I have found this Asana AI function to be extremely unreliable - am I missing something? For example, I asked it about tasks completed by my team within the last week, and it pulled tasks that were three years old and attributed their completion to a team member who wasn’t even here at that time. It can be helpful when it gets it right, but it seems very prone to confusion. Just me?
@Emily_Rens
I believe that some inacurate answers happen when the context is too large. I believe that’s a typical LLM issue.
To get the best result, I would suggest to be specific on the scope, use @mentions (for example, instead of “completed by my team”, try to mention the name of your team); or ideally mention @users.
Remark: for such data use-case I would rather use the very useful “Advanced search”
A random example. of “Chat with Asana AI” that works great
I use it to create a formated recap of all tasks in a project to prepare a “all hands meeting”.
100% of the tasks are always taken into account, included all the details.
My prompt is very specific and I mention the project.
The outcome format is exactly what I need to create some slides.
My prompt
Create a summary with all non completed tasks in the project [ All Hands 2026]
The format should be in 4 chapters, based on the field “EOS Fonction (mono)”
Company, strategy, business (that you can call “ Visionary”
Sales
Operations
Systems
Marketing
Finance combined with HR/RH
Then, each chapter should contain 4 sections based on the field “All Hands format”
iDO humans to celebrate
(Past month) Key milestones
(Next month) Goals & Events
A few topics you might have missed
Each task should come with a brief summary of the task (if there is description)
This IS helpful to see, Arthur, thank you! I do prompt it more specifically than I illustrated in my above comment, but maybe not to the level of detail you have shown here. I will do some experimenting!