# Asana's AI Support Agent is broken and misdirects users. Worst I have ever encountered

**URL:** <https://forum.asana.com/t/asanas-ai-support-agent-is-broken-and-misdirects-users-worst-i-have-ever-encountered/1134484>\
**Category:** Product Ideas\
**Tags:** agentic-work-management\
**Created:** [2 April 2026 18:49 UTC](https://forum.asana.com/t/asanas-ai-support-agent-is-broken-and-misdirects-users-worst-i-have-ever-encountered/1134484 "2026-04-02T18:49:35Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Robert\_Tolmach](https://sea2.discourse-cdn.com/asana/user_avatar/forum.asana.com/robert_tolmach/32/62053_2.png) [@Robert\_Tolmach](https://forum.asana.com/u/Robert_Tolmach)\
**Post date:** [2 April 2026 18:49 UTC](https://forum.asana.com/t/asanas-ai-support-agent-is-broken-and-misdirects-users-worst-i-have-ever-encountered/1134484/1 "2026-04-02T18:49:35Z")

</div>

I came to Asana’s AI Support Agent with a simple question: _how do I submit a UX suggestion?_ The correct answer was one sentence: “Post in our Product Feedback forum at [forum.asana.com](http://forum.asana.com).” Instead, the AI sent me on a 30-minute wild goose chase. Here’s what went wrong:

1. **Wrong answer from the start.** It told me to create a “Customer Feedback project” in Asana — a completely made-up process that doesn’t exist for submitting feedback to Asana.

2. **Never mentioned the forum.** The actual answer — your own Product Feedback forum — was never mentioned once in the entire conversation.

3. **Ignored screenshots repeatedly.** I sent multiple screenshots. The AI repeatedly described UI elements that weren’t there, then when corrected, gave the same wrong description again. And again. [Transcript of the worst AI support chat ever](https://docs.google.com/document/d/1rOkcdkBoPb11blN78GvdvHX3ZdL1ohcKIojhCbVTwqY/edit?usp=sharing)

4. **Looped endlessly.** It repeated identical responses word-for-word even after I explicitly said “I already told you no.” At one point it asked me the same question three times in a row.

5. **Gave developer-level instructions to a regular user.** When I asked how to tell if I’m in an organization vs. workspace, it told me to check the `is_organization` flag via the API.

6. **Never escalated or offered a human.** At no point did it acknowledge it was stuck or offer to connect me to a person.

I’m attaching the full transcript. This AI is causing user frustration and brand damage. It needs significant improvement.