►
From YouTube: Generative AI Demo
Description
Learn more about GitLab's AI features:
- Code Suggestions
- Suggest merge request summary changes
- Generate suggested tests in merge requests
#AI #DevSecOps
A
Today,
we'll
talk
about
the
generative
AI
features
in
git
lab
that
are
powered
by
vertex
AI
I'm,
Renardo,
Caden
partner,
Solutions
architect
at
gitlab
and
I
help
team
members
and
customers
alike
learn
more
about
the
tight
integration
between
gitlab
and
Google
Cloud.
Today's
features
that
we
will
be
focusing
on
are
code
suggestions,
suggest
merge,
requests,
summary
changes
and
generate
suggested
tests
in
merge
requests.
A
All
of
these
are
part
of
the
gitlab
duo,
Suite
of
AI
powered
features
before
we
proceed
to
a
demo,
I'd
like
to
give
a
high
level
overview
of
the
AI
architecture
at
gitlab,
now,
clients
who
communicate
to
the
AI
abstraction
layer
depending
on
their
implementation,
whether
it's
the
SAS
version
or
self-managed
instance
of
GitHub,
and
these
will
communicate
with
our
access
layer.
Now
this
access
layer
will
then
route.
A
This
request
to
the
AI
Gateway,
which
will
then
communicate
with
vertex
AI,
so
code
suggestions
and
many
of
the
other
gitlab
Duo
features
will
work
this
way
now
under
the
hood.
Each
of
these
features
are
using
Google
Cloud
vertex
AI
related
services,
like
with
code
suggestions,
it
uses
Kodi
apis,
suggest,
merge,
request,
summary
changes.
It
uses
text
bison
and
then
generate
suggested
tests
in
merge,
requests,
also
uses
text
bison.
Now,
let's
get
to
the
demo.
A
I
have
to
remind
you
that
you
have
to
enable
code
suggestions
on
the
project
or
group
level
to
be
able
to
use
features
that
we
will
be
highlighting
in
today's
demo,
from
gitlab
I've
created
a
branch
from
the
main
code
from
where
all
of
the
changes
that
we'll
be
introducing
to
the
main
code
base
now
from
there
I
can
access
the
web
IDE
and
our
goal
for
today
is
to
create
a
python
code
that
will
be
a
calculator
function
now.
I'll
just
create
the
file
from
web
IDE
I'll
call
it
calc
2.
A
py,
and
then
now
we
have
a
blank
file.
Now,
what
will
serve
as
prompt
will
be
a
comment
that
I'll
be
making
here
and
let's
just
give
it
a
few
seconds
for
the
code
suggestions
to
kick
in
now
from
there.
We
can
wait
for
these
new
features
or
new
code
to
be
introduced
and
I
just
need
to
hit
tab
to
accept
those
suggestions.
A
So
from
here
we
have
the
calculator
class.
We
have
a
new
add
function.
A
sub
or
subtract
multiplication,
and
probably,
let's
finish
off
with
division,
now
we're
happy
with
this
with
this
set
of
new
codes,
so
without
me
creating
new
code.
It's
just
prompted
by
that
comment
that
I
just
want
to
create
this
class.
Now
from
there.
I
can
go
to
my
commit.
A
A
Now
just
call
this
merge
request
that
will
where
we
will
see
all
the
changes
and
progress
of
that
new
introduction.
You
introduce
newly
introduced
code,
create
wrench
request
right.
So,
let's
just
wait
for
this
pipeline
to
complete,
so
I
have
fast
forwarded
a
bit
to
show
you
that
the
merge
is
complete
and
there
were
several
jobs
that
were
done
to
merge
in
the
price
of
that
merge
now.
I
want
to
also
highlight
that
those
jobs
that's
depicted
here,
then
they
would
be
ensuring
security
and
quality
for
the
code
that
you'll
be
introducing
now.
A
This
is
something
that's
not
AI,
but
would
happen
across
gitlab
now
now
that
the
merge
is
done,
I'll
just
go
to
the
merger
Quest
and
show
you
the
two
other
AI
powered
features,
so
I'll
go
to
the
merge
status.
The
first
one
is
the
view
summary
notes
in
here
now:
ai
will
look
at
the
all
of
the
information
that
is
in
the
merger
Quest
and
would
summarize
it
because
typically,
as
a
merger,
Quest
we'll
have
more
information
and
sometimes
could
take
months
to
complete.
A
So
AI
will
condense
that
and
make
that
more
concise
information
for
you
to
process
and
read
now.
The
other
feature
would
be
the
test
cases
now
in
the
previous
submerge
I
would
be
using
that
to
highlight.
Now
you
will
go
to
the
changes
Tab
and
from
there
you
can
click
here
on
the
changes
of
that
code.
A
To
recap,
we
saw
a
few
generative
AI
features
powered
by
vertex
AI
in
gitlab,
the
first
of
which
is
code
suggestions
where
we
created
python
code
from
scratch
just
using
a
prompt.
This
uses
the
code
gecko
model
for
code
completion.
The
second
is
suggest,
merge,
request,
summary
changes
where
we
saw
a
summary
of
all
of
information
and
collaboration
that
had
happened
on
a
merger
Quest.