baz
Book a demo

Productive in a month. Still three years from judgement.

What a junior needs to know splits cleanly into what a system can hand them and what only time supplies. The first half compresses by 92%. The second half moves by 5%, which is inside the noise and should be read as zero.

Months to competence, six things a first-year learns

The faint bar is how long it used to take. The solid bar is what is left.

The bottom two rows are the ones worth staring at. Judging whether a deviation matters, and knowing which deadline the other side will move, are learned by watching people — and the table says so rather than finding a way to shade them slightly.

The search was the lesson

Three hours hunting through the precedent bank taught a junior what the bank contained. Nine seconds teaches them that the tool works.

Both produce the clause. Only one produces somebody who knows where it came from, and a firm that deploys this without changing how it supervises will get productive second-years who plateau quietly.

The search was the lesson
Three hours hunting through the bank taught a junior what the bank contained. Nine seconds teaches them that the tool works. Both produce the clause; only one produces a lawyer who knows where it came from.
Answers arrive without their context
A retrieved position is correct and unaccompanied. The junior never sees the eleven near-misses they would have read on the way to it, and those were most of the education.
Plateaus are quiet
A productive second-year who cannot yet judge a deviation looks fine on every metric a firm tracks. The gap shows up in year four, at the point it is expensive to fix.
So the provenance is not optional
Every answer carrying its source is a partial mitigation — the junior at least sees which agreement it came from and can open it. It is genuinely partial, and a firm relying on it alone is relying on curiosity.

Four things to do about it

None of them are features. All of them are supervision decisions, and the first costs nothing.

01 Make them predict first
Ask for a position before showing them the distribution. Five seconds of commitment turns a lookup into a calibration, and it is the cheapest intervention available.
02 Review the spread, not the answer
Supervision that asks why the median sits where it does teaches more than supervision that checks the clause. The tool makes that conversation possible on day one instead of month twelve.
03 Keep some things manual on purpose
A firm that runs every first-year through one unassisted diligence exercise pays for it once and gets a cohort that understands what the ranking is doing.
04 Measure judgement separately
Time-to-productive will improve and say nothing about the two rows at the bottom of the table. Those need their own assessment or nobody will notice they have not moved.

A first-year who is useful in a month is worth a great deal, and it is half a claim. The other half is that they still need the four years, and the firm has to decide deliberately where those come from now that the easy hours have gone.

The other outcomes

Keep the file in the Kingdom

Where documents go, and where they do not.

Cut review time

The first pass on a document set.

Draft in both languages

Arabic and English that agree with each other.

Never miss a limitation

Counted from the trigger event.

Bring research in-house

Stop paying counsel to look things up.
Talk it through The spread a junior sees on day one