A recap of our conversation with Kyle Forrest (Deloitte) and Claire Schmidt (AllVoices)

Most companies aren't actually failing right now. They're running a playbook that no longer matches reality.

That was the framing for the kickoff session of Rebuilding the Modern Workplace, our new webinar series for HR leaders trying to figure out what to keep, what to throw out, and what to build next. In this first session, host Rebecca Taylor sat down with Kyle Forrest, Future of HR Leader at Deloitte, and Claire Schmidt, Founder of AllVoices, to dig into what's actually broken inside HR right now, why copy-pasting another company's "best practice" usually makes things worse, and what HR leaders should be doing instead.

The short version: the ground moved. AI showed up. Hybrid redrew the org chart. Employee expectations evolved. And most of the frameworks HR teams are running today were built for a different era of work.

Below is the full breakdown of what was discussed, with the strongest moments pulled out as block quotes you can actually use.

Why HR's Old Playbook Stopped Working

The opening poll set the tone: when your playbook stops working, what do you do? The audience was split between Google what other companies are doing, ask my network, and throw it out and start from scratch. Almost nobody picked tweak it and hope for the best.

Kyle's read on the moment: there isn't a playbook for what we're living in, so HR leaders need to actively go look for inspiration outside of their own four walls.

"There's no playbook per se for what we are living in. You need to get outside of your four walls. Whether that's Ask Your Network, whether that's Google what other companies are doing, whether it's Ask AI. People need to look outside of their current environment to get some inspiration for what may come next."— Kyle Forrest, Deloitte

He's not wrong about the urgency. According to Deloitte's 2026 Global Human Capital Trends report, which surveyed more than 9,000 leaders across 89 countries, 70% of business leaders said their primary competitive strategy for the next three years is to be fast and nimble.

The problem? Speed without direction is just motion.

"Everyone's used to navigating in their rearview mirror using lagging indicators, and it's really hard to look forward. This understanding of the trends, what bets do you make that are safe, is really one of these friction moments that organizations are trying to navigate through right now."— Kyle Forrest, Deloitte

Claire pointed out a related and very specific pressure HR leaders are feeling right now: the assumption that because AI is so accessible, HR should just build everything in-house.

"There's a lot of expertise out there that you can tap into where people have already done all the heavy lifting for you. You don't have to start from scratch for everything. AI is not perfect. You don't have to feel the pressure to go it on your own and do everything by yourself, and then have to clean up the messes after the fact."— Claire Schmidt, AllVoices

The takeaway here isn't that HR shouldn't experiment with AI. It's that experimenting with sensitive HR data without the right guardrails, training, and compliance backing creates risk you can't unwind. Especially when employees are the ones who eventually feel the consequences.

Best Practice Is Dead. Welcome to "Next Practice"

One of the strongest reframes of the session came from Kyle when Rebecca asked how HR teams should be thinking about benchmarking right now.

"There's no such thing as the best practice. There is the next practice, meaning things continue to keep moving. Let's go get a sense of what others are doing that's next, and let's go try it. Let's literally move."— Kyle Forrest, Deloitte

The shift from "best practice" to "next practice" matters because best practice implies a settled answer. Next practice acknowledges that nobody has the answer yet, and the smart move is to find what's emerging, test it in your own context, and learn fast.

But that doesn't mean you should copy whatever your favorite LinkedIn creator is doing. Kyle laid out a practical filter for evaluating outside signals before you act on them.

How to Evaluate Outside Signals Before You Act on Them

When deciding whether to trust a trend, a case study, or a peer's advice, Kyle's filter has three parts:

What expertise does the source have, and for how long? A founder posting hot takes from six months of building is not the same as a CHRO who has shipped this transformation across three companies.
Where did the data come from? Real research with sample sizes and methodology, or "I asked five friends at dinner"?
Is it contextualized to your industry, company size, geography, or function? Broad sweeping generalizations are interesting but rarely actionable.

"Watching someone build on LinkedIn using an AI tool, how they're automating all their hiring and screening, is not helpful for me if I work in a regulated bank or if I work in a company that has countries in Europe or the Middle East, which have certain regulatory requirements related to employee data and privacy."— Kyle Forrest, Deloitte

Translation: copying what someone else does is never going to work inside your own organization, because it doesn't account for your business strategy, your culture, your compliance footprint, or your tech stack.

The Black Square Problem (And Why It Keeps Happening with AI)

Claire used a memorable analogy that landed hard with the audience: the 2020 black square moment.

"Companies were all kind of looking at each other being like, what are we doing here about this DEI thing? And one started posting the black square, and then other people started posting the black square simply because each other were doing it. It almost created more criticism and more controversy because it seemed so performative. A lot of the companies that had been posting the black square were some of the worst offenders in terms of internal DEI practices. If they had taken the very small amount of time it took them to decide to post the black square and done one thing internally to improve DEI at their company, it could have gone so much further."— Claire Schmidt, AllVoices

The parallel to AI is hard to miss. Companies are racing to roll out AI initiatives because everyone else is, without doing the internal work to understand what their employees actually need or how those decisions will land.

Intent doesn't equal impact. And in HR, the gap between the two is exactly where culture problems start compounding.

Decisions That Echo: How AI Choices Today Shape Culture for Years

Kyle introduced a phrase from Deloitte's 2026 trends report that stuck with everyone: decisions that echo.

"We are making decisions today that we're going to have to live with two, three, four years from now. If we're not really intentional with our design choices as a company, as a country, as a community and society, there are going to be implications that we have to live with that we might not really understand are the outputs of choices we're making today."— Kyle Forrest, Deloitte

He flagged two specific risks that any HR leader rolling out AI should be tracking right now.

Cognitive Offloading and Cognitive Surrender

This is what happens when employees let AI do all the thinking for them. The convenience feels great in the short term. The long-term cost is a workforce that's lost the muscle to think critically, push back on bad answers, or solve problems without a prompt.

Cognitive Overload

This is the opposite failure mode. AI takes the easy work. Humans are left with only the high-cognitive-load work. There's no recovery time built in, and burnout accelerates.

Both are real, both are showing up in organizations right now, and both are largely invisible to leadership until they show up as attrition spikes or sudden engagement collapses. This is exactly the kind of early-warning data HR teams should be tracking, ideally with a real-time view of employee sentiment instead of a quarterly engagement survey that lands six months too late.

Culture Debt: The Overdraft Fee Nobody Sees Coming

Speaking of compounding consequences, Kyle introduced one of the sharpest concepts of the entire session, also from the Deloitte report: culture debt.

"Just like with the bank, if you overdraw your account, you get hit with a penalty. Just like in an organization, if you take a series of moves that have a negative impact on culture, you will have an overdraft. Either attrition will spike, or engagement goes down, or you name it."— Kyle Forrest, Deloitte

According to the report, 65% of organizations believe their culture needs to change significantly because of AI, and 34% say their culture is actively blocking their AI transformation goals. The companies racking up the biggest culture debt right now are the ones treating AI rollouts as pure efficiency plays, without addressing how those rollouts will land with employees.

Kyle's antidote: be more explicit about implicit things.

Why HR Needs to Make the Implicit Explicit

If you're rolling out AI, don't assume employees understand what it means for their job, their bonus, their headcount, or their career path. If you don't say it out loud, employees will fill in the gap with their own narrative. And in a low-trust environment, that narrative is rarely generous.

The questions HR should be answering out loud, on the record, in writing:

What do we want people to do differently with AI in their workflow?
How is performance going to be measured if AI is doing more of the work?
Are we holding headcount flat? Growing? Reducing?
What's the upside for the employee who actually leans into this?

Without explicit answers, the rollout is incomplete, no matter how good the underlying tool is. This is the place HR has the most leverage right now: not in slowing AI down, but in demanding the answers nobody else is asking for.

The Bias Question: Can You Actually Train Bias Out of AI?

A great audience question came in: if AI exhibits bias, doesn't that imply it was trained to have such bias in the first place?

Kyle's answer reframed how a lot of HR leaders are thinking about this.

"We know humans are biased. We know some bias will get into AI, but you can begin to train the bias out of AI because you can look for it. We are never going to train bias out of humans. The way the human brain works, it's all that unconscious mental model matching. We can actually train that out of AI."— Kyle Forrest, Deloitte

He referenced a comment from former EEOC commissioner Keith Sonderling: when the EEOC first started examining companies using AI in hiring, they actually became more concerned about companies that weren't using AI, because the alternative was unmonitored human bias with no audit trail.

That doesn't mean AI gets a free pass. It means the design choices matter enormously. Who trained it. What language they used. What context they baked in. Who tested it before launch.

Kyle gave a concrete example of how bias sneaks in even when nobody intends it to:

"I'm working with an organization right now to revisit some of their policies. They actually write the policy documents with some unnecessarily flowery language. So that actually makes it challenging for people who may not be non-native English speakers to interpret a policy document. The introduction of bias could come through a number of ways. Was the language clear? Who was involved in training the first time? Was it a team of people of similar backgrounds, same educations, same upbringings, same countries?"— Kyle Forrest, Deloitte

The lesson: most bias in AI isn't intentional. But because you can audit AI in ways you can't audit human judgment, you can also course-correct it in ways you can't course-correct unconscious human bias. The catch is that the audit only works if you actually do it, before launch and on an ongoing basis.

"It's so hard to build trust but so easy to lose it. If you launch something that just completely whiffs and fails to a level where it looks like you have a ton of bias in an area that you might not mean to, your employees, it's going to take a lot for them to trust the next iteration of it."— Rebecca Taylor, AllVoices

If you're rolling out AI in any process that affects careers (hiring, promotion, performance, comp), the questions you ask vendors about how their AI was trained, monitored, and audited matter as much as the features themselves.

Bringing Stakeholders Along: Where AI Rollouts Quietly Fail

Kyle shared a story that anyone who has ever shipped a process change inside a company will recognize immediately.

"There was an organization sharing with me their journey in trying to help hiring managers not have to review as many resumes. They said, 'we used to give them 50, now we give them 10.' And the reaction from the hiring managers was, 'what the heck happened to my talent pipeline?' Because the hiring managers were not brought along the journey."— Kyle Forrest, Deloitte

The TA team had done the work. The AI was probably surfacing better candidates. But because the hiring managers didn't understand why their pipeline shrank, they assumed the system was broken, lost trust in the tool, and the rollout had to start over from scratch.

This is the unglamorous part of AI deployment that gets skipped most often. The change management. The communication. The "here's what we're doing and why" memo that nobody wants to write but that determines whether the rollout actually sticks.

The Single Most Underrated Source of Insight: Your Own Employees

Claire kept returning to one theme throughout the session, and it might be the single most actionable takeaway of the whole conversation.

She told a story about revamping benefits at AllVoices a few years ago. She did the research. She asked other CEOs. She asked friends. She built what she thought was the perfect list of options. Then, at the last minute, she ran an anonymous survey to her own employees with her shortlist.

"The winner was a write-in. The winner was not on my list. I was so far off, and I thought what I was thinking of was so great. Free gym memberships, mental health reimbursement, all this stuff, but it was stuff I thought of. In reality, people wanted something different that I had not even thought of."— Claire Schmidt, AllVoices

The same principle applies directly to AI rollouts. Employees are using AI right now, in ways their leadership often has no visibility into. They have ideas about where it would actually help in their roles. They know exactly where the friction is in their day-to-day work. Most companies still aren't asking them.

"Employees are the ones that have the insight that's actually needed by leadership to make decisions. Yes, of course those decisions can also be informed by best practice and top-down strategy. But employees, a lot of them are using AI. A lot of them have great ideas about how it can be put to use in their role."— Claire Schmidt, AllVoices

The HR teams getting AI right are the ones running a constant, real-time dialogue with employees about what's working, what isn't, and where the gaps are. The ones rolling it out top-down with no feedback loop are the ones writing the next round of culture debt.

This is also why having infrastructure that surfaces what employees are actually feeling in real time matters so much more than it used to. A quarterly engagement survey isn't going to tell you that your AI rollout is quietly tanking trust in three of your departments. A real-time employee relations platform will.

What HR Leaders Should Actually Do Next

Rebecca closed the session with a question for everyone watching: if you had the chance to ignore best practices for a minute and just look inside your own organization, where would you start?

The practical takeaways from the conversation:

Stop chasing best practice. Hunt for next practice. Find what's emerging, pressure-test it against your own context, and move.
Be explicit about implicit things. Don't let employees fill in the AI narrative for you. Say it out loud, on the record.
Track your culture debt. The decisions you make about AI today will echo for years. Watch for the early-warning signals.
Use AI to find bias you can't see in yourself. But test before you launch, and audit on an ongoing basis.
Ask your employees. They have the answer that most leadership teams are paying outside consultants to find.

If you want to see what your employees are actually thinking and feeling before friction becomes a fire, that's exactly what we built AllVoices for. Modern employee relations infrastructure with real-time insight into the things HR usually only learns about after they've already broken.

Book a demo and we'll show you what that looks like inside your organization.

Catch the full session on demand and register for next week's conversation in the Rebuilding the Modern Workplace series, where we're talking about the metrics that actually matter (and the ones HR teams should stop tracking), with leaders from Netflix, Chili Piper, and Illumina.