Question the Base Assumption Before Build-or-Buy
Before building OR buying a tool, ask whether the process needs to exist at all
- Difficulty
- Moderate
- Time to result
- ~days to results
- Steps
- 4
- Confidence
- 86%
When AI makes it trivial to build internal tools, teams polarize into 'buy the vendor tool' vs 'we can build it with Goose in half the time.' Prasanna's move is to reject both by default and apply portfolio judgment: question the base assumption — do we even need this process? Often changing or eliminating the process beats both building and buying, and AI is specifically bad at this global, cross-company judgment call.
Origin
Prasanna's framing at Block; he explicitly links it to Elon Musk's optimization process ('do we even need this thing before we start optimizing it') and to the classic 80/20 problem.
Core principles
- 01The build-vs-buy debate hides a prior question: is this process necessary at all?
- 02Cost savings from in-house builds are usually not worth the mental bandwidth and technical focus lost
- 03Maintenance is the hidden cost — 'built it in a weekend, now it's years of endless maintenance'
- 04AI cannot make portfolio judgments about what globally matters — that's a human job
- 05Eliminating a process shrinks your surface area (e.g. less to secure) more than optimizing it
How to run it
- 1
Surface the real request
When a team says 'buy this vendor tool' or 'build it ourselves,' pause both and restate the underlying job to be done.
- 2
Question the base assumption
Ask whether the process even needs to exist. Can you change the process so the tool — bought or built — becomes unnecessary?
Pro tip For security teams especially: 'ask the team that's building it to do it differently or to not build that at all if it doesn't matter and then you won't have to increase your surface area.'
Watch out Infosec teams 'twist themselves into knots' securing things that shouldn't be built in the first place.
- 3
Anchor to core purpose, not dollars
If you must choose build vs buy, decide by whether it serves your core purpose, not by raw dollar savings.
Pro tip 'Keep coming back to the thing that matters to you as a company and then the rest will follow.'
Watch out Pure dollar-vs-dollar comparisons pull you off your core purpose and cost more in lost focus than they save.
- 4
Weigh the maintenance tail
Price in the long tail of edge cases and ongoing support before committing to build in-house.
Watch out The 80/20 trap: the first 80% ships in a weekend, the last 20% of edge cases takes far longer.
In the wild
One team wanted to buy a new vendor tool because their current one lacked features; another said they could just build an equivalent app with Goose in half the time. Prasanna's response was to ask whether either was necessary — could a process change remove the need entirely?
→ Reframes the decision away from tooling and toward eliminating unnecessary work, especially for infosec teams whose 'surface area' shrinks when the thing simply isn't built.
Block built the entire functionality of Cash Card in roughly a weekend, then spent 'a really long time' ironing out edge cases — someone tipping twice the bill breaking the backend, gas stations billing differently.
→ Illustrates the 80/20 maintenance trap: fast to ship the core, expensive to harden the long tail.
Common mistakes
Framing it as build-vs-buy
Both options assume the process is needed; the higher-leverage question — 'do we need this at all?' — gets skipped, and AI can't ask it for you.
Optimizing for dollar savings over purpose
Replacing a vendor tool to save money can cost more in lost technical focus and mental bandwidth than it saves, pulling the company off its core purpose.
Is it for you?
Best for
Leaders fielding build-vs-buy requests in the AI era where internal tool-building is suddenly cheap
Not ideal for
Decisions where the process is unambiguously core and required, and the only real question is execution
From the transcript
“is any of this necessary like if we just change the process do we even need to think about building tools and this is the…”
“just ask the team that's building it to do it differently or to not build that at all if it doesn't matter and then you…”
“the savings and cost that there might be in replacing a vendor tool by something you build in house is probably not worth it in…”
“Elon has this whole process... one of the steps is like do we even need this thing before we start optimizing and automating it”
From the episode
How Block is becoming the most AI-native enterprise in the world
Dhanji R. Prasanna