►
From YouTube: Verify Think Big #1
A
A
goal
that
I
have
is
to
populate
some
validation,
backlog
issues
for
our
investments,
and
this
particular
topic
came
up
out
of
a
discussion
that
I
was
having
with
darren,
and
we
were
thinking
through
what
were
some
of
the
ways
that
we
can
consider
challenging
ourselves
to
invest
in
our
innovation
for
ci
and
really
to
answer.
Questions
like
how
can
git
lab
do
ci
better.
So
some
of
the
challenges
that
we're
looking
at
when
we
think
about
git
lab
and
ci
our
customers
gartner
analysts.
A
The
market
are
saying
that
customers
in
the
market
are
perplexed
by
development
in
ci
environments,
the
time
that
it
takes
to
a
troubleshoot
ci
when
we
think
about
passing
integration
tests
and
even
just
the
burden
of
writing
tasks.
A
So
there's
lots
of
different
meaty
problems
that
we
can
solve.
So
what
I'm
hoping
to
accomplish
in
this
think
big
is
to
take
a
couple
of
these
challenges
and
think
through
what
we
can
do
and
get
lab
ci
and
and
take
them
apart
and
potentially
take
some
action
items
forward
in
issues
and
research
them
validate
them
and
create
some
mbcs
sound
good.
A
I
pre-populated
some
questions
that
we
can
start
with
broadly
on.
How
can
we
do
gitlab
ci
better,
but
I
also
have
some
bullet
points
around
the
problems
and
challenges
and
then,
of
course,
from
the
issue,
the
high
points
that
we
previously
discussed.
So
we
can
start
with
any
of
these
areas
I'll
leave
it
up
to
how
the
group
would
like
to
to
start
this
conversation,
I'm
I'm
flexible.
We
only
have
30
minutes,
so
I
want
to
be
efficient
with
our
time,
but
I'm
also.
B
I
don't
know
yeah,
I
guess
I
would
frame
up
the
conversation
and
I'm
sure-
and
hopefully
everybody
hear
me-
okay,
I'm
sure,
fal
and
and
dove
thought
about
this.
But
just
if
I
was
thinking
about
this
and
just
kind
of
stepping
back
would
be.
If
all
of
us
got
here
today
were
working
in
a
startup
in
some
mythical
place,
mythical
remote
land,
and
we
were
trying
to
do
ci
differently.
We
were
trying
to
disrupt
the
existing
players
in
the
ci
space,
gitlab,
github
and
so
on,
we'll
be
doing
anything
differently.
B
B
And
if
the
answer
is
no,
then
we
just
continue
down
the
path
in
terms
of
occurring
like
strategic
direction,
but
I
always
ask
like
to
me
that's
kind
of
how
you
disrupt
the
market
right.
That's
kind
of
what
happens
when
you
see
someone
you
show
up,
it's
like
they're,
always
like.
Oh,
we
didn't
think
about
that
and
they
didn't
drop
them
up.
So
can
we
disrupt
ourselves
and
that's
the
answer
is
no
it's
fine,
but
that's
the
kind
of
the
place
I
was
starting
from.
B
And
then
to
make
it
more
tangible
for
the
engineers
in
the
room
kind
of
going
to
the
brass
tacks
at
the
end
of
the
day,
right
now,
a
ci
platform
is
still
for
the
it's.
Not
still
I
mean
it
is
it's
executing
a
script
right,
it's
basically
executing
workflows,
and
I
guess
I
don't
know
if
we
can
get
away
from
that.
I
don't
know
if
you
can
get
away
from
having
to
script
a
workflow
to
build
a
piece
of
software
right.
So
if
all
else
has
been
equal,
can
you
truly
disrupt
this
space?
D
Myself,
so
I
I,
I
saw
a
comment
that
darren
had
written,
I
think
somewhere
in
this
issue,
and
I
can't
find
it
right
now,
but
it
made
me
think
about
the
idea
of
like
the
the
like,
the
user
of
the
ci
system
and
and
it
being
kind
of
like.
D
I
can't
I
can't
think
of
the
right
word,
but
it's
just
like
it
just
works.
It's
like
they
don't
have
to
spend
a
bunch
of
time
on
it.
D
It's
just
sort
of
like
I
I'm
really
I'm
not
awake
yet
today,
like
like
it's,
it's
kind
of
it's
kind
of
transparent
like
like
you
just
you
put
your
code
there
and
it
does
the
thing
that
it's
supposed
to
do
and
I
think,
like
that's,
that's
like
it's
not
really
disruptive,
because
I
think
that's
what
other
competitors
do,
but
I
think
that's
like
the
the
thing
that
people
want
is
they
want
to.
They
want
to
get
the
job
done.
D
They
want
to
like
get
their
code
and
run
it
leverage
the
features
and
they
they
don't
want
to
have
to
like
understand
a
lot
of
the
internals
so
like
the
the
mapping
in
my
head
is
like
comparing
heroku
to
kubernetes
like
heroku.
You
push
your
code
and
it
just
figures
it
out
and
does
it
for
you.
D
Kubernetes
is
you're
writing
a
ton
of
ammo
and
you're,
like
you
have
to
understand
like
what
these
concepts
are
under
the
hood
and
like
what
these
primitives
are
that
you're
interfacing
with
in
order
to
get
it
to
work,
and
I
think
that's
like
that's
like
the
I
don't
know
holy
grail
for
for
ci
is
just
like
here's,
my
code
to
do
the
thing
I
wanted
to
do.
E
I
think
that's
the
holy
grail
for
like
your
startup
user
and
even
then
you're,
like
customer
of
the
ci
system.
I
think
we
should
also
think
about
the
persona
of
the
operator
who,
for
us
may
be
the
person,
ultimately
writing
the
check
and
making
that
buying
decision.
Not
only
do
I
want
it
to
get
green
fast
or
set
up
my
first
pipelines
fast.
I
want
them
to
be
efficient.
Always
so,
can
we
start
to
do
things
like
repair
the
pipeline?
E
Can
we
start
to
do
things
like
tell
you
where
inefficient
pipelines
are
that
are
a
slowing
teams
down
and
be
costing
you
money?
I
think
that's
how
you
would
disrupt
that
space
and
that's
what
rat
and
just
sas
in
general
is
not
just.
Let
me
do
things
but
tell
me
where
I
can
go,
make
things
more
efficient.
D
Yeah,
that's
that's!
That's
a
really
interesting
idea
like
if
you
think
about
the
person
setting
it
up.
You
set
it
up
one
time,
but
then
you
run
it.
You
know
maybe
hundreds
of
thousands
of
times
and
that's
where
that's
where
the
actual
like
impact
is
like
on
slowing
your
team
down
or
your
cost
or
flaky
tests,
like
all
of
those
things
become,
you
know,
amplified
when
it's
done
over
and
over
and
over
again.
C
F
I
think
there's
two
use
cases,
there's
the
user
who's
starting
ci
for
the
first
time,
and
they
would
benefit
from
an
auto
generating
smart
pipeline
kind
of
to
sum
up
what
darby
was
describing
it's
a
smart
pipeline.
It
knows
what
jobs
it
needs
based
on
the
code,
language
or
the
what's
being
tested,
what's
being
the
code
itself,
auto
generating
smart
pipeline,
but
then
there
is,
I
think,
just
equally
important
the
use
case
of
someone
who
has
pipelines
already.
F
It
would
be
nice
if
they
can
put
it
into
some
thing.
That
is
yet
to
exist.
That
says,
I'm
going
to
have
you
do
smart
analytics
on
it,
a
smart
analysis,
my
pipeline
and
spit
out
all
the
suggestions
on
how
I
would
optimize
this
that
for
a
user
that
already
has
a
complex
pipeline
would
be
kind
of
revolutionary
as
like
a
an
algorithm.
That
is,
I
don't
know
smart
optimization.
F
F
F
This
one
status
would
depend
on
the
status
of
that
job,
but
this
other
one,
not
necessarily
I
don't
know
it's
almost
like.
G
I
I've
seen
like
devops
engineers
that
are
struggling
with
ci,
mainly
because
sometimes
it's
not
predictable,
they
they
want
to
to
run
a
pipeline,
and
they
want
to
know
that
this
pipeline
will
pass
and
there
won't
be
any
surprises
on
the
way
we
missed
something
or
something
fail.
And
now
I
need
to
wheel
on
the
whole
thing
so
making
it
very
predictable
yeah.
G
G
They
won't
test
like
subset
of
their
pipeline,
because
there
are
some
areas
that
they'll
not
feel
certain
about,
and
then
they
need
to
run
the
whole
thing
just
to
get
struggled
by
by
like
this,
like,
I
would
say,
like
job
eight,
so
something
around
this
area
can
be
very
helpful
for
us
to
disrupt
the
market.
I.
G
E
E
Actually
didn't
ask
like
why
your
pipeline
is
failing,
I
would
say
yeah,
because
we
could
do
things
like
capture
some
of
the
traffic
and
stub.
Some
things
based
on,
if
you
have
100
pipeline,
runs,
here's
generally
what
the
inputs
into
that
job
look
like
just
reuse,
the
most
common
ones
and
then
run
it
locally
on
a
local
runner
or
something
or
in
a
sandbox,
or
something
if
you're.
H
H
Just
to
add
on
to
what
dov
is
saying
yeah,
the
reason
I
mentioned
ai
was
or
machine
learning
really
is.
If
we
have
the
data
already,
we
should
be
leveraging
this
and
presenting
this
to
our
users,
not
only
like
in
like
an
analytics
or
graphical
form,
but
like
look
as
already
mentioned,
looking
to
optimize
what
they
currently
have,
we
can
start
looking
at
how
we
can
predict
what
they
want.
I
think
I
think
that
would
really
disrupt
our
space.
Jackie
did
you
want
to
add.
A
I
think
it's
complemented
by
the
ml
ops
direction.
I
just
linked
the
direction
page.
There
there's
already
created
issues
about
things
that
you
can
do
with
ml
ops,
on
our
runners
being
able
to
predict
and
optimize
cost
detection
and
cost
modeling
related
to
ci
minutes.
A
I
think
that
if
we
think,
if
we're
looking
at
the
operator,
the
operator's
perspective
and
we
consider
predictive
models
in
creating
bayesian
models,
so
using
ml,
not
ai,
and
we
forecast
out
for
users
before
they
even
run
a
pipeline.
What
could
your
costs
be
based
on
this
pipeline
and
if
you
were
to
apply
or
rewrite
your
yaml
file
using
these
kinds
of
things,
you
could
save
this
much
in
a
year.
A
D
The
next
next
step
there
is
then
making
an
mr
to
their
project
with
an
updated
ammo
file.
C
C
You
know
when
we
found
a
vulnerability.
We
wanted
to
auto,
create
an
mr
that
fixes
it
for
you
and
we
struggle
with
bots
and
things
like
that,
because
it's
quite
intrusive.
So
there's
there's
steps
to
to
get
to
that.
I
think
people
freak
out
a
little
bit
if
we
were
to
create
loads
of
mr
saying
hey
this
fixes
your
things
so
there's
there's.
C
C
E
F
Oh
that
just
gave
me
a
thought.
You
know
there
is.
There
was
a
commercial
a
while
back
about
security,
africa.
I
don't
know
what
the
product
was.
Wasn't
that
good,
a
commercial
can't
can't
run
the
product,
but
the
user
was
saying
it
was
equivalating
to
security
monitoring
versus
security
like
actually
doing
security
patching,
I
guess,
and
the
commercial
was
this
guy
was
in
a
bank
that
was
being
robbed
and
he
says
and
and
one
of
the
people
on
the
ground
that
says,
do
something
and
he's.
F
Oh,
I'm
not
a
security
guard,
I'm
a
security
monitor,
and
he
says
so.
I'm
doing
my
job
to
tell
you
we're
being
robbed
so
same
thing
with
the
linter.
If
we
could
take
it
from
telling
the
user
what's
wrong.
To
tell
you
that's
wrong
and
here's
an
mri
just
created,
and
you
merge
it
to
fix
your
your
your
ci
configuration
that
would
be
even
better
than
than
monitoring
that
that
you
know
that
something's
wrong
with
your
config.
Here's
a
suggestion.
If
you
like
it,
merge
it's
a
one-click
solve
that
would
be
awesome.
A
A
A
I
think,
from
from
a
thinking
big
perspective,
we're
addressing
the
full
life
cycle
here
of
a
developer,
an
operator,
a
an
end-to-end
team,
and
by
integrating
this
ops
for
all
theme
that
we're
trying
to
to
really
solve
the
pain
point,
making
it
really
easy
for
troubleshooting
for
integrating
ci
and
really
disrupting,
but
also
kind
of
thinking
through
this
push
button
piece
if
we
weren't
to
leverage
ml
so
say
that
this
other
group
is
responsible
for
all
of
the
ml
architecture.
A
So
let's
say
that
we
have
to
rely
on
them
for
implementing
ml
architecture
and
we're
waiting
on
them.
For
that,
what
could
we
do
independently
kind
of
thinking
small
in
the
same
session,
given
that
we
have
like
five
minutes
left?
What
do
you
think
would
be
a
next
actionable
step
for
driving
innovation?
Do
you
think
it's
pursuing
things
like
like
more
dashboarding
or
more
cost
integration?
E
If
it's
just
a,
I
want
more,
I
want
faster
resolution
on
pipelines.
Just
tell
me
which
job
fails.
The
most
often,
which
is
give
me
a
rough
count
of
this
job,
is
failing.
The
most
out
of
all
of
your
pipeline
runs
because
that's
gonna.
If
I'm
looking
at
a
stacked,
ranked
list
of
those,
that's
the
one,
I'm
gonna
go
tackle
first
and
try
to
debunk.
D
So
something
I've
been
thinking
about,
I'm
not
sure
if
this
is
like
documented
well
or
not,
but
I
think
we
kind
of
touched
on
this
a
little
bit
with
getting
fast
feedback
on
pipelines.
Some
things
I've
noticed
just
playing
with
things
on
my
own.
Is
it's
not
always
easy
for
me
to
know
like
if
I
make
a
change
to
enable
a
feature?
One
thing
I
enabled
recently
was
the
test.
D
Failure
reporter,
like
it
took
me
a
while
to
get
it
right
because
I
had
to
like
my
script
was
wrong.
I
was
doing
something
stupid
but,
like
I
had
to
keep
pushing
and
running
and
waiting
and
like,
I
know,
there's
ways
to
do
this
locally
and
I
think
we
could.
We
could
probably
make
it
easier
for
for
people
to
understand
how
to
do
that.
Like
kind
of
quick,
you
know
quick
like
copy
these
commands.
D
B
Now
that
you
asked
me
to
tell
you
about
the
matrix
as
soon
as
my
screen
completely
messed
up,
I
was
just
kind
of
like
thinking
of
the
top
of
my
head
really
quickly
and
it's
like
a
really
convoluted
thought
so
bear
with
me.
B
I
want
to
double
check
kind
of
where
our
current
growth
is
happening.
I
know
we
have
a
strategic
thrust,
obviously
on
deflect
sas,
so
assuming
for
a
moment
about
growth
is
happening
against
that
sense
and
then
double
checking
what
kind
of
uses
or
users
or
companies
is
driving
that
growth.
B
I
was
thinking
that
if
we
were
to
kind
of
think
about
it
from
a
perspective
of
just
to
focus
on
gitlab
sas
and
the
first
time
users,
if
we
were
making
to
continue
the
point
of
that
style,
was
making
about
of
generating
slam
pipelines
and
picking
up
the
thought
from
darmy
about
making
it
super
easy
that
first
time
used
on
gitlab
sas.
B
B
I
went
to
the
first
time
user
because
I'm
just
such
a
marketing
geek-
and
I
know
it
doesn't
drive
growth
and
adoption.
But
I
just
love
hearing
like
a
first-time
user
saying
wow
gitlab
ci
is
so
super
cool
and
easy,
it's
the
best
guy
in
the
world.
I
know
that's,
not
gonna
drive
all
of
them
revenue,
but
just
clearly
that,
like
sentiment
in
the
market,
this
is
so
interesting
to
me
anyway.
That's
kind
of
what
I
was
thinking
like
the.
A
That's
something
that's
something
to
that.
We
can
also
pair
with
growth
on
too
so
if
we
create
like
a
a
great
landing
spot
with
the
first
time
user
experience
on
sas,
we
can
really
tie
this
into
growth
and
in
our
tams
and
professional
services,
and
if
we
build
a
usable
product
experience
around
these
auto
generated
pipelines
for
people
on
sas.
A
This
is
something
that
we
can
really
great,
create
a
great
product
and
user
experience
around,
and-
and
I
could
I
could
see
this
being-
this
could
be
a
really
great
user,
generating
opportunity,
at
least
for
an
experiment.
G
D
Hey
so
I
have
a
question.
Maybe
this
is
already.
Maybe
everyone
already
talks
about
this,
but
how
often
do
we
talk
about
like
the
first
time
user
being
like
someone
at
a
company
experimenting,
trying
out
ci
systems
and
like
that
experience
being
like
leading
them
into
further
implementations
with
git
lab?
Like
I
know
at
a
company
I
worked
at,
I
evaluated
a
bunch
of
ci
systems.
A
A
I
know
that
we
are
at
time,
but
is
this
something
as
a
quick
check
that
you
would
want
to
be
recurring
on
a
quarterly
basis,
just
kind
of
a
discussion.
A
You
can
say
yes,
no,
okay,
yes
great,
and
then
I
am
open
to
also
including
engineers,
I'm
open
to
having
this
be
on
a
distribution
list,
this
being
a
larger
group,
I'll
tag
that
in
this
ish
in
the
issue,
the
next
generation
ci.
So
if
we
want
to
have
a
follow-up
discussion
on
this
topic,
where
we
include
technical
resources
and
dive
into
more
of
these
issues,
we
can
do
like,
I
think,
small
and
go
deeper
on
that.