A note for AI researchers and labs, on a kind of data your training corpora do not contain. Read it →
Food for thought

Perspectives from the lab.

Short pieces on cognition, measurement, and the things the data keeps telling us. Written by the people building Agnira.

On measurement

Why a child can read every word and understand none of them

Decoding and comprehension are different cognitive acts. A reader can pronounce every word on a page while the meaning never assembles. We keep testing the first and assuming the second. The gap between them is exactly where readers are quietly lost, year after year, with no test raising an alarm.

On data

The dataset education forgot to collect

Schools hold decades of marks and almost no record of cognition. Marks tell you the outcome of thinking; they never tell you the shape of it. And the shape is the only part you can actually intervene on. We built Agnira to collect the dataset that was always missing.

On the cliff

Comprehension does not fade. It falls off a cliff.

Across every school we have measured, the pattern is not a gentle decline. Benchmark attainment holds, holds, holds, then collapses at the transition years. A gradual slope you can manage. A cliff you have to see coming. Surface testing never sees it.

On reading

More books is not the same as better readers

Twenty books that climb in cognitive complexity build a reader more than fifty that sit at the same level. Volume feels like progress and reassures everyone involved. But volume without direction is motion without travel. Sequence is the active ingredient.

On fairness

Rigour should not be a privilege of well-funded schools

A pen-and-paper instrument that needs no devices can measure a Tier-3 town school as precisely as a metro campus. When the instrument is constant, the differences it records are real, and every school, regardless of budget, deserves to see the truth about its own students.

On the long view

The bottleneck you can spot at thirteen

A cognitive bottleneck identified at thirteen is a bottleneck you can still do something about. Left unmeasured, it follows a person undetected into the job market, the boardroom, and every complex decision they will ever make. Early measurement is leverage.

Have a perspective worth sharing?

We collaborate with researchers, educators, and writers who take cognition seriously. If that is you, we would like to hear from you.

Get in touch