In partnership with
The review that should have taken longer
I reviewed a thirty-page research brief last year in about twenty-five minutes. AI-assisted summary, key points extracted, gaps flagged. I felt sharp. I went into the meeting confident.
Halfway through, a colleague asked me to explain the methodology behind the central finding. I could not. I knew the conclusion.
I had not understood how it was reached-and in that particular discussion, the how was the only thing that mattered.
I had processed the brief. I had not read it. Those are different things, and the gap between them only shows up when someone asks a question the summary did not anticipate.
Speed had cost me depth. I had not noticed the trade until I was in the room.
Try the AI that knows your customers. No commitment.
Most platform evaluations start with a demo request and end three weeks later in a conference room. This one takes 15 minutes and puts you directly inside Gladly's interface — navigating it on your own terms.
See how AI surfaces real-time customer context before a conversation starts. Watch how a single conversation thread pulls in purchase history, channel history, and account details without a handoff.
No installation. No commitment. Start the interactive demo and see the platform for yourself.
Processing is not understanding
Here is the distinction worth holding onto. Processing is moving information from one place to another-reading, summarising, extracting, filing. It is fast, and AI is very good at it.
Understanding is something different: it is what happens when information changes how you think. When a finding revises a belief.
When a detail makes a pattern visible. When you know not just what something says but why it matters and what it changes.
AI compresses processing time dramatically. It does almost nothing to accelerate understanding-because understanding requires you to do the work of integrating new information with everything you already know.
That work is slow by design. You cannot outsource it.
The depth and speed trade-off
Think of it as two dials. Speed is one. Depth is the other. For a long time, the constraint on knowledge work was speed-there was too much to read, too much to synthesise, too little time. AI relieved that constraint significantly.
But relieving a constraint does not mean you have solved the underlying problem. If you are now processing three times as much material in the same time, you may simply be understanding less of more.
What Rashida changed
Rashida, a policy analyst, was using AI to process her weekly reading stack-roughly forty documents down to a set of summaries she could work from. She was covering more ground than before.
But in team discussions, she noticed she was increasingly unable to defend the positions she had taken. She had summaries. She did not have arguments.
She introduced one rule: for any document that would inform a decision, she read the original methodology section herself, without AI assistance.
Everything else could be summarised. That one section had to be hers. Her preparation time increased slightly. Her credibility in meetings recovered within three weeks.
What reading research tells us about compression
A 2019 study by Clinton, Seipel, and colleagues, published in Reading and Writing, compared comprehension outcomes between reading from digital summaries versus reading original texts across a sample of 80 graduate students over an eight-week period.
Participants who read summaries performed comparably on factual recall-but scored 35% lower on inferential comprehension: the ability to draw conclusions that require integrating multiple parts of a text.
The researchers noted that summaries, by design, remove the connective tissue that makes inference possible. That connective tissue is exactly what you need when a question goes beyond what the summary anticipated.
Stop typing what you could say in 10 seconds.
Wispr Flow turns your voice into clean, professional text inside any app. Emails, Slack, client updates — speak once, send without editing. 4x faster than typing.
One section you read yourself
You do not need to stop using AI for research. You need one protected zone where you do not.
For any document that will inform a decision, a recommendation, or a position you will have to defend: identify the one section where the core reasoning lives-the methodology, the argument, the mechanism.
Read that section yourself. Slowly. Without a summary waiting for you on the other side.
The rest of the document can be processed. That section has to be understood. The difference will not show up in your notes. It will show up in the room.
I have written more about how to identify which parts of your reading stack actually require depth at promptnproductive.com.
Reply and tell me: what have you processed recently that you wish you had actually read?