26 No/Low-code ML tools to check out
An assessment of no/low-code tools for the machine learning process by their complexity & coverage
_________________________________________________________________
PSA: It is a key value for us at AIxDESIGN to open-source our work and research. The forced paywalls here have led us to stop using Medium so while you can still read the article below, future writings & resources will be published on other platforms. Learn more at aixdesign.co or come hang with us on any of our other channels. Hope to see you there 👋
Together with Benjamin Flader and Harits Abdurrohman, we did a little AI x Design Community collaborative session to evaluate the state of No-code and Low-code ML tools. This article documents our process, the metrics that we assessed the tools on, and insights we were able to draw from our assessment.
The process of decluttering tools
Everything started with a very long list of no-code tools. We filtered out the ones that we believed were not relevant — for example, if it was a data analytics tool rather than a machine learning tool. Initially, although our focus was on no-code tools, but we also included low-code tools as well. In the end, we ended up with 26 tools altogether.
We defined no-code tools as tools that do not require any coding, but also don't require prior knowledge of programming or software development concepts.
Our next step was to discuss the assessment metrics. We chose coverage of ML process and tool simplicity. In two rounds, we split up to briefly research the tools on our list, and to make sure that the assessment is not based on a sole opinion, there were always at least 2 of us looking into each tool.
After browsing through tool features, documentation, tutorials, and company landing page, we came back together to discuss and assess the tools based on the two metrics we had decided upon. Evaluations of the tools were from first impressions, although we were all familiar with about half the tools and we’ve evaluated from personal experience.
We noted that some of the tools were more process-focused, while some were use-case focused. The ones that focus on the process helps users easily manage the machine learning process or parts of it. The ones that focus on use cases seem to be more "out of the box", allowing for people to quickly train and deploy a model for a specific use case. We classified the tools into these 3 categories of focus: process building, single use case, and multiple use cases.
Aside from tool simplicity, we also assessed how much Data Science knowledge a user would need to have in order to be able to navigate these tools and use it properly (without wanting to throw their computers out the window).
After mapping the tools in a quadrant chart, we analyzed the results of our assessment and grouped the quadrants as Power Tools, Accessibility Promotors, Expert Tools, and Pocket Tools.
Results & Presentation of assessment factors
- Coverage of ML process: From data wrangling to model training to deployment, how much of the machine learning process does the tool cover? If the tool is focused on a specific use case, how much of the process is behind the tool's "black box"?
- Tool simplicity: How simple is it for a user with minimal to no coding experience to use the tool? Are the results shown by the tool easy to interpret? How is UX/UI? Does it require download or is it plug-and-play?
- Tool focus: Is the tool designed for specific use case(s) or is it meant to aid the machine learning development process?
- Scale of Data Science knowledge needed: At which level of data science knowledge should the user have to be able to effectively use the tool?
Quadrant groups
- Power Tools: Tools found in this quadrant are mostly intended for data scientists and engineers (or those with a good understanding of data science concepts). These tools will often aid users in the ML development process, either by providing in-depth insights or helping to increase productivity.
- Accessibility Promotors: Tools here often aim to make AI accessible to the general public.
- Expert Tools: If a tool finds itself in this quadrant, then it would be considered underdeveloped for a no-code tool and would most likely be a tool used by industry experts to perform specific tasks in the ML process.
- Pocket Tools: These tools are designed with a very specific focus of certain parts of the ML process in mind.
Assessment analysis & Tool recommendations
There is no no-code ML tool that covers both the whole ML process from data collection to integration and is extremely simple to use. All tools have different benefits, but the one that most stood out to us was Obviously AI (that we are proudly announcing that we are hosting our future event with). As it is in an early stage, we'll have to keep an eye on it. We are looking forward to see how these tools will progress!
If you already have a specific use case in mind there are specific tools for one or several use cases available. Try out Teachable Machine for image or audio recognition. It is easy to use but lacks representation of the whole ML process, although it does come close. Similarly, try out Runway ML for image generation.
To get some ideas, check out our other articles in which some of these tools are used:
Lastly, while there are a lot of tools trying to be the tool for everything (like H2O), the market for no-code ML tools is still young and does not provide many options for specific phases of the ML process. There is lots of room here for the development of tools intended for testing, data preparation, or deployment and integration of ML models in other tools.
Overview of assessed tools
- AI Builder from Microsoft
- AutoML Tables from Google
- Azure AutomatedML from Microsoft
- BigML
- Create ML from Apple
- DataRobot
- Fiddler
- H2O
- Lobe
- MakeML
- Metaranx
- mljar
- Obviously AI
- Orange
- PerceptiLabs
- RapidMiner
- Weka
- What-If Tool
Use-case specific tools
- Chatfuel for Facebook Messenger (chatbots)
- Dialogflow from Google (conversational AI)
- Cognigy (conversational AI)
- textengine.io (language generation)
- Playform (image generation)
- RunwayML (image and text generation)
- Teachable Machine from Google (recognition & classification)
- Wekinator (recognition & classification)
Huge thanks to Benjamin Flader and Harits Abdurrohman for helping with the assessing the tools and contributing to this article, and to Matthijs for contributing to the list of tools.
About AIxDesign
AIxDesign is a place to unite practitioners and evolve practices at the intersection of AI/ML and design. We are currently organizing monthly virtual events, sharing content, exploring collaborative projects, and developing fruitful partnerships.
To stay in the loop, follow us on Instagram, Linkedin, or subscribe to our monthly newsletter to capture it all in your inbox. You can now also find us at aixdesign.co.
