Workforce datasets are static. A snapshot. A list. A moment in time. But companies are not static. They grow. They contract. They shift role composition. They reallocate talent before revenue changes show up. So instead of building another database, I built a longitudinal company-year panel. ~2.5M normalized U.S. companies. ~387M company-year rows reconstructed from historical experience timelines. Median 7 years of workforce history per company. Not profiles. Not contact records. Company-year intelligence. For each company and each year: • Observed headcount • Growth rate • Role distribution shifts • Structured entity normalization The real asset isn’t volume. It’s the ability to ask: – When did this company actually start scaling? – Did engineering grow before sales? – How did workforce composition change pre-funding? – Which segments show consistent multi-year expansion patterns? Longitudinal structure turns raw records into signal. Investors call it alternative data. Strategists call it market intelligence. AI teams call it training infrastructure. I call it organizational time-series intelligence. Building this in public. #infrastructure #database #patternrecognition

what a time to be alive...

finally made it into the club....🛰️ @Starlink @elonmusk

Make it make sense...🪽

Crypto can squeeze an entire company lifecycle into a weekend. Accumulation. Hype. Growth. Collapse. At Vivameda, we study how companies evolve over decades. Now we’re applying that methodology to crypto. Track wallet behaviour, liquidity and ownership over time. Find comparable trajectories. Test what happened next. And keep the dead tokens. Especially the dead tokens. If your dataset only contains survivors, your model is learning a fairy tale. The real test: could you identify the opportunity BEFORE the move....and actually trade it after fees and slippage? That’s the experiment. I want receipts.

I am using the methodology of a dataset with 70 years of company evolution to train a model on meme coin breakouts....and it works. What a time to be alive.

the 4 hour work week was never more real than now....yet we all fail to make it happen. 🤖

Good stuff! @obie

If the code tells a story, why can't the story tell the code? Vibe coding' just killed an entire industry while bringing it much more to life at the exact same time. Honestly, I don't think anyone has a clue what we've gotten ourselves into.

Everyone is chasing the freshest signal. We spent the last year training on the opposite: career histories going back to the 1950s. Five things that changed how I think about it. Historical data is not a historical record. A profile gets edited today and reads as if it was written in 2019. What you actually hold is a present-day reconstruction of the past. If you do not enforce what was knowable on a given date, your model is quietly reading the answer sheet. Survivors write the archive. Companies that failed leave thinner traces, and the people who worked there tidy them away. Every base rate you compute from history is more optimistic than the world actually was. Sequence carries more information than composition. Two companies can arrive at the same headcount mix and mean opposite things depending on the order they hired in. A snapshot throws that away. Most long-run patterns are regime, not skill. A hiring signal that looked predictive in 2013 was often a signal about cheap capital. More depth means more chances to mistake the macro cycle for something a company did. The gaps are data. Missing years are not missing at random, and they correlate with exactly the outcomes you are trying to predict. None of this makes prediction easier. It makes it defensible, which is what matters the moment a client asks why the model fired.

Prediction markets have a data problem nobody wants to talk about. They are obsessed with now. Current odds. Current sentiment. Current news. Current positioning. But prediction without history is just a very sophisticated opinion about the present. The real edge will come when prediction markets start reasoning longitudinally. What happened the previous 50 times similar conditions appeared? What did companies look like before they failed? Which workforce structures preceded acceleration? Which capability shifts appeared before an industry changed direction? Markets price expectations. Longitudinal data provides reference classes for those expectations. And that's where this gets interesting. AI will make prediction markets faster. More participants will arrive. Public information will be absorbed almost instantly. The obvious informational edges will get arbitraged away. What remains valuable? Data that cannot simply be scraped tomorrow. History. Deep, structured, proprietary longitudinal datasets will become incredibly valuable because they give prediction systems something the crowd doesn't have: memory. Prediction markets may become one of AI's most interesting applications. But the winners won't just be better at predicting the future. They'll be better at remembering the past.

Social Media is dead 💀 Performance Marketing is about to die and the GTM space is more disrupted by AI than any other industry…🫣 You all trained your own competitors for years and now wondering why people won’t pay for your services anymore. It’s about time to change the audience if you want to succeed….📈

VMD.21921.120826.224

We tested whether AI is already leaving a footprint in human capital. Since Aug 2022: High AI-exposure social/content firms: -7.1% workforce Low-exposure firms in the same sector: +11.6% Same clients. Same budgets. Same cycle. The divergence starts in mid-2023 and keeps widening. This doesn’t prove AI caused it. But it raises a much bigger question: What if human capital shows us where AI is taking the economy before the financials do?

To understand where AI is taking the economy, watch what companies do with people.

What determines whether a retailer survives structural disruption? We examined this question using longitudinal workforce data covering 225 major retail companies over a ten year period. Rather than focusing on revenue, valuation, or market share, we measured something more fundamental: the capability profile of each workforce before the outcome was known. The findings were striking. Retailers whose core capabilities could be replicated by digital channels experienced a 5.5 times higher failure rate than those whose workforce was concentrated in capabilities that remained difficult to digitize. The implication extends beyond retail. Competitive resilience is not determined solely by products or financial performance. It is embedded in the composition of the workforce years before the market recognizes the shift. Financial statements describe the consequences of strategic change. Longitudinal workforce data captures the transition itself. This study is part of Vivameda's ongoing research into how organizational capability evolves over time, and how historical workforce structure can be used to identify competitive trajectories long before they become visible through conventional business metrics.

In a world soon dominated by AI, we should focus on understanding Human Capital.

AI has knowledge. Humans have experience. What's the difference? Experience is knowledge over time. A person doesn't become an expert by reading one snapshot. They become an expert by seeing patterns repeat across years, industries, and market cycles. AI should be no different. If a model only sees today's companies, it learns facts. If it sees how companies evolved over decades, it learns behavior. That's the power of longitudinal knowledge. It gives AI something surprisingly close to experience. And experience has always been the foundation of better decisions. The next generation of AI won't just know more. It will have a longer memory.

The future of deal sourcing is in the knowledge before you hunt the signal, not the other way around. We must train our LM like a human, not like a machine. Ground truth knowledge is the future approach to define predictions much more accurate than guessing from snapshots. # longitudinalknowledge

The easies way to predict the future, is to understand the past.