Talent advisor says AI is accelerating organizational debt, not fixing it
Discover how high-growth leaders use AI to scale decision-making and execution, while avoiding the organizational debt that comes from treating AI as a shortcut.
Heather Doshay
Founder of Waypoint Works
AI Leadership: AI reshapes leadership, necessitating adaptability in high-slope leaders during rapid technological transitions.
Operational Shift: AI encourages companies to shift from a 'buy vs. build' mindset to an experimental 'tinker before buy' approach.
Cultural Transformation: Integrating AI into companies isn't just about tools but requires a broader cultural transformation strategy.
Human Judgment: AI supports decision-making by synthesizing information, but human judgment remains crucial for nuanced decisions.
Innovation Challenge: AI accelerates innovation but often fails in comprehensively replacing human workflows, emphasizing augmentation.
Heather Doshay is the founder of Waypoint Works, where she helps high-slope leaders manage the transition to AI. She previously served as a Partner and Head of Talent at an AI-native, $3B AUM venture capital firm.
We caught up with Heather to learn how strong operators are using AI to scale their impact. Here's what she had to say.
How AI Reshapes Leadership in Fast-paced Environments
When ChatGPT launched in late 2022, I was already a Partner and Head of Talent at an AI-native venture capital firm, where we had invested in and built around earlier waves of AI for years. But the rise of large language models fundamentally reset the playing field.
Keep Reading—and Keep Leading Smarter
Create a free account to finish this piece and join a community of forward-thinking leaders unlocking tools, playbooks, and insights for thriving in the age of AI.
Have an account? Log In
Work we expected to evolve gradually suddenly risked obsolescence. Even for an AI-first organization, the pace of change forced us to completely rethink our operations.
My role intersected two transformations. Internally, we rapidly redefined how an AI-native company works as the technology shifted daily. Externally, my team advised over 200 venture-backed startups facing the same ambiguity.
Founders asked for best practices that did not yet exist, and we helped them navigate everything from talent strategy to operating models in real time.
That period, especially from 2023 through 2025, marked a shift from curiosity to urgency. What started as exploratory conversations quickly became foundational to how companies build, hire, and compete.
At the start of this wave, I was a Partner and Head of Talent at SignalFire, a ~$3B AUM, AI-focused venture capital firm. VC teams are intentionally lean, so my team and I supported over 200 portfolio companies across the full talent lifecycle, from hiring and org design to compensation and scaling distributed teams.
As AI accelerated, our scope expanded from advising on “what” companies should do to helping them rethink “how” work gets done. AI wasn’t a function, it was a new operating layer. I left that role in August 2025 to found my own company.
In September 2025, I founded Waypoint Works, where I coach, advise, and train high-slope leaders and their teams. My work isn’t about AI transformation directly, but AI shows up consistently as part of how strong operators scale their impact. The focus remains the same: developing people and systems that can adapt quickly, with AI serving as one of the tools that enable this.
How AI Helps Startups Navigate Transformation Chaos
Most founders initially wanted a benchmark or a playbook on how to transition to AI. The reality is none existed. So we shifted our work from giving answers to helping them make better decisions amid ambiguity.
We focused first on orientation, not tools. Instead of asking “What can AI do?” we pushed teams to ask, “Where are we inefficient or constrained, and where could AI meaningfully change that?” That reframing alone prevented a lot of wasted spend on shiny tools that didn't solve real problems.
From there, we helped companies define their posture. Are you centralized or decentralized in how teams use AI? Are you optimizing for speed and experimentation, or control and risk mitigation? Those decisions shaped everything from tooling policies to hiring plans.
For example, some companies ran AI “bake-offs” to evaluate tools safely, while others hosted internal hackathons tied to real business bottlenecks to drive adoption and creativity.
Ultimately, this was a change management exercise, like any other, but focused on AI. The challenge was thinking at a systems layer, not just checking the box that says "I get a gold star because I'm using XYZ AI tools."
The companies that moved most effectively treated AI as a cultural transformation, clarified their principles early, and then consistently reinforced those choices across how they hired, operated, and built.
Why AI Redefines Operational Strategies in Organizations
We made a concrete shift from a “buy vs. build” mindset (e.g., do we buy a tool or build an internal process managed manually?) to a “tinker before you buy” approach.
Historically, companies defaulted to purchasing tools because manual processes didn't scale. With AI, we encouraged teams to first experiment with solving the problem themselves, even in a lightweight way, before committing to a vendor.
That led to a second change expanding who gets to build. We introduced internal team hackathons, historically reserved for engineering teams, but with AI advancements, anyone can participate.
AI lowered the barrier to entry, so office managers, recruiters, and people leaders can prototype solutions to real business problems. That shift alone changes behavior, from passive tool adoption to active problem-solving. People often worry that AI will reduce critical thinking, but I’ve found it can actually strengthen critical reasoning when solving problems in novel ways.
Another shift I've seen in the past year is more CHROs becoming Chief People and AI Officers. Ultimately, the first chapter of this transformation shows a shift in how teams organize and how work gets done. It's an interesting people and change management effort, with two logical homes: CTO/CIO or CHRO. I see the former more frequently, but the latter enough that it's notable.
Why We're in the Messy Middle of the AI Revolution
We’re in the messy middle of this AI revolution. There's a real push toward efficiency based on AI's potential, but AI isn't enterprise production-ready yet. In other words, I can't replace the HRIS I hate with my own product built with Claude Code.
So, while we see much more experimentation and novel problem-solving, it isn't the silver bullet that leadership often thinks it is. Teams scramble to maintain legacy systems as their day job, squeezing in time to tinker with AI, but not fully replacing any legacy systems. It's a double-duty job for HR teams right now, which are already leaner than they were in the past.
Companies aren’t more efficient yet. They’re more exploratory with small pocket wins through AI. But that exploration will ultimately unlock the next wave of productivity, and I believe HR tech will be architected more extensibly in the future, allowing for the best of both worlds. Buying and then building on top of it.
When further AI advancements enable this, I'm excited to see larger-scale shifts.
How AI Supports Leadership Decisions without Replacing Human Input
The line is pretty clear. AI can support the system, but humans own the judgment. Decisions that require nuance, taste, and discretion, like who to hire, how to give meaningful feedback, or how to allocate resources, still depend on experienced operators.
AI doesn't make decisions for us — that wouldn't work practically or legally. I consider where it accelerates inputs and improves the process. It powerfully synthesizes feedback, drafts performance reviews, structures thoughts from voice to text, and aggregates hiring and workforce data into usable information.
For most of this, I just dump things into Claude and have it synthesize various inputs. But for 360 reviews, I like a tool called Your360.ai.
Today, AI can get you to a stronger starting point faster, but it cannot replace the human lens that determines what “great” looks like.
How AI Fosters Innovation Yet Challenges Effectiveness
I've seen companies avoid expensive vendors by building lightweight internal solutions with AI, sometimes saving thousands annually, along with a meaningful increase in employee autonomy and creativity as individuals at all levels start solving their own problems.
On the flipside, many organizations are more experimental but not yet more effective, with teams duplicating effort, chasing ideas that don’t land, and delaying investment in existing systems because they expect AI to replace them.
The pattern I’d call out is AI is unlocking bottom-up innovation faster than most companies can direct it. That’s exactly why manager enablement matters more than ever.
The next generation of high-slope leaders will be defined by their ability to translate vision into execution, channel that energy, and turn raw experimentation into real, scalable outcomes that align with topline business goals.
That’s the missing piece today. These efforts are happening in isolated pockets across organizations rather than systematically.
Why AI fails to Fully Replace Comprehensive Workflows
AI hasn't fully replaced roles, especially where work requires end-to-end ownership. It's strong in discrete tasks, scheduling, drafting, and synthesis, but it breaks down across workflows that require judgment, context, and follow-through.
You still need a human in the loop to connect the dots. AI is an overconfident intern who proudly claims it did a piece of work that it never actually started.
I’ll say something controversial, and I stand by it. The large majority of layoffs attributed to "AI efficiencies" are entirely false. We're just not there yet. The unfortunate truth for most is that these layoffs reflect poor headcount planning and previous over-hiring that the business never adjusted to, but companies want a better narrative to inspire shareholder confidence.
AI may compress teams, with one person doing the work of three, but it won't eliminate the need for any functions themselves.
The gap is in ownership and full workflows. AI doesn’t yet reliably handle full job-to-be-done cycles, and it lacks the taste and accountability required to own outcomes. The near-term reality is augmentation and speed, not replacement.
How AI Transforms Hiring from End-to-End
Here's one workflow I’ve used multiple times for end-to-end hiring. AI shows up at every stage, but always as an accelerator, not a decision-maker.
I start with an LLM like ChatGPT, Claude, or Gemini to pressure-test what I need to hire for and draft the job description. From there, I manually edit it until it's ready. Then, I build and launch an interview process inside Ashby, which offers extensive AI functionality across resume sorting, scheduling, and more.
I also paired this with Juicebox for outbound sourcing, using its AI functionality to improve search results, find candidates matching my requirements and preferences, and draft notes to grab their attention about the opportunity on my team.
Once I had candidates to interview, I used Metaview to capture and structure feedback notes in real time. Across all of this, AI handles the operational load—building lists, drafting messages, capturing data—while I focus on evaluation, decision-making, and human connection.
AI generally helps me structure my myriad ideas, automate tedious manual steps, and polish my writing. My discretion, style, and human connection consistently remain my own.
Ultimately, hiring requires a balance of speed and coverage to make the best decisions before losing the most suitable candidate. AI helps me keep up the pace.
I'm excited to see what's coming in the next generation of AI talent tools. Concepts that historically felt wrong to me—like AI running a phone screen—now feel more intriguing.
Does it seem awkward and impersonal? Absolutely. But can it also realistically help you screen 1000 applicants instead of just 20? Yes. The key is maintaining discretion and intentionality to help candidates see the benefit. Interviewing more people leads to fairer outcomes where the best candidate gets hired for the role.
How AI Challenges Traditional Leadership Assumptions
The interesting shift is that while it may feel like a shortcut, it also builds skills. Over time, people who use AI well tend to become better communicators because they are exposed to clearer patterns of thinking and expression.
I’ve had to let go of the assumption that systems need to be fully baked before they’re useful. Historically, if a tool didn’t do exactly what you needed, you didn’t use it. With AI, nothing works perfectly, but it often provides enough directional help to move faster, test ideas, and build something better than before.
I’ve also rethought what “good communication” looks like in leadership. I assumed strong managers had to be naturally great writers to deliver clear, thoughtful feedback. AI changes that. It can take rough thinking and turn it into something more structured and usable, which lowers the barrier but also acts as a kind of real-time coach.
The interesting shift is that while it may feel like a shortcut, it also builds skills. Over time, people who use AI well tend to become better communicators because they are exposed to clearer patterns of thinking and expression.
How to Avoid Common Pitfalls in AI-enabled Initiatives
Early on, I tried to build an AI “extension” of myself to answer questions and scale our team’s knowledge, but it worsened with each iteration. The issue wasn't the idea, it was that I hadn't designed the system up front. I gave a vague prompt and iterated 500 times, and by the end, it was too convoluted.
I learned that if your initial prompt is weak or lacks design thinking detail, adding more instructions later usually worsens the output, rather than improving it. You’re better off restarting with a more fully thought-out prompt than trying to patch a bad one.
Today, I sometimes draft a 16-page Google Doc as a prompt and upload it instead. It gets me much farther, faster.
The takeaway for leaders is simple. Think like a product engineer and apply design thinking elements. Figure out the end state in advance. Most people try it once, it doesn’t work, and they give up. In reality, they didn’t fail at AI, they skipped the upfront thinking required to make it work.
Why AI Faces Challenges with Fairness and Accountability
AI struggles most when people mistake pattern recognition for judgment. Because it’s trained to predict what’s most likely based on existing data, it can quietly reinforce the past rather than challenge it.
Hiring and sourcing provide a clear example. Tools might over-index on candidates from prestigious schools or well-known companies because that’s what shows up most often in the data, which can unintentionally filter out high-potential, non-traditional talent.
Similarly, language models can reflect bias in how they draft performance feedback, using different tones or descriptors based on gender-coded or culturally loaded patterns.
Issues also arise in consistency and accountability. AI can generate confident but incorrect outputs, and it’s not always clear how it arrived at a recommendation, which makes it hard to audit or defend decisions.
That’s why human judgment still matters, because knowing when to trust the output, when to question it, and how to apply context that the model simply doesn’t have is vital.
How Leaders Can Manage AI-driven Transformations
Too many leaders are treating AI like a box to check. “We’re using this tool, we’re experimenting, we’re adopting.” That mindset might earn you a gold star from your CEO today, but it's not going to solve your organizational debt. If anything, it might accelerate it.
Look at your organization and audit what's working and what's not. That means getting sharper on how decisions are made, who owns what, and how work flows across the organization. Figure out what needs to change, then look around corners to 1 year, 3 years, 5 years out to determine what foundational shifts are needed.
Don't try to predict the future perfectly. Focus instead on asking better questions. I'm a big fan of first-principle thinking and second-order thinking. What does a well-run organization look like 5 years from now? What are the changes needed to get there? If this changes how work is done, what happens to our org design? What skills matter more? Where do roles compress or expand?
The leaders who get this right might move more slowly on vanity metrics like AI adoption today, but they'll create more effective organizations in the long run.
Follow Along
You can follow Heather's work on LinkedIn. And check out Waypoint Works!
More expert interviews to come on People Managing People!
