Low-Code and No-Code are already impacting software development and analytics, we describe what else is needed.
By Assaf Araki and Ben Lorica
Recent analyses of technical job postings in the US by Dice and CompTIA show demand for developers and software engineers growing rapidly year-over-year, from 2020 to 2021. Last year’s dip in the number of job postings has given way to double-digit growth, and job postings are slowly approaching levels seen prior to the pandemic. We’ve also seen a strong recovery In our own analysis of job postings in technology hubs in the US. For example, we found the number of job postings that mention “machine learning” and “python” were up more than 70% and 80% respectively, year-over-year (Aug/2020 compared to Aug 2021).
Demand for developers, software engineers, and technical talent is strong across industries and geographic regions. Numerous studies conclude that the supply of software talent cannot keep pace with companies’ needs, and many companies are dedicating more resources to attracting, recruiting, and training software engineers. To help meet the demand for technical talent, some companies have made a concerted push to create internal programs to train a broad pool of their own employees in skills such as coding and data analytics.
Another critical trend is specialization. As software becomes more important and software applications become more complex, technical specialization is evolving. In addition to general job categories like web, mobile, frontend, backend, platform, and full-stack engineers, within data and AI there are specialists like machine learning engineers, data engineers, and data scientists. Depending on the application, there are even specialists in language models, computer vision, search, recommenders, and speech technologies.
While new tools have come a long way towards narrowing the “ninja gap” – the difference between programmers who produce naively written code and those who create highly optimized code – developers still need to master many libraries and APIs. For example, someone building a machine learning application will need to use tools for data processing, model training and tuning, experiment management, and model deployment. So while developers can often take advantage of high-level interfaces that hide technical details like parallelism and distributed computing, horizontal scaling, and fault tolerance, they still have to master a multitude of frameworks.
We describe a new set of tools that aim to boost developer productivity and agility, while also expanding the pool of people who can help build software. These tools allow developers to abstract and automate stages of the application development lifecycle to streamline the delivery of a variety of solutions. More importantly, these tools have the potential to expand the developer community from 27M developers to include the estimated one billion analysts who already use visual tools like spreadsheets and BI tools.
The Advent of Low-Code and No-Code
With an eye towards accelerating time to market, how can companies broaden their technical talent pool and increase the productivity of their current developers? One common approach to upskilling is investing in training, which could include a mix of in-house courses, external trainers and instructors, and online learning platforms.
Beyond training and retraining existing staff, there are promising new technologies that help bridge the ninja gap and relieve talent shortages. We’ll list tools that accelerate software development and increase the productivity of developers. These Low-Code Development (LCD) tools are designed for fast, continuous, and test-and-learn delivery. There are LCD tools for most types of software development, including mobile, data engineering, and Machine Learning (ML). Broadly speaking, LCD tools come in the following forms:
- Visual programming tools allow developers to write software primarily using a GUI, accompanied by some coding.
- Declarative programming interfaces: A recent example is AutoML tools for building machine learning models. Developers describe a model (“classify”, “predict”, “cluster”) they want to build with their dataset, and the AutoML framework does the rest.
- Coding Assistants: These are tools that can generate code for many routine tasks, freeing up development time for more challenging projects. Recent examples include program synthesis tools like Autopandas, and sophisticated coding assistants built on top of large language models (GitHub Copilot, OpenAI Codex).
While LCD tools target developers and staff who do some coding, a new set of tools eliminates coding altogether (“No-Code”). No-Code tools have the potential to vastly expand the pool of people who can build software applications and products.
No-Code has become especially commonplace in website development. There are now many No-Code tools for building websites. For example, a person with no coding experience can launch a simple, no-frills WordPress website using several No Code tools. Because WordPress has a large ecosystem of plugins, advanced features (forms, payments, etc.) can be easily added without writing a single line of code.
Our previous post entitled AI and Automation meet BI lists tools that allow data analysts to create their own data pipelines and reports while reducing the need for IT support. Low-Code and No-Code tools are showing up in many different areas of data engineering and data science as well. There are now several Low-Code and No-Code tools for data integration, analytics and machine learning, and data management:
What Else Is Needed
Developer tools like IDEs, text editors and notebooks have been around for decades. Many of the tools described in this post rely on new workflows and interfaces, and some target new potential software developers. For Low-Code and No-Code tools to really take off, innovations in human-computer interaction and interface design will likely be needed, as well as progress in automation building blocks (language models, ML, and AI).
We are particularly excited and optimistic about the potential of Low-Code and No-Code tools for building data, machine learning, and AI applications. We see a future – akin to what exists now in website development – in which anyone can build simple-to-complex AI and ML applications, end-to-end. As mentioned, there are already Low-Code and No-Code building blocks that hint at this possibility.
As Low-Code and No-Code tools become more commonplace, and as the pool of developers expands to include analysts and domain experts, we need tools and processes that can foster collaboration across different job roles and organizations. This includes developers, analysts and domain experts within IT and non-IT. Tools alone aren’t enough to shorten development time and accelerate delivery. Teams need processes and a common language to enable people with different technical backgrounds to seamlessly collaborate and exchange ideas, so they can build,deploy and maintain applications together.
Companies will also need to manage risks that accompany the expansion of the software developer pool and the acceleration of software development project lifecycles. For example, if more employees can create applications, access data, and interact with IT systems, companies need to have the requisite DevOps and DataOps tools for control and security, monitoring, and incident response. For companies building machine learning AI applications, tools and processes for Responsible AI become even more critical.
Finally, as Sky computing and multi-cloud native applications become more prevalent, we expect many Low-Code and No-Code tools to use multi-cloud native backends and platforms. Ongoing progress towards both No-Code and Sky computing hint at a future when companies can substantially decrease complexity and increase accessibility across their software and IT infrastructure stack.
Related content: Other posts by Assaf Araki and Ben Lorica.
- An Enterprise Software Roadmap for Sky Computing
- What is DataOps?
- The Growing Importance of Metadata Management Systems
- AI and Automation meet BI
- Demystifying AI Infrastructure
- Software 2.0 takes shape
Assaf Araki is an investment manager at Intel Capital. His contributions to this post are his personal opinion and do not represent the opinion of the Intel Corporation. Intel Capital is an investor in DataRobot & Hypersonix. #IamIntel
Ben Lorica is co-chair of the Ray Summit and the NLP Summit, and principal at Gradient Flow.
Subscribe to the Gradient Flow Newsletter: