Chase the Schleps
The best opportunities tend to hide behind work that looks tedious and unpleasant. PG has a name for what keeps us away from them: schlep blindness. We don’t consciously turn down the hard, unglamorous problems. We just don’t see them. They get filtered out before we ever weigh them as real options. PG argues that a company is “defined by the schleps it will undertake,” and that idea had a lot to do with the problem I picked.
I think schlep blindness is mostly about status. The problems people talk about at conferences, the ones that get funded on a first meeting, are the ones that sound clean: an elegant algorithm, a consumer app with a satisfying demo, anything you can describe at a party without watching someone’s eyes glaze over. Schleps are the opposite, full of messy inputs, domain jargon, and unglamorous plumbing. So the ambitious crowd drifts toward the clean problems and leaves the messy ones alone, which is exactly why the messy ones stay open.
There’s a caveat worth stating plainly, though. Not every unpleasant task is a hidden goldmine. Plenty of tedious work is just tedious. A schlep is worth chasing when three things are true at once: the pain is real and expensive, the tedium works as a moat instead of a dead end, and a properly hard problem sits underneath the boring surface, one that’s finally becoming solvable. Miss any of those and you’ve signed up for drudgery with nothing waiting at the end of it. The point was never to love drudgery, only to find the kind of drudgery that everyone’s blindness has left sitting in front of something huge.
Why construction procurement
Construction runs on documents, and those documents are a mess. A single tender can pull in bids from a dozen firms, each in its own format, each carrying its own assumptions. The expensive risks almost never sit in the priced line items. They hide in cover letters, exception clauses, qualifications, and annexes: the “excludes rock excavation,” the “assumes owner-supplied power,” the provisional sum someone carried quietly against a scope nobody has pinned down yet.
And in 2026 teams still compare all of this by hand. Somebody opens ten PDFs, builds a spreadsheet, and burns days trying to line up numbers that were never meant to line up. Often enough the lowest headline bid turns out not to be the lowest bid at all, once you account for what each firm quietly left out. Getting that wrong costs real money, sometimes far more than the software that could have caught it.
Not many people want this work, so there’s much less competition for it than a problem this size should attract. That gap is the whole reason to chase a schlep.
The hard part isn’t the PDFs
You might assume the hard part is reading PDFs. It isn’t. Pulling text off a page is close enough to solved. The real difficulty is making bids comparable in the first place.
Comparability means getting everything onto a common basis: reconciling scopes that don’t match, catching the exclusion one bidder spelled out while another simply left it off, turning unit rates, lump sums, and allowances into figures that can sit in one column and mean the same thing. That work is less document parsing than judgment written down, and the judgment is where the moat is, because it’s the part you can’t hand off to a stock model.
It also has to stay honest. Every normalized number needs to trace back to the exact clause it came from, because a leveled bid is only worth anything if you can defend it. When a big discrepancy gets challenged across a table, “the model says so” won’t do. “Here’s the sentence on page 14 of their submission” will. Traceability is what moves a comparison from guesswork to evidence.
None of it is glamorous and all of it is hard. Boring from the outside, huge once you’re in it. That’s the kind of thing I’d rather spend my time on. Schleps aren’t obstacles on the way to the good problems. A lot of the time they are the good problems.
Referenced: Paul Graham, Schlep Blindness.