►
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
Welcome
everybody
very
excited
that
we
have
tristan
handy
here
with
us,
ceo
and
founder
of
dbt,
just
a
real,
quick
intro
about
tristan.
He
is
someone
out
of
a
philadelphia
startup
started
formally
as
fish
town
analytics
he's
been
pioneering.
The
modern
analytics
engineering
space
dbt
is
currently
used
by
over
5
000
companies
to
organize
catalog
and
distill
knowledge
from
the
data
in
their
data.
Warehouses,
including
companies
like
jetbluehubspot.
A
Of
course,
us
gitlab
and
the
aclu
are
a
few
tristan
has
been
working
in
the
data
space
for
two
decades
in
both
in-house
and
consulting
roles
for
large
enterprises
and
even
small
startups,
tristan,
we're
so
happy
to
have
you
here.
A
Thank
you
for
being
here
and
sharing
some
time
with
us
today
appreciate
you
for
having
me
yeah,
so
I
thought
I'd
kick
it
off
with
just
one
question
that
I
thought
was
really
interesting
that
I
don't
even
know
if
you
thought
about
it
when
you
made
the
comment,
but
you
know
during
a
coffee
chat.
You
mentioned
one
of
the
dbt
values
and
you
know,
values
at
gitlab
are
very
important
and
we
live
by
them
every
day.
A
B
Yeah,
oh
geez,
so
so
the
the
text
of
the
value
is,
we
are
human
and
it
often
does
translate
into
bring
your
whole
self
to
work
or
come
as
you
are.
B
We
we
were
talking
about
it
because
we
were
talking
about
virtual
backgrounds
and
how
we
often
like
to
share
the
mess
that
is
is
present
in
our
lives,
because,
that's
just
you
know,
that's
the
state
of
the
world
it
most
of
our
values
are
were
actually
penned
back
before
the
corporate
entity
was
even
formed,
but
this
one
is
one:
that's
been
on
an
evolution
and
it's
it's
an
evolution
that
that
I
think
is
reflected
like
our
own
understanding
of
like
how
how
we
wanted
to
show
up
at
work.
B
There's
been
a
lot
of
decisions
that
folks
have
made,
as
as
the
world
has
changed
over
the
past
five
years,
and
how
much
of
yourself
do
you
bring
to
work,
especially
in,
and
I
don't
think
I'm
saying
anything
particularly
surprising,
but
the
the
world's
been
an
interesting
place
over
the
past
five
years,
and
we
we
have
to
choose
like
how
much,
how
much
of
that
do.
We
show
up
with
and
share
with
our
our
colleagues
and
how
much
of
that
do
we
try
to
leave
leave
at
home?
B
It's
not
always
easy
to
show
up
and
like
bring
all
of
that
to
work
with
you,
but
I
think
that
it
creates
a
pretty
unique
environment
here
and
we've
learned
how
to
have
hard
conversations
and
sometimes
conversations
that,
like
folks,
are
not
usually
like
not
used
to
having
him
at
work.
I
hope
that
does
at
least
a
reasonable
job
of
answering
that.
A
Yeah
for
me
it
does
I
I
love
that
value
and
I'm
always
interested
to
see
how
other
companies
really
find
that
identity
and
in
how
they
operate,
and
I
think
that
particular
value
certainly
aligns
very
much
to
how
gitlab
operates.
So
thank
you
for
sharing
that,
let's
jump
into
q
a-
and
I
think
I
believe
I
think
that's
rob
parker
rob
parker.
I
think
you
have
the
first
question
on
voiceover.
C
Sure,
tristan
thanks
for
joining
it's
fantastic,
to
see
you
and
have
your
presence.
So
I
appreciate
your
time.
I've
been,
and
my
team
has
been
a
dbt
customer
for
a
number
of
years
and
appreciates
my
my
start
here
at
get
lab.
But
it's
a
very
powerful
technology
coming
to
gitlab
was
my
first
experience
with
it
and
I
can
see
the
community
and
your
customer
base
just
growing
and
growing
and
growing.
C
So
congratulations.
How
did
you
come
about
dbt,
like
what
tools
did
you
use
before
dbt?
What
were
some
of
your
tools
that
you
grew
sick
and
tired
of
or
examples
that
led
you
to
creating
dbt.
B
It's
it's
a
it's
a
very
big
question.
I
will
I'll
try
to
keep
keep
my
answer
to
it.
The
reasonably
confined
set
otherwise
we'll
spend
the
whole
time
on
this
there,
the
actual,
like
specific
journey
that
I
was
like.
I
need
a
thing
that
looks
exactly
like
dvt
involved,
redshift
and
mode
a
little
bit
of
looker
thrown
in
there,
and
I
I
felt
that
the
this
whole
I
I
kept
writing
a
bunch
of
sql
code.
I
kept
writing
you
know.
B
Looker
does
its
own
kind
of
data
transformation.
I
I
felt
like
there
was
a
hole
in
the
workflow
and
I
rather
than
writing
this,
like
150
line
long
query
in
mode.
I
wanted
to
like
work
more
like
a
software
engineer
and
break
this
into
like
distinct
chunks
of
logic,
and
I
I
actually
I
had
used
a
product
much
earlier
in
my
career
called
oracle.
B
Warehouse
builder
owb
was
like
early
2000s
and
you
would
like
you'd
start
with
a
blank
canvas
and
you
would
like
draw-
or
you
would
like
put
these
like
things
icons
on
the
and
then
you
create
a
flow.
You
would
drag
lines
between
them
and
the
anyway.
It
was,
it
was
a
product
of
that
era
and
in
many
ways
a
sub-optimal
experience
and-
and
so
I
was
like
okay,
I
need
that
thing.
I
need
like
the
thing
to
do
that
job,
but
how
does
it?
B
How
should
it
look
today
and
the
the
two
inspirations
that
I
took
were
one
terraform
and
two
ruby
on
rails?
I
had
never
used
terraform
before,
but
my
co-founder
connor
was
just
getting
into
terraform
in
2015
2016
and
he
showed
me
how
it
worked
and
I
was
like
that's
like
literally
the
exact
user
experience
that
I
want,
and
then
I
was
thinking
about.
B
I'm
not
a
very
good
like
software
engineer,
but
I
have
done
some
web
development
and
a
lot
of
that's
been
in
ruby
on
rails,
and
I
I
was
like.
I
want
like
a
framework
like
that.
That
kind
of
gives
me
a
a
kind
of
a
sense
of
how
to
work,
because
I
want
to
be
able
to
write
code
and
express
the
business
logic,
but
I
don't.
B
I
don't
think
about
like
how
a
request
flows
from
like
you
know
the
beginning
to
the
end
of
the
web
stack
so
that
that's
kind
of
like
where
all
the
soup
of
ideas
came
from.
D
Yeah,
thank
you
I
was
wondering.
Is
it
harder
to
grow
an
open
source
product
compared
to
a
traditional
product
and
especially
in
the
data
world,
because
in
my
opinion,
the
data
world
is
using
quite
some
proprietary
software
currently.
So
how
is
the
data
world
responding
to
open
source
software
here.
B
Yeah,
it's
a
it's
a
very
good
question.
I
think
that
here's
the
like
really
boring
answer
in
some
ways:
it's
easier
in
some
ways
it's
harder
and-
and
I
think
that
you're
like
100,
absolutely
right
that
data
historically
has
been
primarily
proprietary
products.
I
I
do
think
that,
like
especially
data
infra
is
moving
towards
being
more
open
source
focused,
especially
companies
that
are
founded
now
or
within
the
past
year
year
or
two,
so
that
makes
me
happy
just
as
a
user
like
I.
B
I
think
that
that
is
just
like
good
for
the
world.
Okay,
the
things
that
are
easy
easier
because
of
open
source.
We,
I
think,
for
a
proprietary
software
company.
You
have
to
actually
set
out
with
a
very
specific
intention
of
like
building
proprietary
software
company.
Like
you,
you
don't
kind
of
like
accidentally,
build
a
proprietary
software
company
and
with
open
source.
You
have
a
lot
more
flexibility.
B
B
But-
and
I
don't
think
you
can
have
that
journey
as
a
proprietary
software
company,
you
can't
just
like
magically
develop
a
community
around
something
that
people
have
to
pay
for
and
and
so
I've
I've
really
found
the
open
source
journey
to
be
like
much
more
aligned,
with
the
way
that
I
like
to
think
very
iteratively
about
the
formation
process,
the
it's
it's
like,
de-risked,
a
lot
of
it
for
us,
but
but
the
hard
part,
as
my
guess
is
that
a
lot
of
you
folks
are
very
well
aware
of.
B
Is
that
you
don't
get
to
sell
the
core
thing
like
you
have
to
like
develop
this
core
thing,
and
people
have
to
care
about
that
and
love
it
and
and
then
you
have
to
like
build
another
thing
or
multiple
other
things
and
and
those
have
to
be
independently
good
enough,
that
people
will
will
pay
for
them,
because
if
you
show
up
on
a
sales
call
and
you're,
just
like
hey
dbt's
awesome,
why
don't
you
pay
us
for
it?
They're
like
no,
we
can
just
get
that
for
free,
so
it's
I.
B
I
think
that
we
probably
we
like
all
open
source
companies,
have
to
be
doing
multiple
things
all
at
the
same
time
and
it
feels
a
little
chaotic
and
feels
like
there's
never
enough
resources
to
actually
make
everyone
happy.
You
know
you
folks
are
way
further
along
this
journey,
so
maybe
you
have
achieved
that
state
of
nirvana,
where
you
actually
have
enough
software
engineers,
but
we
always
feel
like
we
need
double
or
quadruple
the
number
of
product
people
and
engineering
people
that
we
need.
C
I
don't
think
sushmazon
so
we'll
voice
over
her
question.
What
next
big
releases
or
features
in
dbt
should
be
we'd,
be
looking
forward
to.
B
B
Let
me
try
to
recover
my
train
of
thought:
oh,
we
finally
1.
after
five
and
a
half
years
and
the
so
it's
this
big
milestone.
It's
like
we
get
to
figure
out.
Okay.
What
do
we
want
to
tackle
next,
like
the
the
core,
the
core
dbt
experience
is
performant.
It
is
stable.
We
like
understand
the
the
user-facing
epi,
and
so
one
of
the
big
ways
in
which
we
want
to
push
things
forwards.
B
Is
we
want,
if
you
think,
of
dbt
as
a
essentially
a
programming
framework
or
a
language
or
a
compiler?
B
Dbt
takes
a
templating
layer
on
top
of
sql
and
then
allows
you
to
express
a
bunch
of
things
in
that
templating
layer
that
you
couldn't
express
in
real
sql
and
then
it
translates
it
compiles
all
that
down
into
raw
sql
and
runs
it
against
databases
and
so
right
now
the
technical
underpinnings
of
that
compiler
are
very
suitable
for
a
certain
set
of
workloads,
specifically
batch-based
data
transformation.
So
that's
generally
how
people
use
dbt.
B
Today
they
write
a
bunch
of
pipelines
and
then
they
say,
okay
dbt
go
and
go,
make
it
so,
and
dbt
in
batch
will
run
all
of
this
stuff.
B
That's
a
you
know,
big
chunk
of
data
workloads
that
exist
in
the
world
today,
but
it's
by
no
means
all
of
them,
and
so
there's
this
powerful
community-built
programming
construct
that
can
only
be
used
by
a
certain
set
of
workloads
and
what
we
want
to
do
is
bring
these
same
programming
constructs
into
other
data
workloads,
and
so
the
first
one
that
we
are
looking
at
is
real-time
metrics,
query,
time
metrics,
and
so,
if
you,
if
you
think
about
okay,
there's
a
certain
set
of
like
data
preparation,
you
like
build
a
customer's
table.
B
You
want
a
really
clean
customer's
table,
so
you
can
go
analyze
it.
The
second
type
of
workload
that
we're
moving
into
metrics
is
rather
than
like,
preparing
the
customers
table
ahead
of
time.
You
say
like
okay,
we've
got
10
metrics
they're
all
built
on
top
of
the
customers
table
and
I
might
want
to,
and
maybe
one
of
those
metrics
is
simply
just
customer
count,
and
I
want
to
be
able
to
access
this
customer
account
metric
and
all
the
different
bi
tools
that
I
use.
B
So
you
can
be
100,
confident
that
everybody's,
using
the
same
definitions
for
these
metrics
and
and
so
what
dbt
is
going
to
do,
is
it's
going
to
become
this
proxy
server?
Where
you
can,
you
can
actually
from
your
bi
tool,
say
like
hey.
I
want
the
customer
count
metric
and
in
this
proxy
server
layer.
Dbt
will
actually
intercept
that
query.
It
will
compile
it
to
raw
sql
to
get
the
customer
account.
Now
we'll
ask
database
that
and
we'll
return
it
so
it's,
whereas
dbt
previously
did
not
participate
in
query
time
analytics.
B
You
will
now
be
able
to
use
the
same
kind
of
programming
constructs
in
in
your
your
bi
layer
as
well.
Dbt
is
not
going
to
be
a
bi
tool.
It
was
one
of
the
you
know
you
make
you
make
these
agreements,
you
sign
in
blood,
with
your
co-founders
when
you,
when
you
start
a
company,
one
of
the
agreements
that
that
your
encounter
and
I
made
was
we're
not
building
a
bi
tool.
B
We've
done
that
before
in
our
lives
not
doing
it
again,
but
that
doesn't
mean
that
we
can't
empower
bi
and
analytics
workflows
with
data
infrastructure,
and
so
that's
that's.
What
we're
trying
to
do
with
this
next
phase.
E
Yes-
and
it
actually
fits
perfectly
with
what
has
been
just
discussed,
because
dvd
currently
so
dvt
as
a
tool
has
access
to
snowflake
to
data
to
data
flows,
do
you
have
any
plans
to
move
into
to
expand
into
data
observability.
B
C
B
Makes
sense,
let
me
answer
that
question
from
like
a
metadata
perspective,
more
broadly,
because
I
I
think
of
observability
as
like
a
use
case
within
this,
like
broader
metadata
problem
space.
B
I
think
it's
really
hard
to
to
separate
out
some
of
these
problems
and
say
this
observability
is
distinct
from
all
of
these
other
problems,
and
the
same
would
be
true
of
of
data.
Catalog
same
would
be
true
of
testing
or
data
quality.
So
when
I'm
you
know
I
I
still
you
know
my
bi
tool
of
choices
is
mode.
Not
I
don't
want
to
make
an
argument,
that's
better
than
any
other.
B
Just
like
that's
my
habit,
and
so
when
I
am
like
doing
an
analytical
like
going
through
the
analytical
process,
I'm
primarily
living
inside
of
the
mode
interface
to
do
that,
and
what
I
do
not
want
is.
I
do
not
want
to
jump
to
four
different
experiences:
to
try
to
navigate
a
data
catalog
a
data
observability
solution,
a
data
quality,
so
I
I
don't
that
doesn't
make
any
sense.
From
my
from.
Like
a
user
perspective,
I
think
that
the
reason
that.
B
This
is
like
a
a
category
right
now
that,
like
companies
are
getting
started
and
talking
about
data
observability,
it
is
less
about
that,
like
users
want
a
separate
data,
observability
pro
solution,
it's
more
that
if
you're
a
company-
and
you
say
like
hey,
there's
a
problem
with
observability
you're
like
well
I'll,
build
a
tool
and
it'll
be
separate.
I
think
that
the
way
that
we
want
to
see
this
go
and-
and
I
I
do
believe
that
dbt
will
be
a
part
of
this
solution-
is
that
we
you're
right.
B
We
want
to
solve
this
via
apis
and,
if
you
think
about
how
did
twilio
solve
messaging,
they
didn't
do
it
by
like
creating
a
marketing
automation
tool
but-
and
they
might
have
something
in
that
front
now,
but
they
like
that's
not
how
they
like
went
about
solving
that.
Fundamentally,
they
said:
okay,
there's
a
million
different
reasons
that
people
might
want
to
send
messages.
B
We're
going
to
just
like
make
a
set
of
apis
that
that
enable
people
to
do
that
in
whatever
context
they're
in
and
so
when
we
think
of
observability
or
cataloging
or
quality
or
other
things
like
this.
We
think,
okay,
let's
make
sure
that
we
have
the
whole
universe
of
information
available
and
then,
let's
like,
create
apis
that
bi
tools
that
other
you
know.
B
Your
data
apps
that
you
build
internally
can
very
easily
access
all
of
that
metadata
and
give
you
a
cohesive
single
experience
to
say,
like
yeah,
I
want
to
discover
a
new
data
set,
I'm
going
to
do
it
inside
of
mode
I'm
going
to
look
up.
You
know
whether
that's
been
built
recently,
whether
it's
a
quality
etc.
So
I
do
think
that
there
is
a
role
for
dbt
to
play
there,
but
I
don't
you
know
I
want
to
be
careful
that,
like
we
do
not
see
ourselves
as
like
building
a
dead,
observability,
competitive
solution.
F
G
Yeah,
hello,
tristan
radwan,
senior
data
engineer
and
thanks
for
creating
a
great
product,
really
make
our
life
easier
and
nicer,
and
what
I'm
interesting
for.
Actually
I
have
two
questions,
the
first
one
in
how
I
feel
about
and
how
you
manage
drive
versions
and
releases.
I
know
you
have
paid
free,
also
a
lot
of
plugins.
G
B
I'm
trying
to
I'm
trying
to
decompose
that
into
different
questions,
so
you're
right,
there's
some
different
surface
area.
We
have.
We
have
dbt
packages
so
like
we,
we
maintain
the
most
widely
used
dbt
package
out
there
called
dvt
utils,
that's
in
use
by
maybe
40
of
all
dbt
projects.
Then
we
have
the
adapters
that
connect
to
the
databases
and
those
have
recently
gotten
split
out
into
their
own
repositories
and
we
maintain
like
the
four
or
five
biggest
ones
and
then
the
community
does
the
rest
as
a
percentage
of
like
total
effort.
B
Those
two
things
are
like
reasonably
small
they're:
they
they're
not
large
ticket
items.
So
then
it
comes
to
really
like
dbt,
core
and
dbt
cloud.
These
are
like
the
two
main
code
bases
that
we
maintain
and
it
turns
out
that-
and
you
know
you
probably
I'll,
I'm
preaching
the
choir
here.
It
turns
out
that
it's
actually
there's
more
work
involved
in
building
and
hosting
maintaining
a
cloud
service.
Then
there
is
a
piece
of
downloaded
open
source
software.
B
There's
a
bunch
of
like
when
you
put
all
of
the
operations
and
maintenance
in
in
the
hands
of
the
users.
Then
it's
just
like
a
lot
of
stuff
that
you
don't
have
to
think
about.
So
it
ends
up.
We
end
up
spending
more
engineering
bandwidth
on
cloud,
but
but
the
the
choices
that
a
lot
of
open
source.
B
I
you
know
I've
had
the
very
I'm
very
fortunate
to
have
had
the
opportunity
to
talk
to
a
bunch
of
open
source
founders
and-
and
I
get
the
sense
that
this
question
like
what
what
do
we
put
in
free
versus
paid
or
open
versus
closed,
that
that
comes
up
a
lot
and
is
like
can
be
a
contentious
issue.
B
I
I
think
that
the
way
that
we
have
architected
things
from
the
ground
up
makes
that
a
little
bit
less
of
a
a
challenge
for
us
and
and
it's
not
because
I
think
that
we're
like
so
clever
and
we
did
it
all
right.
I
think
it's
more
that
we
got.
We
are
fairly
recently
started
open
source
company.
B
You
know
if
we
got
to
watch
confluent
and
data
bricks
and
and
elastic,
and
you
folks,
like
we
got
to
watch
all
these
companies
make
these
decisions,
especially
a
lot
of
our
thinking,
gets
2018
or
2019
that,
like
all
the
big
re-licensing
happened
anyway,
the.
B
There
are
very
clear
scopes
for
core
versus
cloud:
for
us
core
does
two
things:
it
does
language
and
it
does
database
connectivity.
So
if
you
want
to
create
a
programming
construct,
it's
always
going
to
go
in
core
100
of
the
time.
If
you
want
to
do
something
relative
to
like
how
dbt
interacts
with
databases,
it's
always
going
to
go
into
core.
B
If
you
want
to
do
something
that
require
requires
statefulness
or
persistence,
or
just
all
the
things
that
cloud
services
are
good
at
or
or
user
experience,
those
are
kind
of
the
two
categories,
like
persistence
and
user
experience,
then
it's
going
to
go
into
cloud
and,
and
then
cloud
actually
consumes
core
and
so
by
building
core
functionality
it
it
rolls
out
into
the
commercial
product.
So
this
is
like
worked
pretty
well
for
us
for
our
community.
We've
never
really
had
anybody.
B
I'm
that's
a
big
statement,
but
I
I'm
not
aware
of
the
kinds
of
tension
that
have
run
through
some
open
source
communities
in
the
past
about
how
this
line
is
drawn.
So
I
I
think
that
that's
an
answer
that
we're
going
to
be
pretty
consistent
with.
G
Yeah
thanks
a
lot
for
that.
I
have
one
more
question:
it's
just
a
follow
up
on
observability
tools
and
I
know
dbt.
It
stands
for
transformation
or
transformer
tool
for
now,
do
you
have
any
plans
to
catch
up
that
wave
to
provide
full
retail
solutions
like
kobi,
as
you
mentioned,
like
talent,
or
something
like
that,
or
you
want
to
stay
only
in
this
niche
in
transformation
world,
or
you
want
to
create
some
strategic
partnership
with
for
parties
to
provide
the
full
set
of
of
functionality
for
retail
processing,
realty
processing.
G
B
B
Good
question
and-
and
I
love
that
this
is
being
archived
on
youtube
forever,
because
I'll
give
you
my
best
answer
today
and
then
in
ten
years.
Somebody
will
point
back
to
this
and
be
like
you
said
the.
B
There
are
like
real
differences
between
the
engineering
task
of
moving
data
around
versus
building
a
compiler.
B
We
have
an
organization
that,
like
knows
how
to
build
compilers
knows
how
to
build
user
experiences
around
ides
that
that
kind
of
stuff.
We
would
have
done
a
lot
of
things
very
differently
if
what
the
problem
that
we
wanted
to
solve
was
how
do
we
move
data
from
salesforce
to
snowflake
or
whatever?
And-
and
I
honestly
don't
if,
if
I
you
know-
had
a
vision
tonight
and
woke
up
tomorrow
and
was
like
the
future,
is
data
you
know,
extract
and
load.
B
I
actually
don't
think
we'd
be
particularly
good
at
that
as
an
organization,
that's
just
like
not
how
we're
wired,
and
so
I
I
think
that
one
of
the
reasons
that
I'm
most
bullish
on
the
modern
data
stack
is
that
we
have
reasonably
clear
layers
and
areas
of
responsibility
where
different
solutions
interface
nicely
together,
and
they
they
all.
B
We
we
all
partner,
pretty
well
and
we
co-sell
pretty
well,
so
I
I
hope
that
that's
the
way
that
things
continue
to
be
for
a
long
time
now
in
the
fullness
of
time,
will
there
be?
B
You
know
some
acquisition
wave
where
you
know
snowflake
buys
every
company
that
has
data
in
it.
I
I
have
no
idea,
we
will
see
where
things
go,
but
we
have
no.
F
B
I
don't
think
that
I
am
at
liberty
to
share
more
information
on
that
right
this.
Second,
I
am
very
excited
that
it's
funny
you
can.
You
can
like
read
into
who
kind
of
got
the
avalanche
started.
There's
been
a
lot
of
interest
in
the
metrics
layer
in
the
past,
whatever
like
eight
months,
and
you
can
map
a
lot
of
that
interest
back
to
a
newsletter
that
was
published
by
ben
stancil
sometime
over
the
summer.
I
forget
that
he,
I
think
he
just
called
it.
B
The
metrics
lair
I
forget,
but
his
he
writes
at
bend.substance.com
he's
a
co-founder
of
mode
analytics.
So
I
won't
tell
you
exactly
who
is
going
to
be
a
launch
partner,
but
I
will
say
that
ben
has
been
very
formative
in
his
public
writing
and
you
can
take
some
assumptions
from
that.
H
Yeah,
I
think
I'm
next
thanks
ryan
yeah,
thanks
for
taking
your
time
today,
tristan
my.
H
Maybe
taking
a
slightly
different
angle
and
they're
thinking
about
yourself,
I
did
a
little
bit
of
linkedin
snooping
but
yeah
you
kind
of
described.
H
But
yeah,
I
think
for
me,
you
know,
as
someone
who's
caught
on
this
school
who's
been,
you
know,
moving
towards
more
management
role
in
the
last
few
years.
It's
something
that
I'm
kind
of
always
trying
to
like
get
the
right
balance
between
thinking
about
stuff,
and
you
know
executing.
I
just
wonder
if
you've
got
any
tips
on
how
you
get
a
good.
B
It's
balance.
It's
such
an
interesting
question.
I
I
am
really
worried
anytime
people
ask
questions
that
like
have
me.
Giving
kind
of
broad
advice
that
my
my
sample
set
of
one
is
is
not
actually
a
great
way
to
to
base
advice
on.
I
will
say
that.
B
I
was
a,
I
was
a
video
game
player
when
I
was
in
my
like
young
years
and
I
played
all
kinds
of
different
games
until
I
was,
I
don't
know
like
12
or
something
and
warcraft
2
came
out,
and
it
was
this
category
called
a
real-time
strategy
game
and
literally
the
minute
that
I
played
warcraft
2.
I
was
like.
B
I
have
no
interest
in
ever
playing
another
video
game,
like
I'm,
only
ever
going
to
play
real-time
strategy
games
and
the
because
it
they
are
essentially
like
this
combination
of
both
strategy
and
execution.
You
kind
of
start
off
with
this
blank
slate
like
you
can
do,
literally
whatever
you
want,
and
you
get
this
really
satisfying
answer.
At
the
end
of
the
day
like
either
you
won
or
you
lost
the
game,
and
so
your
strategic
and
execution
decisions
were
like
good
or
bad
anyway.
B
It's
like
it
happens
to
be
the
way
that
my
brain
works,
and
I
I
like
love
those
two
things:
the
things
that
people
the
way
that
people
who
have
brains
like
mine,
the
things
that
I
think
oftentimes
we're
not
good
at,
are
things
that
you
know
you
have
to
then
like
tack
on
later
in
life.
So
for
me,
I've
I've
had
to
like
learn
how
to
be
a
lot
more.
B
Intellectually
humble
I've
had
to
learn
how
to
like
see
people
like
truly
listen
to
people
better.
How
to
like
have
hard
conversations.
There's
like
this
whole
other
skill
set
that
I
like
really.
I
have
had
to
work
very
hard
to
add,
but
I
I
don't.
I
don't
know
that.
There's
like
some
magic
trick
about
like
at.
B
I
think
that
if
you
want
to
be
good
at
execution
and
go
to
strategic
thinking,
then
it's
like
put
yourselves
put
yourself
in
positions
where,
like
you,
can
exercise
those
muscles
and
like
continue
going
at
it,
I
I
didn't,
have
a
playbook.
I
just
like
continued
to
take
roles
where
I
had
enough
scope
where
I
could
and
sometimes
in
order
to
get
enough
scope.
You
have
to
join
like
some
tiny
little
company.
That
probably
will
fail.
B
H
Yeah
yeah,
that
was
interesting,
thank
you
and
I
think
I've
got
the
next
one
actually
I'll
try.
I
think
I've
kind
of
been
collating
some
answers
as
you've
been
talking,
but
the
question
was
what
kind
of
high
level
factors
would
you
attribute
to
the
success
of
dbt
to
date.
B
Yeah,
it's
another,
it's
another
one
of
those
trying
to
try
to
determine
causality.
I
I
will
I
kind
of
teased
this
at
the
beginning
of
the
conversation
open
source
gives
you
optionality
like
you,
don't
have
to
be
all
in
to
starting
a
venture
funded
business,
and
you
know
like
the
only
possible
outcome
that
is
successful,
is
a
billion
dollar
you
know
exit.
B
The
biggest
thing
that
has
enabled
us
to
be
successful
is
to
not
try
to
answer
the
question
like
what
does
this
thing
actually
want
to
be
until
we
had
more
data
and-
and
so
I
think
that
as
a
founder
one
strategy
that
is
widely
underemployed
is
optimizing
for
optionality
like
creating
as
many
possible
ways
that
success
could
look
as
as
possible,
and
that
requires
you
to
be
flexible
and-
and
you
know,
loves,
you
forces
you
to
live
with
uncertainty
and
a
lot
of
people
hate
that
a
lot
of
people
are
just
like.
B
I
want
success
to
look
like
that,
and
I'm
gonna
run
it
as
hard
as
I
can,
even
if
that
means
that
I
like
give
myself
a
concussion
running
into
the
wall
along
the
way,
but
I
think
that
that
is
really
a
huge
one
and
and
and
then
like
here's,
here's
the
other
one,
and
this
is
so
overplayed
and
I
apologize
for
even
saying
it.
But
I
think
if
you're
gonna
be
a
community
based
business,
you
actually
have
to
care
about
the
community.
B
B
The
company
actually
exists
to
support
the
project
as
opposed
to
the
other
way
around
and,
and
that
just
is
like
something
that
you
as
a
founder,
need
to
either
believe
or
not
believe,
and
if
you
believe
it,
you
have
to
make
sure
that
you
hire
people
and
that
believe
it
too,
and
that
you
make
sure
to
resist
the
temptations
to
pull
you
in
the
other
direction.
A
By
the
way,
tristan,
I
really
like
that
I
think
there's
a
little
sound
bite
there
that
you
should
a
little
bit
there
well
done
rob.
C
Go
ahead:
hey
tristan,
just
sharing
the
fun
fact
here
we
have
published
a
lot
of
great
content
in
our
handbook.
Gitlab's
handbook
is
one
of
our
superpowers.
I
believe
in
helping
to
spread
knowledge
and
information
just
around
the
world
about
how.
C
B
That's
really
neat.
Hopefully
it's
provides
a
good
good
hiring
opportunity
for
you,
folks,
I
I
will
you
can
scrub
this
part
out.
If
I'm
not
supposed
to
say
this,
but
I'm
looking
at
what
you,
the
screenshot,
that
you
shared
the
dbt
guide,
ranks
number
12
and
then
security
practices
ranked
number
22..
I'm
I'm
just
I'm
just
being
funny.
It's
probably
much
more
widely
applicable
for
other
people
throughout
the
dbt
community.
C
Cool
miles
you're
next.
I
Great
thanks
tristan,
I
attended
the
user
conference
back
in
december
and
one
of
the
cool
things
about
that
conference
was
when
they
showed
the
growth
of
dbt
projects
over
time,
and
just
I
think,
maybe
the
first
first
year
was
maybe
2016.
You
can
just
kind
of
see
the
hockey
stick
growth,
especially
in.
I
Years,
I
actually
started
using
dt
back
in
2018
and
one
of
the
things
back
then
that
is
still
true
today.
That
I
really
appreciate
is
how
just
the
the
feel
of
community
and
my
question
to
you
was
about
how
you've
been
able
to
dvt
live,
has
been
able
to
kind
of
maintain
this.
The
similar
community
feel
from
2018
when
I
first
joined
two
now
when
you've
had
all
this
growth.
I
B
I'm
I'm
happy
to
say
some
some
words
on
this.
I
I
you
know
it
just
strikes
me,
though,
that,
like
you,
might
have
more
insight
into
the
answer
that
question
than
than
I
do,
because
your
experience
as
a
community
member
is
is
like
actually
what
what
is
interesting
here.
Do
you
actually
feel
like
that's
true?
Do
you
feel
like
the
community
that
you
joined
in
2018
is
has
has
much
the
same
positive
characteristics
that
the
community
does
that
you
see
today.
I
So
from
the
very
beginning,
I
noticed
I
observed
the
slack
group
to
be
incredibly
helpful
and
it's
gotten
a
lot
bigger
now
like
there
are
a
lot
more
people
posting
in
channels
like
asking
questions
discussing
things,
but
like
that
same
attitude
of
asking
questions,
people
being
very
generous
with
their
time
and
like
giving
responses
like
really
cool
collaborations
being
set
up,
that's
yeah,
that's
like
those
are
characteristics.
I've
noticed
since
the
beginning
and
things
I
still
see.
So
I
it's
kind
of.
I
About
with
being
like
letting
the
community
kind
of
decide,
the
direction
of
the
product
is
something
you
touched
on
so
yeah.
I
just
wanted
to
hear
your
thoughts.
B
Yeah
one
of
the
things
that
I
think
is
is
a
struggle
for
me
and
I
don't
know
that
there
is
like
a
solution
for
it.
It's
just
like
how
numbers
work
that
you
know
you
2018.
The
community
was
still
reasonably
small,
but
let
me
tell
you
in
2016:
it
was
like
five
of
us
sitting
around
the
kitchen
table
or
something
no,
not
that
extreme.
B
But,
like
you
know
you,
you
kind
of
knew
everybody
and-
and
that
was
still
appreciably
true
in
2017,
and
one
of
the
things
that
that
helps
with
is
that
new
members-
that's
not
fair,
not
new
year
members,
like
literally
everybody
when
you
go
to
ask
a
question
in
a
public
space,
you
always
have
this
second
thread
going
on
in
your
brain.
That
says,
like
are
people
going
to
think
that
this
is
a
really
dumb
question
and.
B
I
literally
have
that
thread
in
my
brain
today.
I
I
just
asked
something
about
some
new
functionality
that
snowflake
launched,
and
I
was
just
like.
I
wonder
if
everybody's
already
talked
about
this-
and
I
just
missed
it-
I
wonder
if
everyone's
gonna
be
like
this
guy's
really
out
of
touch
at
this
point
and
and
when
you
have
a
small
community,
you
don't
worry
about
that
as
much,
because
you
like
literally
know
everybody
and
you
no
one's
gonna,
be
no
they're.
Just
gonna
judge
you
anyway.
B
B
I
don't
know
what
we're
at
seven
people,
something
like.
Seven
people
on
the
community
team
and
we're
gonna
grow
that
team
pretty
aggressively
over
the
coming
year.
One
of
the
ways
that
we're
really
trying
to
solve
this
numbers
problem
is
is
kind
of
localizing
conversations
into
micro
communities,
as
opposed
to
just
like
this
is
the
beginner's
channel.
B
It
has
25
000
people
in
it,
and
you
know
if
you
ask
a
question
you
have
to
that
question
is
going
to
be
seen
by
25
000
people
instead
to
like
you
know
whether
it's
by
location
or
interest
area,
like
you
know,
there's
a
I
forget
the
channel
name,
but
there's
a
data
team
leaders
channel
now
for
people
to
talk
about
their
struggles.
B
There
there's
database
communities,
there's
there's
all
these
like
different
micro
communities
and
and
we
we
do
a
lot
of
analytics
to
figure
out
which
of
these
micro
communities
are
healthy
or
which
ones
need
some
love.
But
I
think
that,
as
of
right
now,
the
dbt
community
really
is-
or
at
least
the
slack
part
of
the
dbt
community
is
more
appropriately
seen,
as
you
know,
on
any
given
week
between
15
and
25
different
micro
communities.
B
Anyway,
all
this
is
to
say
that,
like
I,
I
think
that
the
problem
with
scaling
community
is
very
hard.
It's
not
impossible,
and
I
think
that
we
have
to.
B
We
almost
have
to
think
about
it
like
an
anthropological
problem,
more
so
than
like
software,
where,
like
literally,
we
have
conversations
about
like
how
do
humans
productively
interact
with
one
another,
and
how
can
we
like
nudge,
the
interactions
in
in
that
direction?.
I
Thanks
for
that,
brian,
I
think
you're
up
next.
A
Yeah,
I
do
have
another
question
around.
You
know
we
are
devops
company
and
can
you
tell
us
a
little
bit
of
your
thoughts
on
the
data
space
and
either
how
it
has
or
it
needs
to
embrace
devops
mindset
and
the
benefits
of
that.
B
I
can
it
is
it's
funny
when
we
said
that
in
2016,
so
I
published
this
blog
post
that
essentially
became
the
vision
for
the
company
and,
and
you
know,
for
what
we're
trying
to
do
in
the
world.
I
published
this
in
2016
and
the
central
thesis
was
data
data
people
needed
to
work
more
like
software
engineers
and
a
big
part
of
that,
although
I
don't
think
I
used
the
term
devops
in
this
post.
A
big
part
of
that
was
essentially,
you
know,
devops
related
workflows,.
B
It's
it's
almost
accepted
wisdom
that,
like
that
is
how
things
should
work,
but
the
the
word
in
that
sentence
that's
doing
the
most
work
is.
Should
I
think
that
people
see
that,
as
almost
an
idealistic
like
okay
dbt
has
created,
has
like
carved
out
a
part
of
the
data
workflow
that
does
len.
You
know
now.
B
We've
we've
brought
that
the
data
transformation
workflow
into
the
devops
world,
but
you
know
if
you
look
at
ingestion,
if
you
look
at
any
of
the
well
almost
any
of
the
different
analytics
layers,
these
parts
of
the
the
workflow
are
not
do
not
really
allow
themselves
to
be
thought
of
in
the
devops
workflow
really
at
all,
and
so
I
think
that
while
practitioners
would
like
to
push
that
further
most
of
our
tools
don't
really
encourage
that
today.
B
I
I
don't
want
to
like
call
out
anybody
specifically
but
like,
if
you
think
of
the
bi
tools
that
you
spend
time
with
like
most
of
them
do
not
support
you.
Do
not
think
of
them.
As
like
code,
first
experiences
or
submitting
to
pull
request,
merge,
request
processes
very
effectively.
You
don't
think
of
you
know
any
of
these
like
classically
devops
things,
and
but
there
are
companies
being
started
today
to
like
rebuild
those
layers
as
well
in
a
in
a
more
devops
workflow.
B
So
I'm
hopeful,
like
my
here's,
here's
the
way
that
I'd
like
to
see
this
work.
I
I'd
love
for
there
to
be.
B
You
know
a
single
repo
or
or
as
a
series
of
repos
that
like
span
the
entire
analytic
journey
at
a
given
company,
and
if
you
want
to
release
a
new
feature
and
that
new
feature
involves
data
ingestion
and
transformation
and
analysis
that,
like
that,
is
one
or
a
set
of
merge
requests,
and
you
approve
those
you
merge
those
and,
like
literally
all
of
it
just
happens
now
we
we
can't,
we
don't
have
the
ability
to
control
that
ourselves
but,
like
I
would
love
to
to
play
nicely
in
in
that
environment.
B
A
A
B
I
think
that's
totally
true
keeping
this
short.
I
think
that
my
sense
is
that
data
leaders
at
large
organizations
want
that
to
be
true,
but
they
don't
see
a
path
between
like
where
they
are
now
and
that
actually
becoming
true
and
partially
that's
a
like
training
problem,
because
maybe
data
transformation
is
done
by
more
technical
folks
and
data
analysis
is
done
by
classically,
less
technical
folks,
and
so
maybe
there's
a
skills
gap
there
to
actually
do
things
more
in
a
data
ops
fashion.
B
A
Tristan,
thank
you
so
much
for
spending
some
time
with
us
today.
I
know
it
was
a
big
commitment
and
it
was
very
appreciated.
I
hope
that
you,
you
found
it,
as
you
know,
stimulating
as
we
all
did
so.
Thank
you.