When the typewriter was invented, it did not change what people wrote. It made them write faster. The ideas were the same. The arguments were the same. The memos, the letters, the reports — all the same. Just produced more quickly and more legibly than before.
When email arrived, it did not change what people communicated. It made them communicate more. The same updates, the same requests, the same status reports — now sent to ten people instead of one, with a CC list that grew every quarter.
When spreadsheets replaced ledgers, the calculations got faster. When word processors replaced typewriters, the drafts got easier to revise. When video conferencing replaced phone calls, the meetings multiplied. The pattern repeats with every tool: the medium changes, the volume increases, and the underlying quality of thought stays exactly where it was.
AI is the most powerful tool most organisations have ever adopted. And most of them are using it to type faster.
The trap
The typewriter trap is straightforward to describe and remarkably difficult to escape. It works like this: a new technology arrives that is genuinely transformative. The organisation adopts it. And then it applies the technology to its existing workflows, producing more of what it already produced, faster than it produced it before.
More emails. More reports. More documents. More proposals. More meeting summaries. More Slack messages. More status updates. More slide decks. The volume goes up. The dashboards look healthy. The organisation feels productive.
But the quality of thought behind those outputs stays flat. The proposals are drafted faster but informed by the same incomplete understanding of the client's actual problem. The reports are produced in minutes instead of hours but contain the same analysis the team would have reached anyway. The meeting summaries are generated automatically but nobody reads them, just as nobody read the manually written ones.
This is the trap. Not that the organisation is doing something wrong. It is doing something natural — applying a new tool to familiar problems. The trap is that this natural response captures approximately five percent of the tool's value while creating the illusion that the full value has been realised.
The CEO reads a report about AI adoption. Ninety percent of staff are using AI tools weekly. Productivity metrics are up. Time-to-draft has decreased by forty percent. Everything looks correct. Nothing has fundamentally changed.
Volume vs intelligence
The question is not whether your organisation uses AI. Every organisation will. Most already do. The question is whether it responds to AI with volume or with intelligence.
Volume means doing more. More outputs, more deliverables, more communications, more content. Volume is the default response because it is the easiest to measure, the easiest to implement, and the easiest to celebrate. Volume shows up in dashboards. Volume feels like progress.
Intelligence means doing better. Not more proposals — better proposals, informed by everything the firm has ever learned about this type of client, this type of engagement, this type of objection. Not more reports — better reports, grounded in the accumulated institutional knowledge that exists somewhere in the organisation but that nobody can find when they need it.
Volume is a race to the bottom. When every firm can produce the same volume, volume stops being a differentiator. It becomes table stakes. The firm that drafted thirty proposals last month has no structural advantage over the firm that drafted thirty-one. Intelligence compounds. The firm that drew on twelve months of governed, validated institutional knowledge to craft fifteen deeply informed proposals has an advantage that cannot be matched by producing more.
Volume produces more. Intelligence produces better. The gap between these two responses is invisible in month one. By month six, it is measurable. By year one, it is a moat.
What compounding looks like
Consider two consulting firms. Both adopt AI tools in the same quarter. Both invest in training. Both integrate AI into their daily workflows.
Firm A uses AI to draft proposals faster. Their consultants paste client briefs into AI tools and get first drafts in minutes. They use AI to summarise meetings, generate status reports, and produce documentation. Their output per consultant increases by thirty percent. The managing partner is pleased.
Firm B does something different. They build a governed knowledge layer — a practice brain — that captures what the firm learns from every engagement. When a consultant encounters a client objection, the resolution is captured, validated, and classified. When a technical approach fails, the failure is documented with context. When a regulatory change affects an engagement, the impact assessment is stored and attributed. Every solved problem feeds the knowledge base. Every lesson makes the next engagement richer.
In month one, Firm A looks faster. Their output metrics are higher. Their consultants are producing more.
By month three, Firm B's consultants are not just producing — they are producing with the accumulated intelligence of every engagement the firm has ever captured. A consultant preparing a proposal for a financial services client does not start from a blank page or a generic template. The practice brain surfaces what worked in the last twelve financial services proposals — which objections arose, which pricing structures were accepted, which technical approaches the client's IT team pushed back on, and what the winning strategy looked like. The proposal is not just drafted faster. It is structurally better informed.
By month six, the gap is visible to clients. Firm B's proposals address objections before they are raised. Their delivery teams avoid mistakes that Firm A's teams are still making for the first time. Their junior consultants operate with the institutional knowledge of the firm's most experienced partners — not because they are more talented, but because the knowledge is available to them at the moment they need it.
By year one, it is a moat. Firm A has twelve months of faster typing. Firm B has twelve months of compounding institutional intelligence. A competitor who starts building their knowledge layer today will spend twelve months catching up — and during those twelve months, Firm B's knowledge base continues to grow. The distance increases monotonically. It cannot be closed by working harder or typing faster.
How to escape the trap
The escape begins with a different question. Instead of asking "how can we do more?", ask "what do we already know that we are not using?"
The answer, in almost every organisation, is: a lot.
The expertise exists. The patterns have been observed. The lessons have been learned. The problems have been solved — many of them multiple times by different people who never knew about each other's work. That knowledge is not missing. It is trapped. Trapped in inboxes nobody will search. Trapped in individual memory that walks out of the building every evening. Trapped in chat threads nobody will scroll back through. Trapped in project documentation that was filed once and never opened again.
"Stop asking how to do more. Start asking what you already know that you are not using. The answer is almost always: more than you think."
Freeing that knowledge — capturing it, validating it, governing it, and making it available at the moment of need — is not a technology problem. Any capable AI model can generate text. Any vector database can store embeddings. The problem is architectural. It requires a system that governs what enters the knowledge base, maintains what is already stored, enforces access controls, handles PII, and compounds intelligence over time instead of just accumulating data.
That is the work. Not typing faster. Not producing more. Building the infrastructure that makes your organisation genuinely smarter with every engagement, every solved problem, every captured lesson.
The typewriter made people write faster. Email made people communicate more. AI can make organisations think better — but only if they use it for thinking, not typing.
The question is whether your organisation responds with volume or with intelligence.