My 2024 Review
This is my seventh end-of-year blog, but it’s the first infused with AI.
Let me know if my writing has been taken over by the “professional but approachable” tone of ChatGPT. And if you prefer listening over reading, here’s an AI-generated podcast!
Each year there’s one big change.
2023 was about the dog- we’ve now fully settled into life with Miso, trained him more, and resolved his icky stomach issues. Happy two years to the best short-legged companion anyone could have!
2024 was about getting engaged and married- more on that later.
However, the arguably bigger change as a married couple was Bridget leaving her job of nearly six years. She’s currently exploring operations-focused roles at small to mid-sized companies after spending a decade in client services at marketing agencies.
If you have leads, please reach out!
Table of contents
- On AI
- On community
- On code -> culture
- On morale
- On families
- On marriage
- Memorable moments
On AI
I came out firing with the densest material first. I was close to breaking away from written form and creating slides instead, but I tried to provide plenty of visuals here!
As a self-diagnosed technophile, it’s obvious that AI is the next major platform shift in how we build and interact with technology. I never said this about VR/AR, blockchain, or autonomous transportation, so I probably mean it? Or I’ve just been reading too many posts about AI.
Like with any technology, I try to apply a technorealist lens to AI:
Technologies aren’t neutral
Every dataset has bias, these biases compound exponentially at scale, and they’re a black box to us.
AI: Long-term differentiation in AI comes from proprietary training datasets that are going to be inherently biased- LLM model improvements are already plateauing and the major models are converging on the same performance benchmarks.
The internet is revolutionary but not utopian
We can’t look to the internet to solve all our problems.
AI: While AI will improve productivity, create new jobs, and spur economic growth, it’ll also displace workers, amplify biases, and further widen socio-economic gaps.
Government has an important role to play
Innovation shouldn’t be stifled, but power also can’t be left unchecked.
AI: We’re still in the early days, but OpenAI’s conversion to a for-profit entity is about to be a litmus test for regulatory involvement.
Information isn’t knowledge
We don’t need more information; we need to reason and make sense of what we already have.
AI: With the proliferation of genAI tools, “AI slop” is already rampant online. And it’s getting much harder to determine whether content is generated by humans or not.
Wiring the schools won’t save them.
Technology in the classroom hasn’t meaningfully improved learning outcomes.
AI: AI-only classes don’t strike me as the best way to teach what students may increasingly lack: problem solving, adaptability, how to learn, resilience, and collaboration.
Information wants to be protected.
There’s a growing awareness and legal resistance to the fragmentation and exploitation of personal data.
AI: Eight major news publishers already sued for copyright infringement- the same fight will inevitably widen to include individuals’ sensitive data.
I have three big questions about what’s next:
1. When does the bubble pop?
Investment in AI has significantly outpaced the actual revenue being generated.
Revenue multiple = valuation / revenue. So a company generating $10 million revenue with a $100 million valuation has a 10x multiple.
Revenue multiples for AI startups are 4–5x higher than other SaaS startups. The last time revenue multiple were this high, the bubble burst after interest rates started going up and COVID subsided in early 2022.
According to an analysis from Sequoia Capital, $600 billion in annual revenue is needed to justify the amount of money that’s been invested in AI. At $4–5 billion in annual revenue, OpenAI leads the industry by a wide margin.
That’s quite the revenue shortfall to cover... I’m bullish on AI but those numbers don’t make sense to me.
2. How does a product find a problem?
Normally, you see a juicy problem and you create a product. We call this product-market fit.
AI is the opposite. It’s a solution looking for a problem.
There’s a long list of factors but I think the more important ones are: 1) the pace of model improvement exceeds our ability to identify and implement practical applications, combined with 2) the immense market hype and our innate curiosity (FOMO) about what generalized artificial intelligence looks like.
In other words, we’re riding the early wave of the Gartner hype cycle.
3. What gets disrupted first?
Up to this point, I’ve laid out a somewhat bearish case on the current state of AI, but it doesn’t change the fact that AI is still a major platform shift long-term.
As I see it, there’s three general markets to shift:
- Personal- how AI changes the way we go about daily life
- Processes- how AI changes the way we build products internally
- Products- incorporating AI into external-facing products
All markets are already getting impacted, but my bet is AI truly disrupts 2→3→1 in that order.
I’ll also acknowledge two massive biases: 1) I work at a B2B software company that’s small enough to quickly experiment with AI, and 2) outside of writing a couple blogs each year, I don’t have any creative pursuits, which is where I think genAI is poised to disrupt the personal market.
While many applications of AI (e.g. generating high quality video based off a couple sentences) captivate our imagination, the reality is that the highest-leverage near-term applications of AI are far more mundane… helping businesses run better day-to-day 🙄.
In other words, OpenAI has a clearer path to being the next Salesforce than the next consumer-facing brand.
Okay- enough armchair prognostication! It’s not as fun as getting hands-on with the shiny new AI tools myself.
The best summary of my experience so far has been tempered excitement, which seems to be slightly above the baseline.
There’s an ever-growing list of AI tools to try- many of which undoubtedly won’t exist in a couple years time. They range from cross-application search, AI controlling my browser, analyzing videos, summarizing meetings, generating diagrams, creating images / videos, writing product requirements, and even cloning myself 👯♂️.
Unfortunately, I stop using half of them after an hour.
I’d get more use out of them if I invested way more time, but it’s not time I want to spend when I don’t have a clear problem to solve.
And if you’re wondering where all the social/fun/non-productivity use cases are- don’t bother asking me about those! Remember- I’m a productivity nut.
My experimentation has mostly been in the context of being more efficient at work (i.e. Processes market). I’m still looking for the 10x improvement, but there’s already a lot of small optimizations that have shaved off minutes here and there.
At Persona, we leverage AI in the product itself (i.e. Products market), but the bar has to be much higher. If a customer can touch it, not only does it have to be accurate, but if it is inaccurate, we have to be able to explain why.
The development of reasoning models and interpretability will help, but I think there’s still a fundamental bottleneck to scaled adoption. I want to understand how something works as opposed to blindly trusting it- maybe that’s just me?
Outside of experimenting with different use cases, I’m trying to spend more time with each of the major LLMs instead of defaulting to ChatGPT.
Note: Perplexity isn’t a LLM but it’s become my starting point for any research, and I’m experimenting with it as my default search engine.
On community
Lately, I’ve been thinking more about the communities I’m a part of.
From a personal standpoint, we started planning a wedding and we definitely want to celebrate with our friends and family. The idea of bringing our communities together is thrilling- there’s only so many times (1?) we’ll have the energy and excuse to rope everyone in.
As we’ve embarked on this journey, I’ve realized that my community is wide. Heck, I have a Personal CRM I maintain monthly.
But Bridget’s community is deeper.
Why?
- The primary communities that shaped us are: family, childhood, high school, college, work, and extracurricular. I think the latter two account for the main difference between us. When I joined a small tech startup, I decided that my coworkers would be my friends. Stripe had a Sunday test- if a candidate were alone in the office on a Sunday, would that make you more likely to come in and want to work with them? For startups, colleagues have to genuinely enjoy each other’s company and share a common drive. And with extracurriculars like playing organized sports or joining a (community service) fraternity, I’ve also been more embedded in larger communities with shared interests.
- I’m naturally introverted but I’ve always been a social chameleon. A close friend calls me the “normal guy” from high school. That’s a mild insult at best. But at a math and science boarding school, it enabled me to move between social circles in a way that my hyper-intelligent but eccentric peers couldn’t.
- I like to actively stay in touch and catch up with folks I haven’t seen in a while. I’m quick to reconnect, but I’m not as inclined to regularly talk with those closest to me. Sorry mom and dad! Bridget is infinitely better at continually keeping in touch with smaller, close groups.
From a professional standpoint, there’s been two times in my product management career where I’ve felt like I had a community to really push me:
- Persona’s Head of Product joining in 2021- working with the CEO is invaluable for aligning on product vision, but there were many more topics (e.g. how to scale up the product functions) we didn’t have the experience — or in the CEO’s case, time — to unpack together. So when the Head of Product joined, we built rapport quickly because we were able to navigate way more ambiguity by not only bouncing more ideas off each other but also having deeper discussions.
- Joining the Skip Community in 2023- after every one of our monthly meetups this year I’ve thought to myself, “these aren’t everyday conversations I can have at work.” In one meetup, we advised startups on their product challenges- something I’d love to do more of. It was eye-opening to see how a dozen different product leaders answered questions about hiring. There were starkly different answers but they were all well-reasoned and insightful.
Meanwhile, the Persona community continues to grow too- we’re almost 300 strong now. There’s so many new faces and this was the first year where I legitimately felt anxious about introducing myself, compounded by what I call the Cumulative Awkwardness dilemma- you pass by someone everyday which only makes it harder to introduce yourself the next time.
There’s two small ways I’ve tried to fight the awkwardness:
- Move desks- the 14th floor of our office is primarily engineers/PMs/designers; while the 13th floor is primarily sales/post-sales/marketing. I relocated to the 13th floor to break out of the bubble and interact with coworkers I normally wouldn’t, with a goal of meeting one new person each week.
- Support early career folks- this year we ramped up heavily on entry-level hiring across the business functions. I wanted to make a concerted effort to bridge generations by getting to know the new hires and acting as a mentor to set an example that anyone at Persona is accessible.
When I reflect on my personal and professional communities, I wonder how can I make these communities more effective and closer?
I read a book called “The Art of Gathering” that inspired the above question and a host of ideas I’m excited to experiment with.
Personally, I want to use our wedding as an opportunity to not only spend time with ones we love, but also for fostering stronger connections between our guests.
Professionally, the age-old question is how to get more out of meetings. Running effective meetings is arguably my most important skill, and I’d love to keep up-leveling.
On code -> culture
For the past couple years, the main challenge of building Persona has shifted from product (i.e. code) to process and people (i.e. culture).
How do we operate, communicate, and grow?
Product
The biggest challenge is not whether we’re building the right things and winning in the market anymore.
Over six grueling years in a rapidly changing market, we’ve proven that we deeply understand the market and know how to build great products.
Rather, the challenge now is playing defense. When we were small and irrelevant, we could just play all-out offense.
This year, I felt the first tug of playing it safe — not to lose.
As we’ve established ourselves as a market leader in identity verification by prominent industry voices, we’ve needed to start playing defense.
Being recognized as the leading product was a massive accomplishment, but competitors have subsequently ramped up their efforts to attack us by cherry-picking examples and blasting us on public forums.
Compliance and stability are other great, more silent examples.
As we work with larger companies, geographical scale, industry variance, and legal implications ramp up exponentially. We could be staying on top of 99% of relevant regulations, but missing the 1% at scale is the difference between winning and continuing to work with enterprise customers or not.
Similarly, the baseline expectation for selling to any enterprise is that your service will be available. Any downtime of critical-path software could mean hundreds of millions of dollars lost.
Process
I’ve talked about embracing speed and chaos, which I still believe and practice.
However, I’m also increasingly aware of and compelled to address the tradeoffs of that mentality.
Over time you see two trends:
- The speed of shipping goes down as you constantly revisit products that were shipped too fast.
- The quality of the product declines and becomes increasingly inconsistent across teams.
I’ll share some unique factors at play and what we’re doing about it.
Factors
- Remote- on the whole, we’re not much different from other tech startups. We’re neither fully remote nor in-person, and our engineering team is more distributed than other functions. But the specific teams I work with daily are primarily remote, and being able to quickly tap folks on the shoulder significantly lowers the barrier for seemingly less urgent conversations, which is exactly what product polish tends to be.
- Tenure- an absurd number of early hires are still at the company (which I love), but it’s triggered many second and third order effects. As it relates to process, many managers and leaders — myself included — who are supposed to be focused on establishing process and scaling teams are often overloaded doing individual contributor work because: 1) that’s how they started at Persona, and 2) now they have too much context to not chip in.
What we’re doing about it
Over the past year, we’ve pushed a “Unify” narrative. Instead of pursuing different business and product goals at all costs as startups within a startup, we want the different products to be consistent and work together as a seamless platform.
I always perceived the narrative as focusing on outputs (i.e. what product we build), but I’ve realized it’s actually about inputs (i.e how we build and do the hard things).
One example is product polish. It’s impossible to prioritize small improvements when your biggest customers are banging on the door for features they wanted months ago.
Another example is cross-product implications. It’s easier to blindly ship your own feature than to put yourselves in other teams’ shoes, gather context on product surface areas you’re unfamiliar with, and chase down stakeholders, only to have seemingly created more work for yourself when there’s inevitably push back.
But that’s the crux! It should feel ever harder to glue everything together as we evolve from a startup to an established company. The most impactful Personerds constantly embrace rather than shy away from the hard things.
On the less fuzzy, more tactical side:
- Leverage AI- see On AI 🙂
- Invest in supporting functions- while we started building a revenue ops team (e.g. set up our Salesforce instance) a couple years ago, we just started building out dedicated ops teams for other functions like marketing and product management.
- Systematize documentation- we have a strong culture of documenting, but we need to do a better job building systems that can accurately point to the relevant documentation.
- Establish key metrics with robust monitoring- one of my regrets is relying too long on excuses like “we don’t have statistically significant data” and “we’re data-informed, not data-driven.”
- Train launching as opposed to just building- we now spend more time rolling out and launching products than actually building them, yet most of our training and discussions still revolve around how to build faster.
People
2024 was the first year we had true performance reviews.
Besides reminding us that each individual is unique and we should put them in environments that accentuate their strengths, it also reminded me of the many challenges and learnings that come with evaluating performance, including my own.
- The more people, the more variance in expectations and performance (e.g. junior vs senior, role specialization, low vs high tenure, organizational structure, career aspirations, management styles). Regardless, the golden rule is to err on the side of too high versus too low expectations.
- Prioritization is as much about people and performance as it is about product. We’re just as likely to work on a feature because a high performer is at risk of burning out and needs to work on a passion project, as we are to work on the most requested feature from customers. The former can still be what’s best for the business, but it’s less obvious or explicit.
- Every day, I get further from the product details. And when I do have to dive into details, it’s usually because something’s wrong. How am I supposed to evaluate my team’s performance if I don’t understand the details? One of the most important tactics to combat this paradox is to dogfood the product whenever I can.
- High tenure is again both a blessing and a curse- many of the highest performers are folks who’ve been at Persona a long time. But past success isn’t a guarantee of future success. Similarly, expectations for new hires shouldn’t be capped by their previous work experience or even their previous month at the company- different environments call for different expectations.
- How about my own performance? A coworker once told me that my receptiveness to feedback and bias for corrective action was a superpower. However, in my position, I so willingly give feedback to others but receive so little of it myself. Yet, to advance in my career, I’m supposed to get 360° feedback to identify and shore up my weaknesses.
- My personal conundrum is that I sometimes wish people would expect less of me to counterbalance my own relentless drive to expect more of myself. It’s a tactic I’ve developed over the years — alongside more common ones like setting boundaries — to mitigate repeated burnout, which is ultimately fueled by one’s own expectations.
On morale
👍 On one hand, Persona is doing really well. We exceeded our revenue goals again, and we’re about to hit milestones only 1% of venture-backed startups ever hit.
And at a macro level, the rise of AI undoubtedly amplifies the importance of identity- are you really who you say you are? We feel the tailwinds already, and I expect those winds to grow stronger in coming years.
👎 On the other hand, morale isn’t as high as one would expect given our success. Every year is harder than the last, but this year felt especially hard.
Why?
This paradox has been the source of countless conversations and sleepless nights, but I’ll summarize:
- Powerlessness- many of the company’s biggest challenges affecting everyone can only be realistically solved by a much smaller subset of people.
- Scale- we’re processing tens of millions of verifications every day and working with companies many orders of magnitude bigger than we are.
- People- there’s more churn and attrition compared to years past, and the core of the company is aging (e.g. the difference in energy between tenured and newer folks widens, important people are out on parental leave).
- Political environment- highly polarized and sensitive events silently but heavily weigh on us mentally (e.g. Israel/Palestine, US elections).
- Fraud- as much as AI is a tailwind, it’s also accelerated fraud at enormous scale. Creating fake IDs or deepfakes of stolen identities is 10x easier and cheaper than it was before.
I don’t have many clear cut solutions, hence why this is the problem of 2024.
But I’m optimistic that the problem is fueled by a potent mix of short-term factors and that the underlying strength of the business will ultimately win out.
😟 My biggest worry is actually that low morale becomes a self-fulfilling prophecy.
Because it’s the topic we talk about most, I sense my mood dampened by default and an over-sensitivity to related problems. Combined with my already unreasonably high bar, it’s easy to feel like the company is crumbling.
I continually try to remind myself and other leaders at the company that our negativity or positivity is infectious and amplified. And I’m uniquely bad at being positive- just ask my dog!
💪 We shouldn’t act like everything is fine and be tone deaf by over-celebrating success- these are really difficult problems to solve. But we also shouldn’t ignore that it’s inherently hard to build meaningful and successful companies. If it was easy, everyone would do it.
On families
Since Bridget’s parents live in San Jose, I’ve been able to spend a lot of time with them over the years. Not to mention, they’re unequivocally our first choice dog-sitters!
One experience we hadn’t shared together yet was going on a family vacation; we’d already traveled to Hawaii with my family back in 2021.
This year her parents celebrated their 45th👏 wedding anniversary with a family Rhine River cruise in Europe over the summer.
Fun fact: it ran right through where I had studied abroad just over a decade ago!
Needless to say it was a blast and I’m grateful we were able to celebrate together.
Plus, spending a highly concentrated period of time with them always affords interesting insights about family and cultural dynamics (see Great Big Christmas 🎄).
And this year was particularly interesting because we also traveled with my parents in Asia to attend my cousin’s wedding at the end of the year, so we had recent family vacations to compare and contrast.
McAnany(s)
- At the dinner table, it’s common for multiple conversations to happen simultaneously. I often see one group talking about topic A suddenly, yet seamlessly, pivot to another group’s topic B. Which brings me to a related observation- interruptions are practically a family sport. Whether it’s a comment, quip, or probing question, don’t expect to finish your story in one go. You may get asked the exact or a variant of the same question, or hear the same story repeatedly because another conversation was happening or someone just sat down.
- On a broader level, it’s clear to me that the McAnanys are simply curious and love to talk- traits I admire deeply because they know so much more about each other. Bridget knows every detail about how her parents met, while I just learned some basic details of my parent’s courtship last week- because Bridget asked, of course!
- Each McAnany has a strong personality, wildly different preferences, and isn’t afraid to express their opinions. There’s always some debate or disagreement but you’ll also always know where each person stands- nothing is hidden or silent.
Tsao(s)
- Instead of communicating with words, we communicate with food- a whole lot of it. When you look closely, food tells a story about our values. Chinese food is incredibly time intensive to prepare- you mince, chop, slice, and marinate everything. We don’t just eat food- we take pride in preparing intricate dishes for hours, comprised of dozens of ingredients. Cooking requires patience and discipline. We eat communally as a group- dishes are shared on lazy susans and the elders are served by others and eat first. At a table of 12 adults, there’s only one real conversation happening and some folks may largely be silent during dinner. We get three orders of a dish we didn’t want simply because we can’t miss a deal. Even the fight for the bill, which can get hilariously contentious, is an act of contributing back to the wider group.
- In general, gifting is a major part of the culture. We brought back way too many snacks from Asia gifted to us from family and friends. I’m admittedly not the best role model- I’m terrible at both giving and receiving gifts. The one thing I’m any good at is paying for the bill, which I thankfully learned from my mom.
- After we eat, we like to go on meandering walks to help with digestion. Bridget despises any sort of physical activity after eating, while I’m happy to go for a jog. In Asia, we walked through dozens of malls because they were all over the place- and since that’s also where all the food was, we naturally wandered through the megaplexes after meals.
While Bridget and I occasionally find ourselves annoyed by some of these cultural differences, both of us agree that discovering and bridging the gap between our families is an incredibly rewarding and meaningful process.
On marriage
Bridget doesn’t like surprises.
Don’t talk to her about the time I made her trudge up Mount Tallac in Tahoe but woefully underestimated how long and hard the hike would be.
So when I planned to propose on an August weekend, I needed to give her enough signal that it was happening so she could do her nails, pack the right outfits, and just be mentally prepared for the big moment. But I wanted to sprinkle in spontaneity too: where and when I would pull out the ring, how I might incorporate the dog, only making a dinner reservation.
I ended up picking Mendocino, a three hour drive north, and a place that holds special meaning for us since it was one of our first real weekend getaways after COVID lockdown. Despite the predicable coastal fog and one nerve-wracking hour where I was desperately searching for the imaginary perfect cliff to propose, it went swimmingly!
We got some beautiful photos to cherish the memory.
I assumed that getting engaged fulfilled my 2024 duties as a good partner, but it wasn’t to be.
We also got legally married so Bridget could get healthcare, and secondarily because I hated the idea of signing and submitting paperwork as another to-do of an actual wedding.
I’m happy we got that out of the way because I was getting seriously confused about what it meant exactly to “get married within 90 days”.
Thankfully for us, it just meant getting a marriage license, asking Bridget’s mom (in her case, a religious officiant) and dad (i.e. a witness) to sign said license at our kitchen table, and then submit it to City Hall which is conveniently 300 feet from our front door.
We ended up being extra happy we got married though, because we’re saving a bunch of money on taxes too.
Honestly, it hasn’t fully sunk in yet that I have a whole other side of the family now, even though I’ve already psychoanalyzed the crap out of them.
Or for that matter, that I have a wife. Feels weird to even write it, even though I had no problems yelling at the TSA agents last year that my wife had been stuck at security for ten minutes and we were going to miss our flight.
Now that we’re both engaged and married, we’d really like to actually celebrate with more than just Bridget’s parents!
Fortunately, Bridget has made wedding planning her full-time job and we’re way further ahead than I thought possible after three months. It’s quite a luxury to skip all the hours of research and simply be presented with the best three options, leaving only the choice to be made.
Additionally, after attending almost two dozen weddings together, we’ve already pre-decided many aspects of our own by identifying the elements we liked most from others.
From my perspective, decisions are being made lightning fast!
Again, I’m just incredibly appreciative of the disproportionate share of planning Bridget’s taken on. That’s on top of everything else she’s already taken on more of — like making dinner and walking the dog — while being incredibly accommodating of my schedule during her break from work.
Conclusion
I haven’t written pithy conclusions in previous posts because it’s hard to pull out a top-level theme from a wide range of topics.
But there was a actually a bit of a theme this year— duality.
- On AI: excitement vs skepticism
- On community: wide vs deep
- On morale: optimism vs pessimism
- On code-> culture: speed vs quality
- On families: loud vs quiet
- On marriage: spontaneity vs planning
- Bonus: Progressive vs conservative (Note: politics isn’t really my thing and this wasn’t a section, but I highly recommended What’s our Problem?: A Self-Help Book for Societies from one of my favorite bloggers)
After all, it’s very me to balance the tradeoffs and embrace the gray area.
Memorable moments
As is tradition, here’s an overly simplistic collage of more highlights from 2024 to wrap up.
- EDC for the first and last time
- Family cruise down the Rhine
- Cousin’s wedding in China
- Cabo and Oaxaca for a wedding (plus making mole)
- My dad’s 69th birthday (Note: in Chinese culture, birthdays ending in 9 are auspicious and celebrated with more fanfare)
- Dominican Republic with the high school crew
- Money2020 conference in Vegas
- Hans Zimmer Live concert
Wishing you and yours a happy and health 2025!
