This
year,
“agentic”
became
the
sexiest
buzzword
to
hit
vendor
PowerPoints.
It’s
everywhere
from
specific
products
to
the
era
itself.
At
times,
it
seems
that
no
copy
can
leave
the
door
without
the
word
“agentic”
crammed
in
there,
despite
it
hitting
the
ear
with
roughly
the
same
credibility
as
“putting
the
law
on
the
blockchain”
or
“building
a
metaverse
practice.”
And
then,
Clio
did
something
remarkable
at
its
2025
conference:
it
didn’t
say
it.
Much
like
jazz,
sometimes
the
most
important
part
of
a
conference
is
what
you
don’t
say.
As
CEO
Jack
Newton
unveiled
an
ambitious
future
for
the
company’s
plan
to
take
on
the
world
as
a
more
or
less
everything
app
for
lawyers,
he
wasn’t
talking
about
agents.
By
my
count,
Newton
mentioned
the
term
exactly
twice
during
his
keynote,
and
both
times
in
passing
reference
to
broader
industry
trends
as
opposed
to
describing
Clio’s
own
products.
Instead,
he
opted
for
terms
like
“automation”
and
“teammates.”
These
may
seem
like
semantic
differences,
and
to
some
extent
they
are,
but
the
absence
of
agentic
—
the
conscious
omission
of
a
ubiquitous
term
—
says
a
lot
about
Clio’s
strategy
and
engagement
with
its
users.
As
someone
who
has
staked
out
a
position
as
an
aggressive
hater
of
2025’s
most
overrused
empty
signifier,
I
couldn’t
have
been
more
pleased
by
this.
In
the
legal
industry,
the
term
“agentic
AI”
means
one
of
two
things,
and
neither
particularly
useful.
Either
it’s
describing
a
truly
autonomous
system
that
takes
user
goals
and
some
vague
constitutional
guidance
to
chart
out
its
own
workflow
that
it
goes
out
and
pursues
before
delivering
a
final
product.
This
is
what
we
in
the
business
would
call
“malpractice.”
Agentic
can
also
describe
a
series
of
vetted,
cascading
prompts
we’d
otherwise
just
call
“automation”
but
for
Silicon
Valley
gloss.
Mercifully,
most
products
calling
themselves
“agentic”
in
the
legal
space
fall
into
the
latter
category
—
competent
workflow
automation
that
lawyers
would
embrace
if
it
weren’t
wrapped
in
terminology
that
suggests
their
AI
might
go
rogue
and
file
a
motion
without
them.
While
every
other
company
at
legal
tech
conferences
this
year
has
been
tripping
over
themselves
to
hype
their
spin
on
agentic,
Clio
seems
to
have
read
the
room
—
or,
more
precisely,
the
lawyers.
Chief
Product
Officer
John
Foreman,
confirmed
that
this
rhetorical
choice
was
very
much
intentional.
“If
you’re
saying
agentic,
who
are
you
talking
to?”
Foreman
asked.
“Investors?
Certain
media
publications?
What
if
you
want
to
talk
a
solo
lawyer
in
the
audience
that
needs
to
use
this
stuff?
‘Agentic,’
as
a
term,
does
nothing.”
Average
attorneys
don’t
want
to
send
their
work
to
agents.
An
agent
is
someone
you
hire
to
go
out
on
your
behalf
and
get
you
a
better
deal
while
hiding
how
the
sausage
is
made.
They
do
your
work
instead
of
you
and
then
ask
for
10
percent.
A
“teammate”
on
the
other
hand
is
someone
who
works
with
you.
An
associate
or
paralegal
is
someone
who
does
work
for
you
that
you
—
based
on
your
actual
experience
—
then
redline
into
oblivion.
“What
do
agents
do?”
Vice
President
of
Legal
Content
and
Migrations
Chris
Stock
asked.
“Agents
do
stuff
for
you,
but
they
don’t
always
get
it
right.
What
do
members
of
your
team
do?
An
assembled
team
works
together,
they
get
to
the
right
conclusions
together,
they
support
each
other.”
“Human
in
the
loop”
is
the
vogue
pushback
that
agentic
advocates
make,
but
this
is
a
superficial
nod.
You
say
“we’ll
keep
you
in
the
loop”
is
what
you
say
to
the
most
annoying
guy
in
your
group
while
planning
the
after
party.
Lawyers
shouldn’t
be
in
the
loop,
they
should
be
the
center
of
the
whole
conversation.
On
the
surface,
what
most
companies
are
calling
agentic
might
not
differ
from
what’s
being
called
a
teammate,
but
it
carries
a
ton
of
subconscious
baggage.
If
vendors
set
out
to
be
agentic,
the
pressure
is
always
on
them
to
move
more
tasks
behind
the
veil.
But,
as
we’ve
put
it
around
here,
this
“GPT-sus
take
the
wheel”
mentality
just
further
moves
the
lawyer
and
their
professional
judgment
out
of
the
center.
It’s
not
enough
to
“edit
at
the
end,”
because
lawyering
is
an
iterative
process
that
requires
those
breaks
in
the
process
where
the
team
can
—
to
quote
from
the
Bard
—
stop,
collaborate,
and
listen.
A
lot
of
the
magic
happens
when
a
junior
shows
a
senior
the
work-in-progress.
A
system
that
jumps
from
input
to
final
product,
will
still
get
an
edit,
but
it’s
a
different
mental
process
than
engaging
with
work
product
over
and
over
throughout
its
production.
“It’s
all
about
checkpoints,”
Foreman
told
me.
“You’re
not
necessarily
going
after
automating
this
whole
thing
into
a
Rube
Goldberg
machine
of
AI.”
Figuring
out
ways
to
keep
the
AI
actively
leveraged
while
not
losing
the
appropriate
interruption
points
is
a
key
difference
between
thinking
in
agent
vs.
thinking
in
teammate.
This
might
not
impress
investors
and
podcasters
as
much,
but
it
should
make
lawyers
much
more
comfortable.