There's growing chatter about people being cut due to less token usage.
If you weren't heavily using your copilot, it's being seen as "you're not growing" and "adapting" to new technology.
To me it's seems mo--nic,
It's another metric of punishing people who are efficient while people who used dev assist for simply changing a label on UI are being rewarded.
@fp I'm @ce -- Are things really as bad in general at other financial companies? FFIO seems anomalously bad even within Fidelity. I just graduated college in 2025 and I'm having to teach basic computer skills to senior software engineers. Our api endpoints in production have typos in their paths. We hand-roll json parsing in multiple different places. Entire folders are copy pasted between repositories. How did this happen ?!?!???
With regards to AI more specifically, Bill seems like he is fairly level-headed and has explicitly said that it's a bubble. I think that it's Maha who is pushing this and then causing all of his underlings to amplify it.
Outsourcing isn't cheaper. A team of one American will produce more than a team of ten offshore if you are measuring functionality instead of Confluence documents.
@ce Yeah , My boss asked me to attend that meeting to view this "sausage-maker" and frankly I wasn't impressed at all .. I'm currently working with few AI tools and it does improve my productivity to some degree ;- however the bad part is that they know the truth ... and they know they'll get zero return on investment but keep spending crazy bucks bc potentially they see it as a nice and friendly mechanism to kick you out and outsource your job somewhere else where the actual labor is much cheaper . So , both components of this puzzle AI and RTO are perfectly aligned together with only one goal ... Get rid of people without paying severance either for RTO violation (will kickoff at September) or negative review due to AI stupidity . Unfortunately it's not just Fidelity issue ... The whole Fintech industry aligning with this roadmap .. Spoke with my buddy from BNY ... Same S**T .
When I become that efficient, I will leave to go somewhere where I'm appreciated.
anybody else in the ffio meeting today where the presenter straight up admitted that his "custom agent" could be replaced with a python script, but that he chose copilot because it is "more developer friendly"?
Only at Fidelity are developers so incompetent they need a fu--ing chatbot because running a python script is too difficult.
If fidelity leaders were smart, they'd see the token usage vs the throughput from the associates before sounding alarm bells.
There are members in my team who's usage didn't go past 20% and they delivered more than 2-3 team members (as highlighted by @OP).
If token usage is the only criteria to stay in this firm, then
1) Soon fidelity will realize that token costs are higher than someone's salary over a year.
2) The high usage folks will get the axe for not being efficient
It's like you hook someone to high speed wifi and all they do is browse questionable content, only to move back to dial up connection, where they can barely google stack overflow.
Just my two cents, but in the long run I'd rather be efficient and have less tokens with more output to show for than absurd about of token and nothing to show for
Have been penalized for too much token usage, am not supposed to go more than 20% ! It's in my mid year review, better to go somewhere else
This firm thinks they know how to measure output and productivity. Changing tracking tools and letting people enter their own time because they have no idea how to measure. They look at teams to figure out productivity. Better make sure you are green or they will think you aren’t productive even though you are sitting in a zoom meeting and the status on teams is yellow.
This firm is beyond imagination.
First they make us heavily use AI, then add limits to our usage.
And now this, efficiency was always frowned upon in this firm when a person doing job of almost 3-4 team member is penalized while pencil pushers get prompted.
So efficient users of AI, with more output to show for would again be penalized.
I believe in this post.