big players in the low code space. Microsoft’s main advantage right now is that it is owning the narrative. Neither AWS or GCP (Google Cloud Platform) are talking about low code as a key competitive advantage for their customers - despite both having most of the pieces in place to take on Microsoft.
AWS, for example, has recently released their low code app platform Honeycode, but isn’t expounding on how it fits in with the other pieces they’ve put together. Last year Google acquired App Sheet as its low code offering, but has not convincingly set out how they’ll combine it with storage, workflow or CI/CD.
The best statement I’ve seen from Google detailing how all of their pieces can be put together to make a credible low code environment was, ironically, part of a press release announcing a partnership with Automation Anywhere, one of the top three RPA companies.
What do AWS and Google need to do to take on Microsoft in the low-code space?
AWS can put together a collection of components that rivals the microsoft stack:
- Data storage: AWS has lots of options here. The difference between their options and Dataverse is ease of use for non-developers. They are making a number of attempts to bridge this gap but no single approach seems to be rising to the top.
- Integration and transformation pipelines: AWS arguably has too many options here. They have Step Functions, Glue, App Flow and several other tools that can be used for this purpose. The thing they lack is a coherent position on what should be used and when.
- App maker: Honeycode
- Reporting: Quicksight
- RPA (AWS has no solution here but acquiring one of the smaller RPA companies is an option)
- Data extraction: Textract
And similarly with Google:
- Data storage: Like AWS, they have lots of options. Appsheet’s data storage will go a long way to handling many low code use cases
- Integration and transformation pipelines: Google’s offering is pretty confusing here. They have a partnership with Trifacta, an Apache Airflow offering called Cloud Composer and the recently launched Workflows. (Our recommendation: Google, just pick one.)
- App maker: Appsheet
- Reporting: Looker.
- RPA: Google’s partnership with Automation Anywhere may actually be problematic here. Whilst Google has no other way to screen scrape legacy applications, most of the other capabilities of Automation Anywhere (data storage, workflow, apps etc) have comparable products in Google’s suite. I foresee sales challenges here for Google and Automation Anywhere with pre-sales consultants pulling clients all over the place).
- Data extraction: Document AI
The missing pieces for AWS and Google
AWS and Google do not have much of a technical gap to cross to compete with Microsoft in the low-code space. Their two biggest challenges are:
- Setting out a clear strategy, and
- Closing enterprise deals
Unfortunately for them, Microsoft has a long history of executing well in both these areas so the odds look stacked against them. But they have the tech in place - they just need to arrange it into a coherent picture and start promoting it.
But once they do, the RPA companies will have three massive behemoth’s chasing them down. This will be bad for their margins but, in my view, good for CIOs and IT teams who will have a wide range of excellent automation solutions to choose from - and some of these will fit nicely with their existing deployment and governance frameworks.
The Low Code Road is a series of articles written for InfoQ by Doug Hudgeon, CEO of Managed Functions. You can read the first article in the series here.
About the Author
Doug Hudgeon is the co-author of the Manning book "Machine Learning for Business" which shows your users how to solve real-world business problems using AWS SageMaker. He is also the chief executive officer of Managed Functions, an integration company specializing in helping enterprise low-code and RPA teams deliver projects faster by providing bespoke components to handle the thorny problems they may encounter on a project. Uniquely, the components can be deployed as serverless functions to the enterprise’s cloud (AWS, Azure or GCP) so the solution runs entirely on their infrastructure. You can find him on Twitter.