Just-in-Time Knowledge Work (Low Work-in-Process)
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- —
Forte imports just-in-time manufacturing into knowledge work. Toyota stopped stockpiling parts and instead pulled them into the factory hours before they were needed. The surprising result wasn't just speed and efficiency but quality: with a stockpile of ten thousand tyres, a defect discovered on the line has already been replicated ten thousand times and nobody speaks up, because what could you do? Small batches surface defects while they're still cheap to fix. Applied to knowledge work, the inventory is your work-in-process and your stockpiled research. Just-in-case learning — study everything perfectly in advance — only works when someone tells you what will be on the test. Outside school, uncertainty is much higher, so you learn what you need when you need it, use it fast, get feedback, iterate.
Origin
Extracted from Deep Dive with Ali Abdaal
How to run it
- 1
Treat unused knowledge as inventory, not an asset
Six months of research on a topic you may never need is a stockpile with carrying costs. Recognise that the just-in-case habit is an artefact of schooling, where a professor guaranteed the material would be used.
Watch out Uncertainty outside school is radically higher — the skill, the topic or the knowledge may genuinely never come in handy.
- 2
Cap your work-in-process
Count how many projects are live at once. Forte's point is that 25 simultaneous projects guarantee quality suffers, efficiency suffers and time is wasted. Deliberately keep the number of things actually in progress low.
Pro tip Keep it low, not necessarily at one — Abdaal notes he works better across a small handful so he can switch when he's the bottleneck on one thread.
- 3
Pull knowledge only when a live project demands it
Start the project, hit the point where you need the skill or the information, and acquire it then. This is just-in-time learning: learn something, acquire knowledge right when you need it, use it fast.
Pro tip Your Second Brain is the supplier network — captured material sits there until a project pulls it in.
Watch out Don't confuse capturing (cheap, passive) with researching (expensive, active). Only the second needs a pull signal.
- 4
Ship the small batch and get feedback
Produce the output while the knowledge is fresh rather than continuing to accumulate. Get the results, get the feedback — that feedback is the defect inspection that a stockpile would have hidden.
Watch out Waiting until the material is 'complete' before shipping is exactly the stockpile behaviour that replicates errors at scale.
- 5
Iterate and pull again
Fold what you learned back in, then run the next iteration. Learn, make another iteration, do it again — rather than growing a huge reserve of material before you even start.
Pro tip Each cycle tells you what to learn next, which is far more accurate than guessing in advance.
In the wild
Forte's illustration: stockpile a thousand or ten thousand tyres, then start feeding them onto the production line. You find a defect. What do you do? You can't return them, so you keep quiet and let them all go through — the single defect has already been replicated thousands of times. Pull the tyres in just-in-time instead and the defect shows up in a batch small enough that fixing it is obvious and cheap. Speed and efficiency improve as expected; the unexpected gain is quality.
→ Small batches make defects visible and correctable instead of economically unspeakable.
Forte ran Building a Second Brain the same way. He never made long-term plans and never thought far into the future — he ran one cohort at a time, and for the first seven, eight or nine cohorts the only goal was to make enough money in this cohort to fund the next one. Each cohort shipped, got feedback, and improved slightly. By the time of the interview he'd run fourteen.
→ Fourteen cohorts, each a little better, bigger and more profitable — with no multi-year plan and no stockpiled preparation.
Common mistakes
Researching for months before starting
A six-month research stockpile is a bet that you correctly predicted what you'd need. Outside a classroom nobody has told you what's on the test, so most of that inventory expires unused.
Running 25 projects at once
High work-in-process degrades quality, efficiency and time simultaneously. The relationship isn't linear — ten parallel projects don't finish in the time of ten sequential ones.
Front-loading perfection
Doing everything as far in advance and as perfectly as possible is a school habit. It only pays off under low uncertainty; in real work it locks in errors before any feedback exists.
From the transcript
“Another way of saying this is keeping inventories low. You don't want a lot of work in process.”
“When you have 25 projects at once, quality is going to suffer, efficiency is going to suffer, time is going to be wasted. So, you…”
“And basically, you can think of it like just-in-time learning. Learn something, acquire knowledge right when you need it, use it fast.”
From the episode
Confronting my Productivity Guru - Tiago Forte
Tiago Forte