Reality leaves clues.
Better decisions begin by noticing them.

I want to tell this as a thread, not a timeline, because the timeline is unremarkable and the thread is the only part of it worth writing down.

It starts before I ever touched a rock. My father and grandfather were kirtankars — men who spent their lives reciting and singing within the Varkari tradition, a path built largely by farmers, labourers, and ordinary householders walking together toward Pandharpur, asking for nothing more elaborate than an honest look at oneself along the way. There is no ornament in that tradition, no promise of a shortcut, no pretence that the work can be skipped. You walk, you sing, and you sit with the question of who you actually are, stripped of whatever you've been performing for everyone else. I grew up watching that happen around me long before I had any language for what it was teaching me.

Years later, we built a small Varkari education centre in Pune, one that still teaches without charging fees. It had been my father's vision long before it was anything I helped with; what I got to do was help make it real. I mention it not as a credential but because it was the first time in my life I helped build something whose entire purpose was to give people something useful without asking them to pay for the part that mattered most. I didn't know it yet, but that instinct would end up underneath everything I've built since.

I didn't arrive at geology through any early calling. I arrived at it the way most people arrive at a degree — through interest, circumstance, and the slow discovery that I was actually good at something I hadn't chosen for poetic reasons. What I found, once I was deep enough into it, was a discipline built on the same refusal I'd grown up watching: don't trust the surface. The surface lies, gently and without meaning to. You have to go underneath it to find out what's actually true. I happened to find a profession that asked the same question of rock.

Varkari devotion taught me to distrust the surface. Geology gave that instinct a method.

My first real proving ground was on one of India's high-speed rail projects, as a construction geologist. It was the first time the discipline I'd studied stopped being academic and became a responsibility other people's safety actually depended on. You learn things on a project of that scale that don't transfer cleanly into words — what it costs, in attention and humility, to be the person whose read of the ground beneath a structure has to be right. Not approximately right. Just right. I carried that standard with me afterward, mostly without noticing I was doing it.

After that, I spent a long stretch of years analysing technical documentation from mining companies scattered across the world. It was a strange vantage point — rarely on the actual sites, but reading, report after report, how an entire industry made its highest-stakes decisions, and what it told itself when the data was inconvenient. I began noticing, at a scale I couldn't have learned any other way, that good outcomes and good intentions are not the same thing, and that almost every expensive failure traced back to someone choosing not to look closely enough, early enough, at something they had every chance to check.

Eventually I left that desk. I had spent the better part of a decade analysing other people's ground risk, and increasingly other people's life decisions, on someone else's behalf. What I actually wanted was to put that same discipline directly into the hands of the people who needed it, instead of leaving it inside a report only a specialist would read. That is the honest origin of the two things I am building now — one concerned with the ground beneath a piece of land, the other with the ground beneath a decision a person is about to build the next several years of their life on. They look unrelated until you trace the thread back far enough to see they are not.

So when I'm asked why I build, the truest answer isn't "I'm a geologist who became an entrepreneur." It's narrower than that, and older than my degree. I grew up around people who refused to accept the first, most comfortable story about who they were. I happened to find a profession that asked the same question of rock. I spent years learning what it costs when nobody asks that question early enough. At some point I stopped being satisfied analysing that gap for other people and decided to spend my time closing it directly, for whoever is willing to use it.

Every catastrophic failure I have studied closely — across years spent reading technical reports from mining companies, working as a construction geologist on one of India's high-speed rail projects, and sitting through more land due-diligence conversations than I can count — failed at execution. That is the story the postmortem tells. The wall cracked during construction. The mine flooded during operation. The company ran out of money trying to grow too quickly. Execution is where the damage becomes visible, so execution gets blamed.

Almost none of those failures were actually caused during execution, though. They were caused earlier, in a decision nobody flagged as a decision, made by someone who was probably in a hurry, who probably meant well, and who likely never knew they were setting a price that someone else — often themselves — would pay months or years later.

In geotechnical work you learn this the hard way, because the ground does not forgive optimism. Before anyone pours concrete, there is a site investigation: boreholes, soil testing, groundwater readings, an attempt to understand what is actually down there before anyone builds on top of it. Done properly, it is slow and unglamorous, and produces nothing worth showing off — just a report full of caveats and recommendations for further investigation that nobody outside the technical team will ever read.

Done in a hurry — fewer boreholes than the site really needs, sampling at convenient points instead of representative ones, a timeline too tight to wait for a proper groundwater reading — it produces a report that looks the same. It is simply wrong. And the trouble is that nobody finds out for a long time. The investigation phase ends, looking complete. The design phase begins, looking confident. Construction begins, looking like progress. It is usually a year or two later — once a foundation has settled unevenly, once a wall has cracked, once water has found its way somewhere it was assured it wouldn't — that anyone goes back and asks what the original investigation actually checked. By then, the cost of correction is nowhere near the cost the investigation would have been.

The mistake was never made on site during construction. It was made at a desk, months earlier, when someone decided a handful of boreholes were probably enough.

I've watched the same shape of mistake happen to people buying land who had done everything they were told mattered — clean paperwork, a promising location — only to discover, a year or two after taking possession, something a proper assessment of the ground would have told them on day one. By the time it surfaces, they've already paid for the land, and often already started building. The mistake wasn't made when the first crack appeared. It was made the day they signed, when nobody asked the one question that actually predicts whether a piece of land can be built on safely: not whether the paperwork was clean, but whether the ground itself was telling the truth.

It holds for decisions that have nothing to do with land, too. People rarely fail at the job, the relationship, or the move to a new city. They fail at the decision to take the job, enter the relationship, or make the move — usually for reasons never quite examined: a title that sounded impressive, a fear of disappointing someone, a comfort mistaken for purpose. What looks like the execution failing, two or three years in, is really the site investigation finally happening, except now the cost of finding out is a resignation letter or a difficult conversation, instead of an honest hour spent asking the question earlier.

Pre-execution diligence is hard to value for a simple reason: it produces no visible deliverable in the moment that matters. A thorough investigation looks identical to a rushed one until something goes wrong. There is no applause for the question that prevents a disaster that, because you asked it, never happens. The reward for doing it well is an absence — nothing went wrong, against a baseline nobody can see.

The mistakes that ruin projects, companies, and decades are rarely failures of effort. They are failures of timing: looking at the right question after the cost of being wrong has already been set. The harder, and more useful, discipline is finding the cheapest possible place to be wrong, and choosing to be wrong there, on purpose, before anyone has to pay for it.

For close to four years, a meaningful part of my working life was spent reading other people's findings about the ground — not from one project, but from mining companies scattered across continents, each trying to tell the world, and more specifically their investors, what they believed was beneath a particular patch of land. Exploration updates, resource estimates, feasibility studies: a great many of these passed through my hands. It is an odd kind of expertise to develop. I never set foot on most of those sites. What I learned, I learned from how people write, what they choose to leave out, and what they decide to say first.

I wasn't looking for lessons about decision-making. I was checking technical accuracy — whether a resource classification was actually supported by the data behind it. But after enough reports, certain patterns kept recurring, until they stopped looking like coincidences and started looking like the actual subject I'd been studying the whole time, hiding behind the geology.

The data is rarely the problem. The story built on top of it usually is.

A geological report is, underneath the technical language, an argument: here is what we measured, here is what we believe it means, here is what we recommend doing next. Almost every report carried, somewhere near the back, a section quietly listing its own uncertainties — limited drilling, assumptions about continuity between sample points, places where more data was needed before anyone should commit serious capital. The summary built on top of that same report nearly always read with more confidence than the report itself supported. Not dishonestly, usually. Just selectively. The caveats were real. They were also the first thing to disappear once the findings had to become a story for people who would never read past the summary.

Confidence in language seemed to rise exactly as the data got thinner.

Early-stage exploration reports — written when a company had drilled only a handful of holes and was mostly hoping — tended to read with more certainty than mature feasibility studies written years later, after far more drilling. There's a kind of sense to it: early on, there isn't enough data yet to argue with the story. By the feasibility stage, there's enough data that the writers are forced into precision, and precision is humbling. The pattern seemed to generalise further than geology. The moment a project sounds most certain is often the moment it has the least information to be certain about.

Sunk cost shows up in rock the same way it shows up in people.

Once a company had drilled dozens of holes and spent years telling its shareholders a particular story about what was in the ground, the report recommending a pause — or worse, a stop — became, visibly, the hardest document anyone on that team would write. I read reports where the data, read plainly, pointed toward caution, and the recommendation still found a way to keep going. Not usually through dishonesty. Through the very human difficulty of being the person who tells a room of people who've already spent years and a great deal of money that it may not have found what everyone hoped.

The teams willing to pay for bad news early seemed to be the ones still standing later.

A quieter pattern, the one I came to trust most: the teams who commissioned the unflattering study — the assessment nobody wanted to pay for, the independent review that might come back saying "not yet" — tended to be the same teams whose later filings kept appearing, rather than going quiet. The teams that treated due diligence as a box to clear, rather than a real question to answer, tended to be the ones I'd eventually stop hearing from. I can't claim certainty about why any single project succeeded or failed. But across enough reports, paying for bad news early was the one habit that kept correlating with projects still around to file another report.

A report is not a decision. It only informs one.

This is the one that mattered most, and the one I nearly missed because it seemed too obvious to be worth noticing. A perfect report changes nothing by itself. Someone still has to read the inconvenient finding, feel its weight, and choose to act on it instead of around it. More than once, a technically excellent report sat beside a decision that looked past its own central finding. More than once, a modest report sat beside a decision that took every caveat seriously. The quality of the document and the quality of the decision built on top of it turned out to be two different things, only loosely related — and the second one was always the one that decided what actually happened.

I didn't come away from those years with strong opinions about rock. I came away with a long memory of how often the truth was already sitting in the file, stated plainly, available to anyone willing to let it change their mind — and how rarely that was enough.

A geologist's job, stripped of the technical language around it, is to read a record that has already been written and is simply hidden from view. The landscape above ground tells you almost nothing on its own — a gentle slope, a flat field, a calm stretch of river. What you learn to do, slowly, and by being wrong often enough to stop trusting it, is to set that surface aside and look for the record underneath: the borehole log, the water table sitting at a depth nobody can see by standing on top of it, a fault line below that the topsoil has no opinion about.

Core samples make this strangest and clearest. A column of rock pulled up from underground, laid out tray by tray, each segment sometimes only centimetres apart in space and millions of years apart in time. Occasionally one layer sits directly against a layer many millions of years older, with everything in between simply missing — eroded away, or never deposited at all, gone without leaving a trace. Geologists call this an unconformity. It sounds, the first time someone explains it to you, like a technical curiosity. It becomes, after enough of them, the most honest metaphor I know for how much of any history — a piece of land's, a company's, a person's — goes quietly unrecorded, while the layers on either side of the gap look perfectly continuous to anyone not looking closely.

Calm topsoil can sit directly above ground that won't hold weight.

This way of looking is trained, not innate. You learn that what's visible at a glance carries almost no information about what actually determines whether anything built on top of it will stand.

Once the habit sets in, it doesn't stay confined to rock. The same shape of mistake shows up everywhere, made by people who have never touched a geological hammer. A company looks stable right up until the one client, or the one regulatory grace period, or the one person's energy quietly holding it up gives way — and the collapse, looked at honestly afterward, was visible as a slow shift for a long time to anyone checking beneath the numbers. A person looks fine right up until they don't, and the people closest to them are often the most surprised, because they were reading the same calm surface everyone else was reading, never the layer underneath that had been moving for a while.

None of this means the surface is meaningless. It's the first clue, not the whole answer — the way a geologist still walks a site before drilling a single hole, because what's visible tells you where to look harder, not what you'll find once you do. The mistake isn't noticing the surface. The mistake is stopping there, mistaking the most recent page for the whole record.

I don't think this habit of attention arrived with a geology degree. It feels older than that — closer to a question I grew up around long before I had a vocabulary for it: what is actually here, underneath what I've been told, or what I'd like, to believe is here. Geology simply gave that question a method, and a place to practise it, on ground that doesn't care what story you'd prefer to be true.

Most advice about decisions optimises for one of two things: speed or confidence. Move fast. Trust your gut. Decide and adjust. There is real value in that, in the right context. But after years of watching expensive decisions go wrong — in geotechnical reports, in mining projects I had no stake in, in land transactions, and in the slower, quieter failures people carry around in their own lives — I've come to think both speed and confidence are the wrong things to optimise for first. Something has to come before either one, or neither is worth much: reality.

I think of this as a framework rather than a belief, because I wanted something I could actually run a decision through, not just something to feel right about. Six principles, in no particular order of importance, though the first is the one the rest depend on.

1. Real problems before real systems

Nothing gets built well that starts from a solution looking for a problem to attach itself to. The problems worth solving reveal themselves slowly, by sitting with something unglamorous long enough to see its actual shape — not the version that's convenient to solve, the version that's actually there. The system, the plan, the structure: all of that comes second. The problem has to be real first, or nothing built on top of it will be.

2. Verification is always cheaper than correction

This is close to a law of nature in geotechnical work, and I have never seen it fail to generalise. The earlier you check something, the cheaper the truth is. A borehole costs a fraction of a foundation repair. An honest conversation costs a fraction of a year spent in the wrong job. Almost every expensive mistake I have studied closely could have been caught earlier, for less, by someone willing to ask an inconvenient question before momentum made the answer costly to hear.

3. An untested assumption is a liability, not a fact

Engineers don't build on an unverified load calculation; they treat it as an open risk until it has been tested and closed. I try to hold other assumptions to the same standard. "This will work." "They will change." "People want this." None of these become facts simply because they feel true, or because believing them is convenient today. They are claims, and claims carry risk until checked against something outside your own head.

4. Build at the pace reality allows, not the pace ambition demands

Ambition wants everything moving at once. Reality moves at the pace of evidence — a conversation, a working prototype, a signal that has actually been earned rather than assumed. Patience here is not the same as caution. It is closer to refusing to mistake motion for proof.

5. "I don't know yet" is a position, not a failure

The hardest sentence to say in any high-stakes room is "we don't have enough information to be sure." It sounds like weakness. In the decisions I've watched hold up over time, it was almost always the strongest sentence available — the one that bought time to find the truth before committing resources to a guess dressed up as a conclusion.

6. The truth doesn't ask for your belief

Reality doesn't shift because you want a different outcome badly enough, or because changing your mind now would be awkward given how much you've already said. It only changes when you actually look at it, whether or not you like what you see. Most of the discipline here isn't intelligence or courage in any dramatic sense. It's the smaller, harder habit of paying attention, even when a more flattering story is sitting right next to it, free for the taking.

None of this guarantees a good outcome. It guarantees only that the decision was made with eyes open — which is the one part of any outcome that was ever actually within reach.