Shoutout to whoever thought about having a company wide broadcast about leaning into Agentic AI that can do a lot of the workload. Even more so for doing it just days after announcing major cost cutting moves. This company really doesn’t have any sense of its employees sentiment right now.
12 replies (most recent on top)
@11s Yeah, I agree. FLOW is a good example. We brought in consultants to implement it, spent a significant amount of money, and haven't seen meaningful results. When it moved to the internal team in GTO, it didn't improve. The platform has been down multiple times, and people can't rely on it for their day to day work. I think the bigger issue is leadership. There's been a lot of focus on promoting these tools and very little accountability for whether they actually deliver. When people raise real technical concerns, they tend to get dismissed instead of addressed. That team also doesn't have the depth of experience this kind of work requires, and pushing adoption harder doesn't fix that. Most people aren't choosing to use it, and that should tell us something. Right now a lot of folks are finding their own tools because they need something that works. I'd rather we be honest about where things stand, measure these tools on reliability and real usage, and invest in what's actually delivering value. I'm happy to stand behind that feedback.
I have significant concerns about our current AI strategy, execution, and, most importantly, the culture surrounding it. From where I sit, much of the AI ecosystem we have assembled feels duct taped together rather than engineered as a reliable enterprise capability. Microsoft 365 agents, engineering agents, MCP servers, integrations, and supporting tools frequently fail, become unavailable, or simply cannot complete the work they are supposed to perform. Yet we continue talking about these capabilities as though they are already delivering transformational productivity.
There appears to be an enormous gap between the AI story being communicated and the experience of the people actually trying to use these systems. What concerns me even more is the pressure to reinforce the narrative that the strategy is working. When employees provide legitimate technical criticism or point out that tools are unreliable, ineffective, or failing to produce measurable outcomes, that feedback should be treated as valuable data. Instead, there is a perception that uncomfortable feedback gets buried, minimized, or excluded while positive examples are amplified.
That is dangerous. If leadership only hears evidence confirming that the strategy is succeeding, eventually the organization begins making decisions based on its own narrative rather than reality. The uncomfortable truth is that the AI we currently have is nowhere close to replacing the breadth of work performed by skilled employees. In many cases, employees are spending additional time troubleshooting agents, working around failed integrations, validating AI output, or simply doing the work themselves after the technology fails. Calling that productivity does not make it productivity.
We have also moved between strategies without adequately resolving the underlying economics. We pursued vendor based AI solutions despite significant budget constraints. Now there is discussion of moving toward open models, which may reduce certain vendor dependencies but introduces substantial infrastructure, engineering, security, operational, and talent requirements of its own. If we do not have the capital, infrastructure, or specialized talent required to operate that capability at enterprise scale, changing models does not solve the underlying problem.
There is an organizational problem underneath the technology problem as well. AI transformation requires people who are willing and able to challenge assumptions. Leadership should actively seek out the engineers, architects, product leaders, and operators who understand where these systems are failing. Instead, there is a perception that people who challenge the prevailing narrative are pushed aside while people who reinforce it maintain influence. That creates a particularly troubling risk during restructuring. If experienced employees and strong technical dissenters leave while the organization retains the systems and assumptions they were questioning, the remaining workforce inherits the consequences. They will be responsible for operating unreliable systems, compensating for their shortcomings, and eventually reconciling promised productivity improvements with what the technology can actually deliver.
The strangest part is that the narrative can eventually become self reinforcing. If everyone is expected to say the AI works, leadership hears that it works. Leadership then makes additional investment and workforce decisions based on that information. Those decisions become further evidence that the strategy must be working. That is not transformation. It is confirmation bias at organizational scale. We need independent and transparent measurement, agent availability, successful task completion rates, failure rates, human intervention required, hours actually saved, hours spent correcting or troubleshooting AI, total operating cost, and measurable business outcomes. Those results should be visible whether they support the current strategy or contradict it.
AI may ultimately transform how this organization operates. But believing that outcome is inevitable is not a substitute for proving that the technology we have today actually works. If we cannot openly discuss where the technology is failing, we do not have an AI problem. We have an accountability problem.
lol instead of investing in the workforce, they just shell out millions to McKinsey to tell them AI will save money and what we need is more layoffs. What a joke. ELT is so disconnected from the rest of their employees and gobble up whatever “get rich quick” advice consultants give them. Short term profit over long term sustainability is the pattern here…far too many execs don’t even care, they’re ready to retire and are simply collecting their fat bonuses. They know they only need to keep the ship afloat a little longer before it’s not their problem anymore, so they’ll make decisions without caring what the future ramifications are.
@bg while not being able to afford the token bill
They’re pushing AI down our throats to increase productivity for the folks who survive the lay offs. “Hey sorry we got rid of all your coworkers but don’t worry, use AI to help get work done!”
For something that is supposed to end humanity in 10 years that’s a lot of trust to put into such technology. Seems sus.
AI hasn’t worked out so well for other big companies (Ford, IBM, McDonalds to name a few) and callbacks have been made. Comcast throws trust into anything and anyone saying this will save a buck Lower level internal tasks seem to be the only way, but even for things such as CS, clients want a workaround (rep) when a bot pops up because they hate it or don’t trust it. Apparently AI’s own creators don’t trust it these days. Hmm.
@a9 I can't believe they let her say "my AI boyfriend".
They did put in a disclaimer at the end: opinions expressed of the guest or whatevs.
I mean...the writing is on the wall lol.
Just make sure we’re not only venting here after Knowledge Fest. Put it in the survey too. If it felt tone deaf and didn’t land, they need to hear that directly in the feedback.
We've seen this movie already at other companies ..big agent rollout …then a token bill nobody forecasted. Have we run the numbers on how we're paying for all this?
More devx garbage