After talking to some people who actually have been forced to work with AI on projects in teams Its been horrifying to find how little time they are given to build out proper safe guards, rules and test the process.
Management seems to be expecting these stupid robots to work the first time every time
Even in small projects for every 8 hours of work there’s a good hour or two of adjusting memory, standing rules, hooks, and countless other things.
It’s endless upkeep. Every shift in work flow tends to create knock on effects.
Management of the bot it self becomes a massive work load if you want them to be even remotely reliable.
I’m not surprised their use in a corporate environment is backfiring massively.
It reminds me of talking to a coworker where I teach about taking on a side job. “All he wants is to take stuff from these 3 databases, and then reformat them to go into this database that he can use with this other software.”
They didn’t like when I told them that they couldn’t just vibe code it in a month and that my previous job was working with a persistent and ongoing team of almost 10 people doing that same type of thing.
To clarify before anyone asks, yes, these are all 3 live databases in production environments.
I sort of know what you mean. The company I work for has been encouraging the use of AI/LLMs. But some of the company leadership has been smart about it. We recently got some AI agents to use as tools in our workflow. However, they tested the agents for a few weeks before allowing us to use them, so that they know the agents do what they’re suppose to do.
However, they tested the agents for a few weeks before allowing us to use them, so that they know the agents do what they’re suppose to do.
I hope they’re planning on doing that for every model change, or pinning it to a specific model. Otherwise, you’re gonna have a bad time when model updates completely change how the bot functions.
After talking to some people who actually have been forced to work with AI on projects in teams Its been horrifying to find how little time they are given to build out proper safe guards, rules and test the process.
Management seems to be expecting these stupid robots to work the first time every time
Even in small projects for every 8 hours of work there’s a good hour or two of adjusting memory, standing rules, hooks, and countless other things.
It’s endless upkeep. Every shift in work flow tends to create knock on effects.
Management of the bot it self becomes a massive work load if you want them to be even remotely reliable.
I’m not surprised their use in a corporate environment is backfiring massively.
It reminds me of talking to a coworker where I teach about taking on a side job. “All he wants is to take stuff from these 3 databases, and then reformat them to go into this database that he can use with this other software.”
They didn’t like when I told them that they couldn’t just vibe code it in a month and that my previous job was working with a persistent and ongoing team of almost 10 people doing that same type of thing.
To clarify before anyone asks, yes, these are all 3 live databases in production environments.
I sort of know what you mean. The company I work for has been encouraging the use of AI/LLMs. But some of the company leadership has been smart about it. We recently got some AI agents to use as tools in our workflow. However, they tested the agents for a few weeks before allowing us to use them, so that they know the agents do what they’re suppose to do.
I hope they’re planning on doing that for every model change, or pinning it to a specific model. Otherwise, you’re gonna have a bad time when model updates completely change how the bot functions.