AI Status Update VS AI smart Summary: Output difference with same prompt

English Forum Ask the Community
1

AI Status Update is Returning Incomplete and Incorrect Task Data Compared to Smart Chat

I’m currently using AI Status Update within my project to generate a structured report of all incomplete tasks. My objective is to produce a table containing specific custom field values for every incomplete task in the project so that the Status Update can be shared directly with stakeholders without requiring any manual editing.

However, I’ve noticed that the output generated by AI Status Update is significantly different from the output generated by Smart Chat, even when both are given the exact same prompt.

Issue 1 – AI Status Update only returns a subset of tasks

My project currently contains approximately 37 incomplete tasks spread across multiple sections. The prompt explicitly instructs AI to include only incomplete tasks and generate a single table containing every incomplete task in the project.

Despite this, AI Status Update consistently returns only a small subset of the available tasks (typically around 8–10 tasks). The remaining incomplete tasks are omitted without any indication that the response has been truncated or summarized.

When I use the identical prompt in Smart Chat, it successfully returns all incomplete tasks in a single structured table with no missing records.

This makes me wonder whether AI Status Update is intentionally designed to summarize project information rather than enumerate every matching task, even when the prompt explicitly requests a complete report.

Issue 2 – Custom field values are missing or mapped incorrectly

Another issue I’m experiencing is that several custom field values are returned as “—” or “Not specified”, even though those fields are clearly populated within the project and are visible in the List view.

For example, my prompt requests the following mappings:

  • Subject PIC → Assignee custom field

  • Country → Country (VR1) custom field

  • Open Task → Open Task custom field

  • Priority → Priority custom field

  • Info. Report Status → Section name where the task exists

In many cases, AI Status Update returns blank values or “Not specified” for fields that definitely contain data.

During one earlier attempt, I also noticed that the AI populated the task title under the Open Task column instead of reading the Open Task custom field. In my project, Open Task is a dedicated custom field (date type), not the task name, and my prompt explicitly distinguishes between the two. Despite that, AI appeared to substitute the task title instead of retrieving the requested custom field value.

Again, when running the exact same prompt through Smart Chat, every custom field was correctly identified and populated with the expected values.

Expected Behavior

My expectation is that AI Status Update should:

  • Retrieve every incomplete task matching the criteria instead of only a subset.

  • Correctly read and populate the specified custom fields.

  • Preserve the requested column mappings exactly as defined in the prompt.

  • Produce a complete structured table that can be used directly as a project status report without manual verification or correction.

Actual Behavior

Instead, AI Status Update appears to:

  • Return only a partial list of matching tasks.

  • Omit many incomplete tasks.

  • Replace populated custom field values with “—” or “Not specified.”

  • Occasionally map the wrong data into requested columns.

  • Produce results that differ significantly from Smart Chat despite using the exact same prompt.

Prompt Used

prepare a report (include only incomplete tasks) in a single table format which should have following format:

Column 1-
Name of the column heading should be "Subject PIC"
Content it should have - "Assignee" custom field value

Column 2-
Name of the column heading should be "Country"
Content it should have - "Country (VR1)" custom field value

Column 3-
Name of the column heading should be "Open Task"
Content it should have- "Open Task" custom field value

Column 4-
Name of the column heading should be "Priority"
Content it should have- "Priority" custom field value

Column 5-
Heading should be "Info. Report status"
Content it should have - Section (name) in which it is present

Questions

  1. Is AI Status Update intentionally designed to summarize or sample project data instead of returning every matching task? If so, is there any way to force a complete, non-truncated output?

  2. Is there a recommended prompt structure or best practice that improves the consistency and accuracy of custom field retrieval within AI Status Updates?

  3. Has anyone experienced AI Status Update returning “Not specified” or “—” for populated custom fields, or mapping incorrect data into requested columns?

  4. Since Smart Chat consistently produces a complete and accurate table using the exact same prompt, should Smart Chat currently be considered the recommended approach for generating detailed, task-level reports? Or is there a way to achieve the same level of completeness and accuracy directly through AI Status Updates?

Any guidance, clarification, or recommendations from the Asana team or community would be greatly appreciated. I’d especially like to understand whether these differences are expected limitations of AI Status Updates or whether there’s a configuration or prompting approach that can help produce the same comprehensive output that Smart Chat is able to generate.

2

Hi @Bhumika_Bihani, I totally understand why that’s confusing, especially when you’re using the same prompt but getting different results from two AI features. A few things that might help explain what’s happening:

What’s going on:
AI Status Updates are designed more for concise stakeholder summaries rather than pulling a full task-by-task list. There’s a limit to how much it can output, so even if you ask for all 37 tasks, it may summarize or only include a portion (like the 8–10 rows you’re seeing).

Smart Chat is better suited for pulling detailed task-level information, which is why it’s able to return the full table.

For the custom fields showing as “—” or getting mixed up, a few things to check:

  • Make sure the custom field is actually added to all tasks in the project and that there aren’t hidden or duplicate versions of the same field.
  • If you have duplicate field names (for example, multiple “Country (VR1)” fields), AI may grab the wrong one. Cleaning up duplicates or renaming them can help.
  • Check that the fields actually have values filled in. Some field types, like dates or multi-selects, may show as blank if no value is set.
  • If tasks are multi-homed, the section may be pulled from another project. You can clarify in the prompt that you only want the section from this specific project.
  • If you recently made changes, give it a little time and try running it again.

A few prompt tips that may improve results:

  • Be very specific with the field names and types: “Use the custom field named Country (VR1), type: single select.”
  • Define the scope: “Only include tasks from this project and use this project’s sections.”
  • Tell it not to infer values: “Leave blank if the field has no value. Do not use the task name as a substitute.”
  • Include task IDs or URLs as a column so it’s easier to spot mismatches.
  • If you have a lot of rows, try asking it to provide results in batches (e.g., “List 15 rows at a time and say CONTINUE for the next batch.”)

So in short: yes, Status Updates are currently better for summaries and key highlights, while Smart Chat is the better option when you need a complete task-level table.

A couple of workarounds:

  • Use Smart Chat to generate the full table, then copy it into your Status Update.
  • Use Advanced Search with the columns you need and export the results as a CSV.

Hope that helps! :blush: