Skip to content
M. K. ELANGO
Architecture 6 min

Your company isn’t slow because your people are slow

It is slow because it is running 1980s CPU architecture in 2026 — and no amount of urgency fixes an arrangement.

The complaint is always phrased as a people problem. We need more urgency. We need better ownership. We need people who move.

Then the organisation runs an urgency programme, and for six weeks things move, and then they do not, and the conclusion drawn is that the people were the problem after all.

They were not. Speed is a property of the arrangement, not of the participants.

Consider what actually happens to a decision that touches four functions. It arrives at the first, which reviews it and passes it to the second, which raises a question that the first has already answered, which goes back, and so on. Nobody in that chain is slow. Every individual response might take a day. The decision still takes five weeks, because it is being processed sequentially, and sequential processing has a floor set by the number of hops rather than by the speed of the hoppers.

Now run the same decision in parallel: all four functions working simultaneously against a single shared brief, with a named decider and a date. Same people. Same working speed. A week.

That is not a motivational intervention. It is an architectural one, and the gain is permanent, because an approval layer you remove stays removed while an urgency programme decays on a schedule.

There are twelve of these properties, and I did not invent them — I took them from the thing that has actually improved a million-fold in a decade. Parallel processing. Distributed intelligence. Scalable architecture. Data as infrastructure. Continuous learning. Pattern recognition. Feedback loops. Transfer learning. Network effects. Self-optimisation. Modular composability. Emergent intelligence.

Not one of them is a technology. All twelve are arrangements, and an organisation can be arranged along every one of them.

The measured gap is not twenty per cent. Scoring real organisations against all twelve on a five-level maturity ladder puts the distance between an accelerated organisation and a traditional one at somewhere between 10x and 100x, and it compounds, because each removed constraint makes the next removal cheaper.

What to do is narrower than it sounds. Do not attempt twelve. Score all twelve honestly, take the three that score lowest and are either Foundation or Quick-Win, and move only those. Measure decision velocity before and after on the same class of decision. Re-score at ninety days with the same instrument.

And if it did not move, publish that internally. An organisation that cannot report an honest negative result on its own transformation has just demonstrated its score on feedback loops.

FAQ

Questions about The twelve architectural traits

Answered plainly, with the real figures. If something here is wrong, tell us and it goes in the log.

Parallel processing, distributed intelligence, scalable architecture, data as infrastructure, continuous learning, pattern recognition, feedback loops, transfer learning, network effects, self-optimisation, modular composability and emergent intelligence. Not one of them is a technology. All twelve are arrangements, and an organisation can be arranged along every one of them.

Scoring real organisations against all twelve traits on a five-level maturity ladder puts the distance at somewhere between 10x and 100x, and it compounds, because each removed constraint makes the next removal cheaper. This is the essay's own claim rather than an audited figure, and it is stated with that qualification.

Three. Score all twelve honestly, take the three that score lowest and are either Foundation or Quick-Win, and move only those. Measure decision velocity before and after on the same class of decision, then re-score at ninety days with the same instrument. If it did not move, publish that internally.

The AQ Score is free: twelve questions, about four minutes, one score out of sixty across five maturity levels. It is benchmarked against an industry mean of 27, which is graded [B] because the constituent studies are public but the compositing has not been externally reviewed, as at 26 August 2026.