What are your thoughts on the AI Assistant?

I've encountered a similar situation. It's unfortunate that it doesn't handle most configurations based on the prompt's details. Personally, I find that aspect more demanding than arranging the applications.

However, this is just the initial release, so hopefully, it's only the start of its potential capabilities.

Currently, I believe the AI Assistant sets unrealistic expectations for new users. I recently encountered a situation where someone attempted to use it to synchronize WildApricot Membership Groups with MS Teams membership groups in a one-way fashion. The outcome provided by the AI Assistant was a minimal, entirely unconfigured scenario with just four modules. While it appears theoretically that this scenario could serve as a starting point, even the basic Webhook → Get Contact Details segment is unconfigured. This is understandable, as the AI Assistant wouldn't know the potential webhook payload from the initial module.

image2280×620 59.5 KB

image1688×754 83.5 KB

Adding insult to injury, when you attempt to run this AI-generated scenario, the unconfigured webhook in the first module results in an unfriendly error message:


* The operation failed with an error. Cannot read properties of undefined (reading '__IMTHOOK__')


It's worth noting that the assistant does mention:

Ready to go! Set up connections and mapping to put the finishing touches on your scenario.

It would be more beneficial if it guided users through the setup process module by module, at least to get the initial part of the scenario functional.

The final implemented solution took approximately 10 hours to build and involved over 15 modules. Upon closer inspection, you'll notice the original 4-module scenario created by the AI is highlighted in yellow within the completed workflow. However, observe the extensive complexity, including filters, iterators, and other crucial modules, required to merely complete that specific path of updating groups. It takes 7 modules just to update the group members (not to mention handling the deletion of groups where a member is no longer associated, which is a separate branch from the initial route).

image3596×1470 489 KB

Conversely, if you instruct it to perform very simple tasks, such as "iterate this array," it might perform better. Currently, it cannot be used on existing scenarios to assist with modifications, as it exclusively generates new scenarios.

I completely agree with the previous point. It's still preferable to handle this manually, as it allows you to learn the proper usage of callin.io and understand how your scenario functions.

Remember to consider all the error handling aspects that AI is not yet capable of. At present, AI can only assemble the most basic structure of modules, and sometimes even that is done incorrectly.

Imagine an AI Assistant that could go a step further: documenting scenarios and filling the Notes section with details about each module and its filters. Check out this neat video showcasing a proof of concept that excels at documenting scenarios using just the scenario blueprint.

Hello everyone - happy to hop on a call with you to further explore this idea.

Let me share some light on what we are piloting - error management, whenever you hit an error we will guide you with AI and we are evaluating automated documentation.

What would make an AI documentation great and valuable for you? What should it include - shoot me a reply here or ping me on linkedin to connect and dive deeper on this: https://www.linkedin.com/in/auto-mate/

I believe integrating the community into a specialized ChatGPT search would be quite beneficial. Often, questions posed have already been addressed, and it would be excellent for ChatGPT to provide intelligent answers based on this existing information. I'm happy to connect via the provided link.

I've attempted to use it multiple times but ultimately gave up. If you have a basic understanding of callin.io, the assistant's intelligence level seems quite low. It's faster, more comprehensive, and more reliable to build the functionality myself.

That said, I would really appreciate an option to disable it, either for the entire organization or for specific scenarios. For some reason, that purple blob in the bottom right corner, which I find unhelpful, is quite bothersome.

It's not all negative feedback from my end; I'm certainly open to trying it again in the future.

Best regards,
Henk

Thank you all! Would you be willing to test our new AI and share your thoughts here:

I've had a really poor experience so far, to the point where I'm hesitant to ask questions. Here are two examples from today:

1. I inquired about modifying the “start from” setting, and it proceeded to UPDATE the settings autonomously. This action reverted the module to an older version, causing me to lose progress.
2. It claimed the setting was within the module. I provided screenshots demonstrating it wasn't, and it then suggested using filters or checking my internet connection.

The best part? I asked Google the same question, and Google's AI provided the correct answer (right-click on the module).