Immigration Advisor
Addendum: why the Immigration Advisor role is built this way
This is the part of a recruitment page that usually gets left out.
We think leaving it out is why roles like this get mistrusted.
The sources are cited at the end so you can check them rather than
take our word for it, which is, more or less, the point.
Why the future of law means the training of lawyers needs to change
Richard Susskind has been making a single argument for the better
part of forty years: that legal work will be broken into its component
tasks, and that the routine components will first be standardised,
then systematised, and eventually commoditised.
Writing with Daniel Susskind in The Future of the Professions,
he extended it beyond law. The bargain society struck with the professions,
in which experts hold a monopoly on practical expertise in exchange for
making it available, comes under strain once machines can make that
expertise available directly.
Daniel Susskind's later work pressed further on what that means for work
itself. Richard Susskind's most recent book takes the same discipline
to AI specifically.
You do not have to accept every conclusion to accept the direction
of travel. We do not accept all of them.
The routine legal work that is disappearing is the same work
junior lawyers have always learned on. It may not even be needed
in a post-AI world once you have qualified.
Legal research.
First-draft documents.
Bundling.
Document review.
Chronologies.
These parts of the training contract were how you learned law:
by doing the low-stakes parts of it badly, repeatedly, until you
did them well.
The task-level evidence has said this for years, and it is far more
focused than the headline "will AI replace lawyers" framing.
Frey and Osborne's occupational study put the probability of
computerisation for lawyers at around 3.5%. For paralegals and legal
assistants they put it at 94.5%.
Whatever one thinks of their method, and it has been criticised for
scoring whole occupations rather than the tasks inside them, the gap
between those two numbers is the entire story of what is happening
to the bottom and junior end of the profession.
Remus and Levy reached the same shape from the other direction.
Using data on how lawyers in large firms actually allocate their time,
they found the automatable share concentrated in document review and
routine drafting, while client counselling and advising were among
the least susceptible.
Their overall estimate of the employment effect was much smaller
than the headlines, and far more unevenly distributed.
More recent evidence sharpens the point rather than softening it.
Brynjolfsson, Li and Raymond studied more than five thousand workers
given access to a generative AI assistant and found the productivity
gains concentrated among the least experienced. The most experienced
saw only small gains in speed, and small declines in quality.
AI compresses the value of the kind of experience that can
be written down.
If your competitive advantage as a junior lawyer was going to be
that you had done something two hundred times and knew the format,
that advantage is now available to somebody on their first day.
So a firm that automates the routine work and changes nothing else
does not produce cheaper lawyers. It produces lawyers who have never
had to sit with a frightened person and work out what is actually wrong.
What Westkin think survives
The economist David Autor made the useful observation, borrowing
from Michael Polanyi, that we know more than we can tell.
The tasks that resist automation are the ones nobody can fully specify
the rules for, and the reason they resist is not that the task is
difficult but that the inputs are unstructured.
We think that describes the beginning of an immigration matter exactly.
Somebody contacts us in trouble. They tell you a story that is incomplete,
sometimes wrong, and almost always missing the fact that matters most.
You have to work out which of dozens of routes they might be on.
Whether a deadline is already running that they have not noticed.
Whether the refusal letter they are describing says what they think it
says. Whether the case is strong, weak, or premature.
And then you have to say the honest thing, including when the honest
thing is wait, or you do not need a lawyer for this,
or the firm that quoted you £900 has not read your refusal letter,
clearly enough that a frightened person can act on it.
That is diagnosis, performed live, under uncertainty, with a regulator
watching and no clean brief.
Producing the clean brief is the work.
It cannot simply be handed to a system, because the system cannot
be given the inputs.
So we start people there, rather than starting them on the tasks
that are being automated and hoping the judgement arrives later.
As an interesting aside, it is the kind of approach used to train
junior barristers: learn to think on your feet as you are training
under live fire.
Where the selling comes in, and why it is in the same job
Because the conversation in which you diagnose someone's problem
and the conversation in which they decide to instruct a firm are
the same conversation.
Separate them and both get worse. The person who diagnoses has no
stake in whether the advice was any good; the person who sells has
not read the file.
Firms that split them end up with salespeople quoting fees on cases
nobody has understood, and we know precisely what that looks like,
because we compete with them and we clean up after them.
A firm that will tell you your case is weak is a firm you
can believe when it says your case is strong.
That is an advantage a price-led firm structurally cannot copy,
because it can only afford honesty when honesty happens to sell.
Our advisors are not penalised for a free review that comes back
clean, and that is written into how the role is assessed, not just
into how it is described here.
What we are not claiming
None of the authors cited below says that lawyers should sell.
That conclusion is ours, and you should test it rather than accept
it because there are footnotes underneath it.
Nor are we claiming to know the timescale. Frey and Osborne's method
is contested. Remus and Levy wrote before large language models.
Richard Susskind has been right about direction for four decades
and wrong about speed more than once, which he says himself.
What we are confident about is narrower: the traditional junior
apprenticeship is thinning, the firms that notice late will train
a generation badly, and client-facing judgement is the last thing to go.
We would rather build the role around that and tell you we have
done it than build it quietly and let you find out.
Sources
On qualifying work experience, see Solicitors Regulation Authority,
Qualifying work experience, and The Law Society,
Qualifying work experience (QWE).