As AppSheet developers you’re fully aware of the extent to which Apps and automations built with the tool can become mission critical to an organisation. With processes from inventory management to invoicing, employee onboarding and asset management to name a few. But with this great power also comes responsibility, and that comes in the form of needing to properly document what you’ve built so that the organisation can maintain continuity if you leave the picture or it requires maintenance 18 months down the track and you yourself need a refresher on how it works. Documentation is more often than not an after thought because human beings have a natural propensity to put the present above the future, especially if it’s not going to affect them. We call this documentation inertia and it finally has the perfect solution in generative AI.

If we think about the impact on undocumented Apps it’s easy to celebrate the victory in the present when some horribly arcane manual process has been digitised or even automated but fast forward into the future when human resources move on or to another department and this can become a highly problematic situation. We’ve even had one senior IT administrator say that the technical debt experienced in this circumstance is more costly than the value the App created in the first place. If the business process itself needs to change, requiring changes to the application and no one is left on the team that understands how it works it will inevitably cost the organisation time and money. This can be avoided however if the App project is documented correctly and developers inheriting the App have the ability to extend or maintain an App with little disruption. We can empathise, it takes a lot of discipline to document a project with the required level of detail, especially if you are a citizen developer who isn’t practised in this art or have taken on the project over and above your normal work commitments. We’ve referred to it as TH;DW [too hard; didn’t write] which acknowledges the mammoth mental effort documentation requires. However that being said, given the potential adverse impacts of not doing it, it is very much a requirement.

By now most people in the technology community are aware of the massive advances in generative AI that happened in 2022 with Dall-e, Stable Diffusion and ChatGPT bursting onto the scene and making a very large impact that many believe is a historic turning point. We agree with this and have found that with the underlying capability of the Teamapps platform that we are uniquely able to utilise this ground breaking technology to solve this very pressing business problem in an amazing way. To explain, Teamapps via its Chrome extension has the ability to pull descriptive characteristics from an AppSheet App such as its name, function, data sources, table names, header names and many others which can then be used to prompt GPT-3 via its APIs to generate App project documentation including among other things natural language explanations of complex expressions. The results are literally extraordinary and accomplished in seconds as opposed to the hours or days that it would take to write them manually. It uses all of the same information the developer would use and generates documentation in sections like; Purpose, Description, Table explainers and their related slices, user actions, key slices that govern user access to data sources and other exciting features like user guides which are coming soon. It also means that it can sense when changes have been made to the App which warrant an update to its documentation. This is a truly breakthrough characteristic in itself given that documentation update inertia seems to be an even stronger phenomenon with the majority of documentation failing to be kept up to date.

Documentation generated with AI Assist is inherently connected to the AppSheet project and the creator and co-creators involved making it a true single source of truth and system of record that the development team can work collaboratively on to make perfect. Whilst also inherently being discoverable by developers that are added to the AppSheet project itself meaning that new developers added will automatically be made aware of the documentation via notifications.

It’s probably fair to say that problems like the insufficient documentation of NCLC Apps has held back citizen development using AppSheet from being the truly transformational force that it can be. These breakthroughs in generative AI have handed us the perfect solution, by giving every developer and their co-creators the ability to create and update highly valuable and accurate App documentation 100x faster and with vastly less effort than how it is currently. We’re excited to be part of the citizen development mission by bringing this amazing technology into the picture.

‍