Thinking about the architecture
Firstly, I decided I wanted to use socket io, because its free and I have a bit of experience with it. And, I want the chat to work on web and mobile, so I created a new board in Standup and wrote a basic prompt:
Then in Standup, you click to assign the task to AI (note you need a Pro membership to use the AI feature):
After a short while, the result gets attached to the task, complete with code examples. I expected a brief overview but it gave me a lot more including code for setting up the client, server and both an Android app and iOS app!
That was more than expected but actually, I didn't really want to go through the trouble of creating both a iOS app in Swift and a Android app in Kotlin. So, I split out the mobile app creation into a separate task:
The result again, was pretty impressive, it created a whole project and listed the contents of each file:
I feel like I could use this as a good starting point to create the app. However, I wanted to learn a bit more before, I get started, for example, how would I actually launch the app on the Android and Apple App Store.
Here, you can see the spinning wheel as Standup does its thinking. I won't post the results here, but it took me through all the steps for publishing to the store including creating the build, App ID, generating certificates, filling in all the listing info on the Google Play Store and App Store Connect etc..
Likes and Dislikes
The thing that impressed me most was the length and detail of the responses. I rarely needed to re-phrase my prompts and everything was written out in clear steps.
The other point that makes it useful is that its all visible on the same Kanban board, so I don't need to go back searching back through bookmarks or previous saved prompts, its all right in front of me. The advantage of this aswell is that I can just get started right away, drag tasks into the Todo column, share the board with other collaborators, attach my own designs, there's no need to import anything. It feels natural when building something to start with a task management app, it maintains momentum between the ideas and build phase.
On the downside, although I was impressed with the code samples, I kind of wish it would attach them as files rather than have them inline. For example, when it creating the React native project, it would be nice if I could download the project as a zip or even just attach each individual file to the task.
The other point I found slightly annoying is that it doesn't have any context of the project I'm working on. Its easy to work around this by adding context to the prompt e.g. "Assuming a socket io server exists" but it would be good to be able to add some context somewhere maybe at the board or column level. For example, imagine I could at the project level say "I'm working on a chat app written in Nodejs using socket io" or maybe "my preferred frontend framework is React". If it could then take that context and include it in every prompt I make, it would feel a lot more fluid. Even better, if it could understand the history of tasks that have already been completed by AI.
Summary
As a big user of task management software for organising my projects, I find it more natural to use AI from inside the board itself rather than going to some external tool and then uploading the results to tasks. The technology exceeded my expectations although I'd like to see a bit more break down of the results rather than one long response. Integrations with other apps would be good to, for example, wouldn't be great if I could deploy the code than came back in the response. Anyway, I've submitted my feedback through the Report an Issue feature, so let's see what happens. Definitely worth checking out Standup.
Posts
