►
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Restart
the
recording,
oh,
this
is
iteration
office
hours
and
I'm
Syd
I'm,
the
CEO
and
co-founder
at
gate
lab
that
gear
lab.
We
discovered
iteration
only
because
we
absolutely
had
to
doing
what
Combinator
there's
a
good
clip
out
there.
Where
we
talk
about
that,
and
we
have
to
keep
kind
of
rediscovering
this
because
iteration
is
so
counterintuitive.
A
It's
so
counterintuitive
to
ship
something
more
minimal
and
these
office
hours
I'd
love
for
people
to
kind
of
bring
forth
things
that
they
think
cannot
be
made
any
smaller
and
then
I'll
I'll,
give
suggestions
that
might
make
sense
or
might
not
make
sense.
But
it's
a
challenge.
The
thinking
I
see
what
we
can
do
better.
A
He's
not
I'm,
not
hearing,
Marc
and
also
I'd
love
to
like
I,
don't
I
can't
think
of
any
by
heart
of
worse
than
best
examples.
Let's
just
go
into
current
things.
I
think
that's
much
more
fun,
because
it's
about
the
process,
not
the
outcome
that
I
want
to
show
Josh
F
the
next
one
top-voted
item
from
last
cab,
yeah.
B
So
the
question
is:
is
how
do
we
like
it's
a
big
problem
and
even
packed
in
there,
is
how
to
even
get
the
data
over
to
these
folks?
One
person
was
like
webhooks
for
everything.
Another
person
said
I
would
love
to
have
a
cough
Scout
like
queue
for
all
the
events
and
just
funnel
the
events
into
our
data
Lake
and
we'll
just
run
our
licks
on
it
and
that's
it.
So
what
we
do
would
be
very
easy,
so
that's
kind
of
where
we
ended
up
in
so
this
is
very.
B
B
C
Did
ask
that,
and
they
basically
said
we
want
it
all
basically
or
anything,
get
labs
seized
we
we
could
potentially
use
it
there.
Well,
at
least
some
of
the
folks
were
like
we
don't
know
what
questions
we
want
to
answer
ask
of
the
data.
Yet
we
would
like
to
store
it
so
that
when
we
come
up
with
questions,
we
have
the
opportunity
to
query
the
data
set,
so
they
didn't
help
us
break
it
down.
Much
live
in
the
cab.
Okay,.
A
Well,
then,
that's
an
infinite
amount
of
work,
so
we
cannot
commit
to
that.
So
don't
have
to
say
what
their
first
thing
is
and
then
we'll
go
from
that
and
I
think
a
we're
poke
sounds.
We
know
how
to
make
web
hooks
sounds
like
a
relatively
simple
thing
and
if
we
can
put
on
a
cue
and
a
question
you
can
ask
us
like
hey,
are
the
audit
logs?
B
D
A
B
They
are
certainly
aware
of
our
logging
I
think
their
concern
is
that
there
could
be
some
events
in
there
that
which
they
don't
know
how
to
capture
necessarily
or
in
a
way
that's
easy
to
us,
and
so
they
parse,
we
probably
get
a
whole
bunch
of
login
statements
around
pipeline
starts
and
ends
and
a
whole
bunch
of
other
events,
though
it
might
not
be
terribly
easy
to
consume.
We
didn't
go
down
the
logging
aspect
in
particular
of
heavily
tried
it
and
have
they.
B
What
were
their
findings?
I
think
these
are
great
questions
to
ask
and
a
follow-up
session
we're
plenty
of
a
follow-up
session
in
December
and
our
next
virtual
call
to
dive
in
here
and
do
more
validation
and
investigation.
That's
a
good
point.
Stan
I
think
also
said.
I
assumed
I
thought
it,
as
you
did,
is
that
this
feels
like
an
audit
stream
and
we're
probably
gonna,
have
similar
requests
for
commerce
to
be
able
to
get
it's
like
a
sim
event
stream
out
of
github
calm
into
their
own
analytic
services.
B
A
A
So
that
might
be
like
we
know,
audits.
The
other
thing
is
very
important
to
our
customers,
so
that
might
be
the
first
step
to
offer
an
API
for
those
audit
events
and
then
the
next
step
would
be
streaming
streaming
instead
of
like
the
API,
where
people
have
to
go
out
and
fetch
them
we're
pretty
good
at
making
api's
I
think
an
API
for
the
audit
events.
How?
How
tough
does
that
sound,
Stan,
I.
D
Think
the
challenge
is
especially
with
our
current
audit
system.
Is
that
there's
so
many
different
things
that
can
happen,
and
how
do
you
represent
that
right,
for
example,
like
creating
a
new
user
you
have
to
you
want
to
blog
all
the
things
that
went
along
with
that
same
thing
with
starting
a
pipeline,
so
there
are
very
specific
events,
types
that
you
have
to
capture.
D
So
you
know
right
now
we
just
kind
of
throw
a
generic,
you
know
class
and
it
makes
it
harder
to
query
in
order
to
kind
of
understand,
because
it's
not
that
well
structure.
So
this
is
kind
of
why
I'm
more
of
a
proponent
of
JSON
logging
to
start
because
you
can
create.
You
know
arbitrary,
structured
logs
for
specific
events,
and
then
you
don't
have
to
worry
about.
How
are
you
I
got
a
query.
Yet?
How
am
I
got
a
story,
it's
just
kind
of
streaming
and
we
can
adapt
as
we
need
that.
A
So
maybe
it
starts
with
having
some
of
our
car
like
a
customer
used
at
Asian
log
and
then
have
them
start
using
it
and
then
complain
about
everything.
That's
wrong
with
it.
I
need
to
filter
I
need
more
days.
I
need
stuff,
combined,
whatever
I
need
to
I
need
to
have
it
directly
into
Kafka,
or
we
can
probably
build
something
that
gets
sense
changes
it
from
a
JSON
log
into
a
Kafka
I
made
her
or
whatever
that's
called.
E
A
E
F
Thanks
so
for
the
packaged
stage,
we've
implemented
a
few
different
package
manager
formats
like
NPM
and
maven,
and
we
have
Conan
coming
up.
These
integrations
typically
take
two,
sometimes
three
milestones
and
when
I
think
about
our
goal
to
cover
ninety
percent
of
our
customers
need
package
management
needs.
We
can't
take
three
months
for
each
integration,
so
I'm
wondering
how
we
can
make
this
smaller,
how
we
can
provide
them
more
iterative
value
when
we
think
about
moving
forward
yeah.
A
These
are
the
iteration
office
hours,
but
also
like
it.
It's
if
you
enable
the
wider
community
to
do
it.
It's
not
a
problem
if
it
takes
two
or
three
releases,
if
you
have
four
of
them
going
on
at
the
same
time,
so,
for
example,
from
our
current
documentation,
where
you
can
view
like
what
formats
we
support,
do
we
link
to
hey
we're
very
what
we
we
welcome
new
formats
and
here's
a
way
to
this
is
an
example,
someone
adding
something.
Here's
hints
and
things
like
that.
We
do.
We
invite
those
contributions.
A
F
Haven't
done
a
great
job
of
that,
yet
we
have
some
documentation,
but
there's
not
clear
call
to
actions,
and
we
have.
We
have
one
community
contribution
for
PHP
composer
in
progress,
we're
hoping
to
take
the
learnings
from
that
and
turn
that
into
better
documentation,
but
I
think
that's
a
great
idea.
F
A
No,
no
we're
now
gonna
figure.
We're
gonna
proceed
to
figure
out
how
to
fix
this,
but,
like
the
lesson
here
is
you
could
change
the
documentation
in
half
an
hour
and
link
to
and
an
add
a
link.
So
we
welcome
all
the
formats
we
especially
would
like
to
see
and
then
just
add
everything
that
the
competition
has.
A
Here's
an
example
of
someone
who
contributed
something
with
a
link
to
an
mr
that
would
cost
you
30
minutes.
It
will
probably
double
the
amount
of
contribution,
so
we
would
get
this
enormously
high
leverage
to
do
that.
But
what
you're
doing
now
is
you're
you're
waiting
for
the
for
the
perfect
thing.
So
don't
wait
ship
ship
what
you
can
do
now
in
half
an
hour
it
will.
It
will
probably
be
over
a
thousand
hours
of
contributed
engineering
you'll
generate
with
that
half
hour
of
work.
That's.
F
A
Some
other
suggestions
were
like:
maybe
he
can
hold
office
hours
for
the
people
contributing
and
invite
all
of
them.
So
if
you
have
4
going
on
at
the
same
time,
stop
maybe
stop
pool.
Take
a
person.
That's
now
like
helping
with
actual
making
of
these
things
and
instead
make
them
an
enabler
of
people
who
want
to
contribute.
There's
a
lot
of
companies
using
gitlab.
We
probably
don't
have
to
make
anything.
We
just
have
to
enable
the
community
to
contribute.
So
the
mindset
should
not
be.
How
do
we
make
this
faster
but
like?
A
A
We
should
have
other
people
contributing,
and
there
just
seems
to
be
some
interest,
but
I
haven't
seen
like
a
blog
post
about
a
person
who
contributed
something
or
a
tweet
asking
for
people
or
like
these
office
hours,
where
someone
is
available
to
help
them
and
ask
questions
before
hey
I'm,
making
you're
gonna
contribute
this
you're
gonna
invest
hundreds
of
hours
of
your
own
time.
You
want
to
talk
to
someone
before
you
take
that
leap
and
make
sure
that
they're
open
to
it
and
everything
else.
A
So
this
is
office.
Hours
I
want
people
to
come
here,
so
you
don't
have
to
make
me
any
promises
to
me,
I'm,
just
giving
and
giving
ideas
to
you
to
decide
what
to
do
I'm,
not
telling
you
what
to
do,
and
then
it
takes
a
couple
of
iterations
to
get
there.
That's
not
good
right.
We
want
to
want
to
get
things
in
three
weeks
in
now,
it's
hard
to
create
value
in
that
time.
A
That
is
there
something
we
can
ship
that
doesn't
create
value,
yet
maybe
a
push
only
where
you
have
a
new
you
have
like.
Let's
take
the
new
get
think
you
can't
push
NuGet
packages,
you
just
can
pull
them
or
maybe
an
API,
only
thing
where
you
don't
have
the
interface,
what
it
reverse
only
an
interface,
but
not
the
API,
or
only
ship,
the
docks
of
how
it's
supposed
to
work.
A
A
A
It
was
a
lot
of
ship,
multiple
docker
contact,
there's
a
lot
of
work
to
do
that,
so
that's
kind
of
the
the
painful
stuff
we
have
to
do
to
make
sure
that
it
ships
in
a
month
like
put
very
artificial
constraints
on
it
that
that
reduce
the
value
but
hopefully
not
take
it
away.
So
I
mentioned
some
things
that
totally
take
away
the
value
like
only,
but
that
only
one
docker
container
per
per
project
that
was
like
there's
still
value.
A
It's
just
there's
a
very
obvious
gap
here
and
that's
the
same
with
like
NPM
and
two-factor
authentication.
Well,
he
can
push
the
NPM
back
home
before
for
security.
How
valuable
is
it
well?
A
lot
less?
Are
there
any
trade-offs?
You
can
think
of
that
people
can
do
who
contribute
a
new
package
format
I.
F
A
D
A
A
It's
it's
like.
The
comparison
is
right
now:
you're,
not
shipping,
anything
so
we're
pretty
sure,
but
pretty
sure
that
if
you
don't
ship
anything
there's
no
value
in
it.
So
the
only
thing
that
has
happened
is
that
there
a
little
bit
of
value
and
that
we
don't
like
lose
people's
data
or
a
security
flaw
or
mess
up
the
technical
architecture
of
Gilad
and
things
like
that.
I
won't
destroy
any
value.
But
as
long
as
there's
a
little
bit
of
value,
that's
enough.
There
might
be
a
good
move.
B
Sure
so
this
one
is
a
request.
We're
hearing
I
would
say
more
often
now,
with
some
more
larger
potential
customers
and
customers,
that's
either
their
multinational
organizations
or
their
highly
regulated
in
their
industry,
whether
it's
public
sector
government
contractors,
what
what
have
you
and
they
have
a
need
for,
did
a
residency
requirements
or
other
types
of
regulatory
reasons
that
they
have
to
sort
of
have
data
and
projects
in
certain
countries
and
not
on
other
ones.
And
this
and
so
for
some
cases.
B
For
example,
you
can
imagine
you
have
an
instance
in
the
host
the
United
States,
when
instance
hosted
in
Europe,
and
they
have
80%
of
the
same
projects,
but
there's
a
20%
on
either
side,
which
is
unique
to
those
instances
and
the
green
like
it
lab
is.
You
can
currently
solve
this
today
by
just
running
two
instances,
but
you
don't
have
problems
with.
B
Like
inner
sourcing
code,
discovery
code,
reuse,
understanding,
which
server
you're
on
which,
where
should
a
project,
be
created?
A
number
of
these
challenges,
which
then
kind
of
spun
out
of
having
a
number
of
these
things
out
there,
and
so
that's
something
that
we're
hearing
from
from
some
customers
who
that
they'd
like
to
do
and
their
current
model
is
generating
issues
around.
You
know,
building
the
same
thing
multiple
times
and
other
things
like
that.
Yeah.
A
A
You're,
probably
a
multinational
you're,
a
multinational
corporation.
We
understand,
we've
we've
seen
this
problem
before
and
this
is
how
people
are
dealing
with
it.
These
are
the
relevant
laws
and
here
are
your
option.
Some
of
these
options
we
already
ship
and
somewhere
in
the
future
and
an
option,
is
just
have
different
instances,
but
the
drawbacks
are
well.
People
can't
collaborate
that
well
anymore,
have
separate
instances
but
use
like
sync
mechanism.
We
have
on
a
pro
repo
basis
and
we
we
can
push
together
the
github
pull
from
github.
A
G
So
a
geo
selective
sync,
so
it's
actually
interesting
from
a
geo
perspective,
because
I
think
the
primary
use
case
for
geo
is
to
you
know,
replicate
things,
whereas
here
some
of
the
use
cases
to
specifically
separate
like
some
bits-
and
so
it's
very
related,
but
it's
almost
the
opposite
of
I-
think
the
GeoDesign
try
to
accomplish.
But
you
can't
use
selective
sync
to,
for
example,
restrict
what
gets
synchronized
on
a
project
basis
to
another
instance.
G
But-
and
you
know
that
is
sometimes
an
issue
and
I've
seen
that
with
customers
we
we
rely
currently
on
syncing
the
entire
post
post
database,
and
we
have
some
future
work
to
do
there
to
figure
out
how
to
overcome
that.
It's
not
necessarily
straightforward,
but
I
just
wanted
to
point
that
out.
So
it
can
help
in
some
instances,
but
it
has
some
limitations
as
well,
especially
first
of
compliance,
conscious,
yeah,
I.
A
A
I
think
that's
the
first
thing
and
I
already
put
on
in
6e
I
put
this
out
before
the
call.
No
Patrick
didn't
put
that
in
I.
Put
that
in
Danielle
it's
hard
to
do
for
the
database
and
easy
to
do
for
github
and
I.
Think
it's
a
strength
we
should
embrace.
So
instead
of
saying.
Oh
we're,
not
gonna
doesn't
solve
all
the
problem.
So
we're
not
going
to
do
this
I
think
there's
a
lot
of
sensitivity
around
source
code
and
there's
less
sensitivity
around
issues.
A
There
is
sensitivity
around
it,
but
still
like
people
are
also
emailing
and
those
emails
end
up
in
different
municipal
in
different
countries
and,
like
there's
perfect,
it's
very
very
hard
in
this
case,
so
we
should
definitely
expose
what
we
can
do
right
now.
So,
apart
from
geo,
selective
sync
I
think
we
should
also
have
a
get
early
forced
location.
You
can
say:
ok,
this
project
should
only
live
on
a
get
early
server.
That's
in
amia
or
something
like
that.
I
think
that's
I
can
see
that
being
built.
We
already
have
multiple
gotelli
servers.
A
They
are
determined
on
a
per
project
level.
It
would
not
be
an
insurmountable
amount
of
work
in
my
opinion,
but
stan
is
in
Nicole,
so
he
can
tell
to
add
labels
to
get
to
these
servers
like
what
region
they're
in
what
the
trust
level
of
that
environment
is
whatever
and
then
say
you
can
add
labels
to
project
and
they
have
to
match
every
label
off
that
server.
This
is
kind
of
the
same
way.
We
do
it
for
runners,
Stan
all.
D
A
A
A
A
feature
for
that:
okay,
I
was
gonna
proceed.
How
that
was
a
dumb
idea,
but
there's
there's
a
there's,
an
issue
for
that
so
store.
The
code
gives
in
object,
storage,
now
obvious
storage.
We
can
have
multiple
types
of
objects
search.
We
can
do
maybe
the
same
labeling
thing
if
it's
amia
and
HIPAA
compliant.
You
can
only
store
it
in
the
AMIA,
HIPAA
compliant
objects,
storage
and
that's
how
you
kind
of
keep
eating
away
and
there's
still
the
problem
of
the
rest
of
the
database
and
everything
else.
A
I,
don't
think
the
solution
we're
gonna
find
for
the
database,
it's
going
to
be
something
that
would
help
that
would
lead
to
a
different
solution
for
the
repositories
or
the
code.
The
code
is
I,
guess
I,
think
those
are
probably
separate
things,
so
we're
not
we're
not
making
we're
not
making
the
situation
any
worse,
and
we
then
have
like
the.
A
If
we
do
this
for
source
code
and
code
snippets,
they
have
like
the
best
or
soon
say
snippets
code
if
we
have
the
best
solution
in
the
market
and
if
you,
if
you're
sensitive
or
where
your
source
code
is
being
stored.
But
you
still
want
to
collaborate
with
the
entire
company.
Our
solution
will
be
better
than
any
other
vendors.
That's
that's
a
good
place
to
be
at
and
then
we'll
get
more
requests,
and
this
will
never
be
finished.
B
B
But
thanks
it
I
think
it's
definitely
agreed.
Ox
is
a
great
place
to
start
and
applies
for
a
lot
of
conversations.
It
also
help
to
steer
people
who
are
interested
in
this
to
some
of
the
issues
to
help
get
better
feedback
and
validation
and
those
issues.
So
it's
a
really
good
idea
and
thanks
obviously,
for
your
next
iterations
beyond
that
as
well.
Thanks.
G
G
A
That's
a
great
idea,
like
download
kind
of
the
knowledge
that
is
now
in
the
heads
of
people
into
documentation
and
tools,
tools
are
kind
of
like
active
documentation,
tests
are
active
documentation
of
the
system
and
it
can
be
run
by
the
computers
anytime.
You
can
make
it
into
a
test
or
a
tool,
it's
better
than
make
it
into
Docs
that
people
have
to
process
themselves.
It
makes
it
make
sense,
and
it's
on
a
higher
level
like
we
have
so
many
people
inside
the
company.
Now
that
it's
automatically
you
start
thinking
about
it
like.
A
B
This
is
Gabe.
One
question
I
think
it's
it's
about,
iterating
our
maturity
models
for
our
categories
have
some
definitions
about
when
things
move
between
different
stages,
but
it's
also
fairly
vague,
and
it's
for
me
being
here
for
months,
it's
hard
to
understand
how
to
define
something
as
moving
between
maturity
levels
based
on
the
definition
which
I
think
is
along
the
lines
of
you
know.
B
A
A
So
loveable
is
now
defined,
as
provides
an
elevated
user
experience,
so
maybe
build
a
bit
of
a
platform
for
people
either
it's
a
platform.
After
all,
so
you
get
an
elevated
user
experience
that
customers
love
is
measured
by
D
NPS
product
Net,
Promoter
Score,
that's
it
that's
it.
The
most
quantitative
criteria
expect,
except
we
didn't
say
what
more
MPs
like.
Would
you
recommend
this
feature
of
get
lab
to
a
friend
will
probably
be
the
question.
A
B
A
Made
up
to
PMPs
I,
don't
know
weeks
or
months
ago,
because
we
set
a
company
goal
in
November,
18
2023.
We
want
to
make
sure
that
half
of
our
features
are
lovable
I
like
okay.
Well,
what
does
that
mean
like
it's
pretty
easy
to
achieve?
If
you
just
make
make
it's
it's
a
soft
definition,
so
I
want
to
make
that
a
bit
harder,
but
because
we
never
measured
a
PMPs,
we
don't
kind
of
know
how
to
where
to
calibrate.
A
So
we
should
probably
do
a
PMPs
survey
at
some
features
where
we
know
they're
lovable,
like
see
I,
add
some
features
where
we
know
we
serve
a
very
long
way
to
go
and
then
see
how
those
scores
pan
out
and
then
we
we
can
add
those
scores
here.
We
can
go
from
these
very
qualitative
criteria
to
more
quantitative
ones.
Yeah.
A
And
our
product
managers
can
stop
wondering
about
where
something
is
and
when
are
when
our
users
say
some
something
is
of
has
a
high
enough
PMPs
or
is
is
good
enough
to
be
recommended.
We
can
celebrate
that.
We
should
you
shouldn't,
be
your
own
score
keeper,
if
at
all
possible
agreed.
Thank
you
thanks
for
the
question
Emily.
Can
you
make
sure
this
ends
up
an
art
agenda?
Yes,.
A
J
J
A
Not
sure
I
thought
of
get
subtree
as
a
one-time
thing
like
that,
you
super
hard
to
add
other
stuff
later
on,
so
I
would
bring
the
images
over
as
well
I
would
bring
us
more
time
to
walk
over
at
the
same
time,
because
otherwise
I'm
not
sure
what
happened
to
your
history.
If
someone
has
market
knowledge
and
I'm
sure
some
well
I
have
more.
If
someone
knows
whether
you
can
do
it,
iteratively
that'd
be
helpful,
but
I
think
this
is
an
example
of
super
hard
to
iterate,
oh
by
the
way.