- Programmers dislike writing documentation ☹️ But it’s necessary! After all, the “old-timers” still remember something they wrote. And new employees have to figure it all out from scratch in this mess. It takes up valuable time.
- This problem has worsened recently as turnover in companies has increased. And programmers typically spend only 1-2 years in companies on average. So, someone always has to get up to speed.
- However, getting programmers to document everything properly is probably practically impossible ☹️ But now we can already create tools for automatic documentation of code using AI!
- The market for programmers is so large that it can accommodate not just one similar tool. Therefore, it’s quite possible to create your own platform, taking this startup as an example to emulate:
Project Essence
MutableAI has decided to create tools that allow documenting workflow processes using AI.
One side of the problem is that for onboarding new employees, documentation on the workflow processes they need to follow is required. And the other side is that none of the employees like to create such documentation.
And they either don’t create it or do it poorly ☹️ And this significantly increases the time for onboarding new employees. Mutable AI’s goal is to reduce the time for onboarding new employees without requiring old employees to create documentation.
On the startup’s website, there is a separate section called “Products” – which suggests that there should be many such tools.
However, so far MutableAI has only created one such tool – Auto Wiki, which can automatically generate documentation for code from the GitHub repository.
A week ago, the startup released the second version of this tool. Auto Wiki, as the name suggests, creates documentation in a “wiki format,” similar to how the popular Wikipedia is created.
Auto Wiki can create not only textual documentation but also visual diagrams reflecting the structure of the documented code. Additionally, Auto Wiki automatically tracks changes in the code and updates the documentation automatically.
The startup claims that Auto Wiki can currently only generate documentation for software code, but it will “soon work with other sources of information” as well. However, it’s not very clear whether they are referring to other sources of information for documenting software code or whether Auto Wiki will learn to create documentation for something else.
A personal subscription to Auto Wiki costs $2 or $15 per month to maintain documentation for one GitHub repository up to date. The rates differ in the quality of the AI model used, similar to the difference in cost between GPT-3 and GPT-4.
A standard subscription for companies will cost $50 per user per month. If a company wants to have code and documentation on its own server and additional customization of the AI model to fit its needs, they need to contact the startup directly for pricing on such services.
Mutable AI went through Y Combinator in the winter of 2022 and received its first investments at that time. In the summer of 2023, it raised another round of investments, but they haven’t disclosed the size of it yet.
What’s interesting
Just a couple of days ago, I wrote about the startup Silo Team, which has developed a platform for onboarding/offboarding programmers. It’s still in the beta stage, operating by invitations, but has already raised $1.3 million in investments.
The essence of the platform is that it:
- “Forces” departing programmers to leave their work in order, including documentation.
- Helps new programmers go through all the necessary stages of getting up to speed so they can start working at full capacity more quickly.
Discussing this startup in our chat, subscribers noticed that getting employees to do something before leaving is very difficult. Especially if documentation hasn’t been properly maintained for months or years, it’s impossible to catch up and fix everything in the last days before departure.
As an alternative, a suggestion was made for managers to conduct weekly or monthly checks of the documentation being created. But nobody likes creating documentation so much that it’s difficult to enforce it at any time, not just before departure 😉
Moreover, creating documentation takes time, and managers have too many tasks that need to be programmed right now and urgently. So, the manager will prioritize controlling the creation of documentation only as long as the work tasks are completed on time 😉
Therefore, the only way to solve the problem is to dramatically reduce the time spent on creating documentation.
One of the methods suggested today is by MutableAI, which allows for automatic documentation creation. However, documentation created in this way will probably still require some manual control for its accuracy, but significantly less time will be spent on it than on writing documentation from scratch.
Another option was proposed by the startup Augmend, which I wrote about last summer. They have developed a tool that allows developers to quickly and easily create documentation for their code by recording their actions on the screen with voice explanations. This startup raised $2.2 million in its very first investment round.
The problem of speeding up the onboarding of programmers is becoming increasingly relevant because companies are experiencing higher staff turnover. Programmers now work in companies for 1-2 years, and the process of onboarding new employees is becoming an almost constant task – one that requires more efficient solutions, which the startups mentioned today are trying to offer.
Where to Go
Considering that the number of programmers has reached nearly 30 million, with many of them not fond of creating documentation, the market for tools for automatic code documentation is huge. So huge that it can accommodate not just one tool like today’s Auto Wiki.
So, the first possible direction of movement is the creation of similar tools for automatic code documentation.
At the same time, it’s worth noting that turnover in companies is increasing across all categories of employees. This means that the task of speeding up the onboarding process for new employees applies not only to programmers. And this acceleration is greatly aided by existing up-to-date documentation on the work processes that new employees must learn.
Hence the question arises – for what other specialties are tools needed for fast or automatic documentation creation? Such tools can be built on the principle of Augmend or Auto Wiki, or using some other technologies.
What other work processes are widespread, poorly documented, and frequently changing – for which documentation needs to be created and maintained up to date?
Find such a niche, create a simple and fast tool for it – and many companies will thank you. Not just with words, but with money 😉
About the Company
MutableAI
Website: mutable.ai
Latest round: undisclosed, 13.03.2023
Total investments: $500K+, rounds: 2