LLenny's Podcast
← All frameworks
LeadershipDhanji R. Prasanna

Conway's Law Reorg: Go Functional Before You Go AI

Restructure engineering into one functional org before expecting company-wide AI or technical depth

Difficulty
Expert
Time to result
~months to results
Steps
4
Confidence
92%

Prasanna argues the single biggest lever in Block's AI-native transformation wasn't a tool — it was collapsing independent GM-run business units into one functional organization where all engineers report to one head of engineering, all designers to one head of design. Because 'you ship your org structure' (Conway's Law), siloed teams that don't share tools, policies, or a technical strategy can never drive a company-wide capability like AI. Fix the structure first, and shared platforms, mobility, and technical depth follow.

Origin

Built on Conway's Law (Melvin Conway, 1967) — 'organizations design systems that mirror their communication structure.' Prasanna explicitly credits Steve Jobs' functional reorg of Apple as the same move, noting Block discovered it independently while investigating what AI-native would require.

Core principles

  • 01You ship your org structure — team boundaries become product and technical boundaries
  • 02GM/portfolio structures treat engineers as an interchangeable commodity you add to ship more features
  • 03A singular technical focus requires a single engineering org, not federated silos
  • 04Shared tools, shared policies, and a common seniority bar are prerequisites for company-wide capability change
  • 05Find your company's DNA and optimize the structure around it, simply and clearly

How to run it

  1. 1

    Diagnose the silos

    Map where independent units (business lines, acquired companies) run separate engineering, design, and technical strategy. Note where no one is aligned on what you want to be in five years.

    Pro tip Look for the tell: teams that share only 'foundational resources like legal' but nothing technical.

  2. 2

    Reset identity to a technology company

    Before restructuring, correct the 'identity drift' — stop describing yourselves as a financial/real-estate/whatever company and reassert engineering-and-design-first DNA.

    Pro tip Seed the spark with visible signals: company-wide hack week, cross-org IC summits, small special-projects teams.

  3. 3

    Collapse into functional reporting lines

    Move every engineer under one head of engineering, every designer under one head of design, with shared policies and a common seniority definition.

    Pro tip A given senior-engineer level should mean the same thing across the whole company so people can move to areas of need.

    Watch out Not everyone will be on board — Prasanna calls it 'quite a painful transformation.' Expect resistance from units with momentum in their silo.

  4. 4

    Reallocate by leverage, not headcount

    Stop adding engineers as a function of feature count. Instead ask where you can build depth, reuse platforms, and modularize across the now-unified org.

    Pro tip Reframe planning around 'areas of optimization' and 'where can we build depth' rather than 'how many people to build product X.'

In the wild

Block's GM-to-functional transition

Block had spun Square, Cash App, Afterpay, and Tidal into a portfolio of GM-run companies with separate engineering and design teams and their own CEOs reporting to Jack Dorsey. No one talked across silos or aligned on technical strategy. Prasanna collapsed them into single functional engineering and design orgs sharing tools and policies.

Teams now 'speak the same language,' share tools and policies, a seniority level means the same everywhere, and engineers can move between teams — which Prasanna calls 'the key to making our transformation into being more of an AI native company.'

Common mistakes

Throwing AI tools at a siloed org and expecting transformation

Without a functional structure, teams have no shared tools, policies, or technical strategy, so AI adoption fragments and never compounds company-wide.

Treating engineering headcount as a commodity

The GM incentive to 'just add more engineers to build more features' is the mythical-man-month trap; it prevents the leverage-and-depth thinking a functional org enables.

Is it for you?

Best for

CTOs and eng leaders at multi-business-unit or post-acquisition companies trying to drive a company-wide capability (AI, platform) but blocked by federated silos

Not ideal for

Early-stage single-product startups that are already functionally organized, or genuinely independent portfolio holdings with no shared technical purpose

From the transcript

it's the law that basically says you know you ship your org structure

12:00

all engineers report into one single team. Now all designers report in one single team and there's single head of engineering, single head of design

10:30

that meant we could really drive forward AI, we could drive forward platform and just technical depth generally

10:30

this is what Jobs did when he came back to Apple as well. He reorganized Apple to be functional

11:00

in our GM structure our incentives were always to think of engineering headcount as a commodity

44:30

From the episode

How Block is becoming the most AI-native enterprise in the world

Dhanji R. Prasanna