►
Description
Demo a application that use both Bacalhau and FEVM
This talk will focus on plans for expanding bacalhau in the web3 space, including ongoing work on the Lilypad Project, and the enabling user apps it can provide such as Waterlily.ai
A
Good
morning,
friends
and
fielders
Falcon
Builders
fielders
yeah
right,
I'm,
Ali
I'm,
a
devrel
engineer
at
the
farcoin
foundation
and
I've
also
been
working
with
bakriao.
Our
team
for
the
last
six
months
as
well
and
I'm
happy
to
be
joining
you
all
in
Boston
today,
and
also
joining
fvm
Dev
Advocate
Zach
ayesh
for
a
Dua
on
compute
at
falcoin
that
I
like
to
call
the
Ali
de
Zak
or
A
to
Z
of
compute
yeah
I
won't
give
up
my
day
job,
so
a
quick
agenda
for
everyone
I'm
going
to
get
into
the.
A
Why
we're
doing
this?
You
know.
Why
is
this
important
for
filecoin?
Why
is
this
important
technology
in
general,
then
we're
going
to
have
an
introduction
to
Fem
from
Zach,
as
well
as
a
demo
of
deploying
a
market
active
smart
contract
and
then
I'm
going
to
get
into
introducing
bakliel
and
I'm
going
to
do
a
quick
demo
of
calling
backward
Yale
from
a
smart
contract
with
a
project
we
have
called
lilypad.
A
A
If
you
think
about
the
printing
press
washing
machines,
things
like
that
that
really
changed
kind
of
how
we
operate
as
society
and,
of
course,
more
recent
ones
like
the
internet.
So
it's
also
why
it's
so
essential
that
we're
creating
a
future
for
the
underlying
technology
that
we
use
daily
to
provide
this
open
and
fair
web
that
protects
our
human
rights
and
helps
us
move
forward
in
a
more
egalitarian
way.
A
So,
as
you
would
all
know
being
that
you're
attending
a
conference
on
competing
over
it,
data
is
an
absolutely
essential
part
of
our
lives
so
enabling
these
distributed
storage
and
compute
mechanisms
for
data
is
part
of
the
fundamental
Mission
we're
working
on
at
filecoin
and
protocol
Labs.
So
building
that
Tech
to
enable
open,
scalable
and
distributed
Data
Systems
and
allowing
for
open
Innovation
to
provide
the
base
infrastructure
needed
for
this
open
web
Vision.
A
So
in
practice
that
means
it's
not
only
essential
that
we
create
the
distributed
storage
and
retrieval
Network
that
enables
that
data
sovereignty
which
we're
doing
with
filecoin
but
as
well
as
and
providing
the
content
verification
system
with
ipfs.
But
it
also
means
enabling
computation
of
data
in
a
way
that
aligns
with
this
open
web
Mission.
So
it's
summed
up
pretty
much
by
this
slide
on
the
four
pillars
of
a
decentralized
internet,
as
we
see
it
at
farcoin
and
we
think
about
some
of
the
data
stored
in
the
farcoin
ecosystem.
A
We're
making
use
of
this
data
in
collaborative
ways,
and
it's
also
never
been
more
relevant
or
important
than
today,
I
think
with
AI
applications
exploding
into
mainstream
use
so
with
programmable
applications
enabled
by
fvm
and
with
verifiable
our
peer-to-peer
data
computation,
we're
kind
of
enabling
tokenized
data,
marketplaces,
open
source,
machine
learning,
algorithms,
potentially
and
other
things
with
when
many
more
applications,
so
I'm
going
to
hand
over
to
Zach.
Here,
though,
to
chat
about
the
filecoin
virtual
machine
and
fvm
Zach.
B
Hey
everyone:
my
name
is
Zach
ayish
I
am
a
developer
Advocate
with
the
filecoin
virtual
machine
team,
so
I
basically
help
developers
build
on
top
of
the
filecoin
virtual
machine.
You
know:
I've
been
in
the
web
3
space
for
about
six
years
now
and
I'm
really
fascinated
and
interested
in
decentralized
infrastructure.
B
So
if
you're
not
familiar
the
file
coin,
virtual
machine
brings
programmability
to
the
file
coin
blockchain.
So
if
you're
familiar
with
a
blockchain
like
ethereum
or
polygon,
they
have
this
concept
of
smart
contracts,
so
they
run
distributed
code
and
filecoin
originally
was
built
for
the
ground
up
for
decentralized
storage,
so
connecting
to
storage
providers.
Storing
data
and
having
the
storage
providers
prove
that
they
were
storing
this
data
onto
the
file
corn
blockchain,
but
there
was
no
virtual
machine
on
top
of
it.
So
that's
that's
what
you
would
do
with
filecoin.
B
Now
we
have
the
filecoin
virtual
machine
and
we'll
get
into
the
architecture
a
little
bit
it's
written
in
wasm,
but
it's
actually
built
in
a
way
to
actually
allow
any
other
kind
of
runtime
to
be
built
on
top
of
it
right.
So
it
kind
of
works
like
a
hypervisor
or
virtual
machine
on
your
computer,
where,
like
maybe
you're,
running
a
Windows
machine
and
you
open
up
a
virtual
machine
for
Linux
or
Mac
OS,
and
so
it's
built
in
wasm,
but
the
first
runtime
we
actually
have
running
on.
B
There
is
the
F
evm
or
the
file
coin,
ethereum
virtual
machine
right,
and
this
allows
us
to
take
advantage
of
all
existing
tooling
around
solidity
evm
byte
code.
You
know
Viper
hard
hat
remix.
All
of
these
awesome
tools,
a
lot
of
smart
contract
developers
are
already
used
to
using
and
keep
your
eye
out
in
the
future
for
other
run
times
down
the
line.
B
B
So
you're
going
to
hear
these
terms
like
we're
at
the
Cod
Summit
compute
over
data
Summit,
but
you'll
also
hear
me
talk
about
compute
over
state
right
and
this
is
actually
the
level
of
the
fvm.
And
what
do
we
mean
by
compute
overstate
right?
So
in
this
diagram
here
you
see
the
l0.
This
is
kind
of
what
filecoin
was
made
for
is
decentralized,
Storage,
archival
storage
and
then
at
this
kind
of
L1
level
we
have
the
smart
contracts.
B
B
B
See
if
the
internet
will
load
so
I'm
going
to
yep,
Phil
Fox
right
here
so
Phil
Fox
is
a
blockchain
Explorer
if
you're
familiar
with
blockchain
explorers,
it's
like
etherscan
for
ethereum
and
we
are
on
the
hyperspace
testnet.
So
that's
one
of
the
primary
test
nets
for
filecoin.
It
is
being
deprecated
in
a
bit
for
a
more
stable
test.
Net
called
calibrationnet.
So
I
should
tell
you
if
you're
doing
your
work
on
hyperspace
prepare
for
calibration
net
we're
moving
all
the
docs
over
there.
B
It'll
have
all
the
same
features:
it'll
be
a
pretty
easy
switch.
But
what
we're
looking
at
here
is
you
know
if
you
go
here
in
this
tab,
blockchain
you'll
see
a
tab
called
deal
list.
I'll
click
on
that
and
you'll
see
a
bunch
of
these
storage
deals.
These
are
the
metadata.
This
is
the
state
that
we're
talking
about
with
compute
over
state
right.
Each
of
these
represents
data
stored
in
a
storage
provider
somewhere
off
of
the
file
called
blockchain,
but
they're,
proving
that
they
have
the
data
in
this
deal
stored
using
zero
knowledge
proofs
right.
B
B
This
is
to
prove
for
the
client
that
they
can
pay
for
the
service
that
they're
about
to
get
from
the
storage
providers
to
store
their
data,
and
the
provider
has
to
put
up
collateral
to
prove
that
they
have
stake
in
the
game
right.
So
they
have
to
make
these
zero
knowledge
proofs
to
prove
that
they
have
the
data
stored
and
if
they
can't
make
those
proofs,
if
they
can't
produce
them,
they
get
slashed
right.
B
B
You
can
see
in
this
diagram
that
you
have
to
prepare
your
files
into
a
special
format,
called
a
dart
dot
car
file
and
you
know,
produces
ipld,
dag
or
kind
of
looks
basically
a
Merkle
tree
right
if
you're
familiar
with
like
how
ethereum
does
State
and
at
the
top
you
get
this
com,
P
or
pcid.
That
kind
of
represents
all
of
this
data.
B
Right
so
originally,
this
is
the
flow
for
how
you
had
to
make
deals
on
filecoin.
It
was
a
a
long
process
with
many
steps,
especially
for
the
user.
You
basically
had
to
find
a
storage
provider
figure
out
what
their
going
rates
were.
If
you
didn't
agree
with
that,
maybe
negotiate
with
that
storage
provider,
so
you'd
have
to
go
in,
like
maybe
a
Marketplace
or
something.
B
Then
you
know
once
you
kind
of
had
an
agreement
with
them
that
you
think
would
work.
You
would
send
them
a
signed
deal
Proposal
with
your
key.
It
would
be
signed,
they
would
review
it
and
if
they
agree
with
it,
you
would
give
them
the
data
to
store
and
they
would
have
to
make
a
call
a
function
in
the
file
coin,
apis
to
publish
that
storage
deal
and
actually
start
activating
the
deal
on
chain
right,
so
kind
of
a
cumbersome
process.
B
But
went
to
a
slide
a
little
too
early
there,
but
you'll
see
going
forward.
The
fem
has
allowed
us
to
make
this
a
lot
easier
and
programmatic
right,
because
that's
the
whole
point
of
the
fem.
We
want
to
make
this
automatic
programmatic
and
make
it
easier
for
users
to
program
logic
around
making
deals
on
the
file
coin.
Blockchain.
B
And
so
this
that's
one
use
case.
This
is
a
slide
of
all
the
different
kinds
of
use
cases
that
we
imagine
being
used
on
the
fvm
and,
as
you
see
right
there
in
the
center,
we
have
decentralized,
verifiable
computation,
which
is
where
baculao
comes
in
and
all
the
other
awesome
Cod
projects
that
are
around
here.
B
Other
projects,
because
we'll
go
into
a
little
bit
more
Ali
will
go
into
more
detail
about
the
verifiable
computation
in
a
bit,
but
another
big
one.
We're
seeing
is
data
Dows,
so
data
dowels?
Are
this
concept
of
you
know
a
dow,
a
decentralized
autonomous
organization,
and
instead
of
having
a
single
user
in
charge
of
what
data
is
stored
on
the
file
coin?
B
All
right
so
I'm
going
to
make
fvm
like
a
little
real
for
y'all
right.
So,
like
we're
talking
about
all
this,
how
does
it
work
like?
Let's
deploy
a
smart
contract
I'm
going
to
deploy
a
simple,
smart
contract
that
reads:
data
from
the
market?
Api
right,
so
all
of
the
deals
on
file
coin
can
be
retrieved
through
traditionally
What's
called
the
market
API.
Now,
there's
a
what
we
call
an
actor
or
smart
contract,
an
inbuilt
actor
that
solidity.
Smart
contracts
can
access
to
get
this
data
right.
B
So
this
isn't
something
you
can
do
on
ethereum.
We
have
very
specialized
solidity.
Libraries
called
filecoin.soul.
These
were
built
by
our
friends
at
a
team
called
zondax
did
an
excellent
job
and
basically
these
contracts,
interface
in
between
the
inbuilt
actors,
the
traditional
apis
on
filecoin
and
the
solidity
world.
So
you
can
bring
all
that
data
in
to
your
solidity.
Smart
contracts.
B
So
I'll
just
give
you
a
little
preview
of
what's
going
on
in
this
contract,
so
we
have
this
hard
hat
kit,
if
you're
more
attuned
to
developing
locally
and
you're
already
familiar
with
the
solidity
world.
This
has
a
bunch
of
pre-built
contracts,
including
the
one
I'm
about
to
show
you
in
it
and
a
bunch
of
configurations
for
automatically
working
with
filecoin,
but
essentially
you'll
see
here.
This
contract
inherits
these
zondax
file
coin
solidity
contracts.
Right
specifically
the
market
API
I
was
just
talking
about
and
all
this
contract
does
is.
B
It
has
a
bunch
of
State
variables.
These
are
the
deal,
information
and
a
bunch
of
functions
to
call
the
Market
API
and
store
them
in
state.
So
you
can
write
logic
around
them.
You'll
actually
see
here.
There's
another
more
complex
contract
called
the
the
deal
rewarder
contract
right
here,
and
this
actually
allows
like
people
to
put
up
a
bounty
and
they
can
retrieve.
They
can
call
a
function
to
retrieve
a
deal
if
it's
the
deal
for
the
Bounty
that
they
propose
like
hey.
B
We
want
this
data
with
this
PC
ID,
it's
this
size
and
that
deal
is
made
on
chain.
This
contract
will
pull
that
data,
see
that
the
data
that's
stored
on
chain
matches
what
the
Bounty
is
and
pay
out.
Whoever
made
that
deal
right,
so
you
can
put
up
a
bounty
for
any
kind
of
data.
Just
to
give
you
an
idea
of
the
kind
of
logic
you
can
build
around.
These
storage
deals
on
the
fvm.
B
All
right
so
remix,
if
you're
not
familiar,
is
an
online
IDE,
integrated
development
environment
for
basically
quickly
testing
and
developing
smart
contracts
is
typically
used
when
you're
just
trying
to
mess
around
fiddle
around
with
smart
contracts
and
get
started,
and
smart
contract
development,
usually
when
you're
really
building
a
project,
you'll
eventually
move
on
to
something
like
hard
hat
or
Foundry.
You
know
those
local
development
environments.
B
So
metamask
is
the
one
of
the
biggest
ethereum
wallets
and
since
the
fevm
is
completely
evm
compatible,
it
totally
works
with
filecoin.
You
can
go
to
chainlist.org
to
add
all
of
the
file
coin
networks
into
metamask
and
yeah,
so
Now
remix
knows
to
go
through
metamask
and
my
meta
mask
is
set
to
the
hyperspace
test
net
future
to
be
calibration
net.
B
Confirm
and
now
you
know,
the
file
corn
block
times
are
about
30
seconds
on
average,
but
file
coin
has
this
thing
called
deferred
execution,
so
even
after
your
transaction
is
included
in
the
block,
it
takes
about
another
30
seconds.
Another
block
for
that
transaction
to
be
actually
executed
on
the
state
right.
So
we're
not
actually
going
to
sit
here
and
like
wait
for
this
contractor
to
be
deployed.
I
put
in
my
notes,
a
pre-deployed
contract.
B
B
B
But
if
you
wanted
to
you
just
have
to
put
in
the
deal
ID
1760
click
store.
All
you
know,
metamask
will
come
up
and
you
know
store
that
information.
We
don't
actually
have
to
do
that
right
now,
since
I've
already
stored
it.
But
if
we
go
through
here
and
look
at
like
the
deal
commitment,
the
deal
commitment
is
the
PC
ID
there
that
you
see
in
the
bytes
and
the
size
of
the
data,
so
pcids
I
think
are
base64.
B
You'll,
see
them
encoded
in
that
ethereum
uses
bytes
hex
hex
bytes
as
their
bytes
encoding.
So
it's
translated
right
there
into
hex,
but
it
is
that
same
deal.
Pc
ID,
you
have
the
deal
label,
which
would
be
the
CID
that
you
would
see
on
something
like
ipfs,
because
they
are
different
due
to
how
the
file
is
compressed
into
filecoin,
and
you
know
the
deal
term
how
long
the
deal
is
meant
to
start
and
end
and
yeah.
B
So
you
can
write
all
kinds
of
cool
logic
around
that
to
enable
those
applications-
and
you
know
that
is
one
thing.
So
we
have
also
the
Falcon
client
contract,
which
again
is
there
to
act
as
a
deal
making
contract.
If
you
want
to
actually
make
the
deals,
so
I've
shown
you
how
to
retrieve
deal
information,
but
that
could
be
it's
a
whole
nother
like
45
to
A
Min
hour
demo.
So
we're
not
going
to
go
through
that
today.
B
So
some
references
for
the
resources
for
the
fvm,
if
you're
interested
in
building
or
don't
already
know
about
these,
this
QR
code
here
will
take
you
to
a
link
tree
that
my
colleagues
Sarah
put
together,
has
a
bunch
of
awesome
resources.
Hackathon
cheat
sheets.
Basically
everything
you
need
to
go
to.
B
So
yeah,
that
is
just
the
introduction
to
fvm
and
fevm.
If
you
have
any
questions,
please
feel
free
to
reach
out
to
me.
My
team
is
also
over
there
of
other
developer
Advocates
Sarah,
Matt
and
longfei
so
feel
free
to
ask
them
any
questions
too,
and
with
that
I'll
let
Ali
come
on
stage
and
demonstrate
how
Cod
and
bahi
allow
integrate
with
the
fvm.
So
thank
you.
Everyone,
foreign.
A
So,
just
briefly,
because
I
think
folks
here
have
probably
already
heard
about
it,
bacquiao's
a
peer-to-peer
network
of
notes
that
enable
users
to
run
Docker
containers
or
webassembly
images
as
tasks
against
data
that
is
stored
in
ipfs
and
in
future,
hopefully
filecoin
as
well,
and
provides
this
platform
for
public,
transparent
and
verifiable
computation.
So
it
was
originally
conceived
to
bring
useful
compute
resources
to
this
data
stored
on
ipfs
and
the
filecoin
network.
Bringing
these
same
benefits
of
like
open
collaboration
on
data
sets
to
generic
compute
tasks.
A
So,
let's
talk
about
now
how
bacliao
and
fem
would
be
work
can
work
together.
So
we
saw
this
slide
a
little
earlier.
Zach
put
it
up
so
just
taking
it
back
to
the
architecture
view.
As
we
said,
farcoin
can
kind
of
be
considered
the
layer
zero.
This
it's
really
robust
internet
scale,
storage,
Network
and
then
when
then,
we've
layered
over
fem,
bringing
like
these
layer.
One
capabilities
to
the
file
coin
stack
and
making
it
really
easy
to
do
that.
A
Compute
over
State
and
bakuyao
is
kind
of
can
bring
the
additional
capability
for
off-chain
verifiable
compute
tasks
on
large
data
to
the
network.
So
this
kind
of
makes
for
some
really
powerful
tooling
to
enable
this
new
open
web.
That
I
was
talking
about
earlier
and
I'll
come
back
to
lilypad
in
a
second.
So
quick
recap
on
the
differences
here.
A
Basically,
so
why
are
we
doing
this
well
outside
of
the
overall
filecoin
Vision
I
introduced
earlier
I?
Think
Peter
Wang
from
Anaconda
said
it
best
yesterday.
He
was
one
of
the
keynote
speakers
yesterday.
If
you
didn't
see
that
talk,
I
really
recommend
looking
it
up
if
you're
at
all
interested
in
AI,
because
he
did
a
really
great
job
of
explaining.
A
You
know
why
we
really
need
to
be
invested
as
a
community
in
building
these
really
Community
orientated
systems
for
data
rights
in
AI
so
definitely
check
that
one
out,
if
you
didn't
and
as
Patrick
Collins,
who
probably
a
lot
of
people
in
web
3,
would
have
heard
of
so
simply
put
it
as
well.
We
need
this
credible,
neutral
technology
of
web3
and
there's
many
more
voices
seeing
the
need
for
blockchain
and
AI
to
work
together
for
a
better
outcome
and
side.
A
Point
I,
don't
know
if
anyone's
read
this
recent
leaked
Google
Document,
who
knows
if
it
really
is
that
it
doesn't
have
a
moat.
So
you
know
it's
basically
advocating
that
open
source
is
winning
kind
of
the
AI
race
and
is
doing
it
better
than
these
commercial
entities
which
to
me
was
really
great
to
hear.
A
To
be
honest,
so
the
team
has
built
this
proof
of
concept
project
lily
pad
for
making
it
easy
to
use
back
Cloud
jobs
from
fem
smart
contract
using
these
contracts,
basically,
as
a
base
layer,
lilypad
acts
as
a
bridge
for
calling
any
back.
We
are
job
and
receiving
those
results
from
bakayal
which
come
back
as
ipfs
cids
directly
from
a
solidity
smart
contract,
so
I
use
a
contract
that
you
would
write
in
solidity
implements
the
lilypad
caller
interface
and
to
call
a
job.
A
They
just
call
a
function
with
the
spec
for
the
job,
so
like
a
Docker
image,
spec
to
call
that
sorry,
they
call
a
function
in
the
lilypad
events
contract
so
that
will
trigger
our
Lily
Pad
Bridge
demon
to
then
go
ahead
and
send
this
job
off
to
balkia
for
computation,
and
then
it
Returns
the
results
back
as
soon
as
it's
finished.
So
once
the
job
is
complete,
it
goes
back
to
the
original
user
contract
that
was
built
in
solidity.
A
So
you'll
get
the
results
directly
back
to
your
state,
so
one
of
the
example
user
contracts
in
the
lilypad
repo.
Is
this
stable
diffusion
example
which
I'll
kind
of
go
through?
Now,
though,
as
an
aside,
if
you're
interested
in
actually
building
the
AI
script,
the
python
script
yourself
or
this
text
to
image
stable
diffusion
example,
there's
a
full
walkthrough
on
that
on
the
backyard
docs
as
well.
So
if
that's
a
side,
you're
interested
in
feel
free
to
go
and
check
that
out
and
because
we
all
know
how
well
live
coding
goes
on
stage.
A
I've
got
a
video
instead,
so
to
call
hopefully
this
loads,
yay
internet,
all
right,
so
I've,
just
loaded
this
up
on
remix,
which
Zach
was
showing
you
before
so
to
call
the
back
of
our
job
on
fem,
you
create
a
smart
contract
that
inflates
that
Lily
Pad
caller
interface
and
that
interacts
with
our
Lily
Pad
this.
This
is
the
back
end,
just
showing
you
kind
of
the
bridge
Daemon
where
the
backlar
job
was
running.
A
Sorry,
that
was
pretty
quick,
but
it
will
show
you
like
there's
nothing
there,
then
all
of
a
sudden
event
comes
through
bakada,
runs
that
job
and
then
sends
it
back
as
a
CID,
and
you
know,
there's
my
stable
diffusion
result
right
there.
So
this
is
the
overall
folder
it
comes
back
in
and
then
you
can
look
up
the
kind
of
output
of
that
job,
which
in
this
case
was
an
image.
So
you
can
make
all
sorts
of
cool
things
with
this.
A
I
initially
made
this
nft
up,
but
I've
gone
further
than
that
since
then,
and
we've
actually
made
a
use
a
user-facing
app
called
water
lily.
You
know
we've
taken
this
kind
of
further.
It
does
use
lily
pad
to
do
this.
So
little
waterlily.ai
is
kind
of
this
first
of
its
kind,
user-facing
app
that's
aiming
to
provide
a
more
Equitable
new
paradigm
in
generative
AI,
so
we're
kind
of
leveraging
the
benefits
of
having
a
blockchain
like
fvm
to
enable
automation
of
payments
to
artists
on
this
chain.
A
So
what
happens
is
as
a
user
you
would
go
in.
You
would
type
in
kind
of
you
know
whatever
prompt
you
want
for
your
stable
diffusion
image.
Then
you'll
choose
one
of
the
artists
that
has
uploaded
their
work
to
the
chain
and
had
a
unique
model
trained
on
their
artwork
and
you
pay
a
small
fee
in
order
to
generate
you'll,
get
four
Images
back
on
water
lily
and
then
that
fee
outside
of
you
know,
the
fees
for
running
on
the
Chain
background
doesn't
take.
A
A
So
yeah
we've
taken
the
concept
A
little
bit
further,
and
this
is
kind
of
one
of
the
things
one
of
the
cool
things
you
could
do
with
like
blockchain
and
AI
with
lilypad.
We
think
we've
got
a
heap
more
use
cases
in
the
pipeline
that
we're
thinking
about
as
well.
Things
like
enabling
storage
deals
or
easier
storage
deals
like
yeah
I
was
looking
for
that
little
way
to
go
yeah
so
things
like
that
or
integrating
with
some
of
our
Dev,
tooling
and
they've
leaned
kind
of
an
oracle-like
service.
A
It
wouldn't
exactly
be
an
oracle
but
an
oracle-like
service,
a
lot
more
Ai
and
ml
use
cases
and
then,
like
kind
of
wider
web
3
use
cases.
Of
course,
the
roadmap.
You
know
we
want
Community
input
on
this
is
how
we
make
a
better
our
future
by
you
know
getting
what
people
actually
want
to
build
out
from
them
so
feel
free
to.
A
Let
us
know
if
you
have
any
ideas
on
what
you'd
want
to
build
or
use,
but
we've
got
some
use
cases
in
the
pipeline
and
like
kind
of
a
road
map
of
where
we're
going
to
go
from
here.
If
you
do
want
to
hear
more,
get
in
touch
with
usual
things
and
we're
having
some
really
cool
discussions
over
in
the
battle
slack
about
where
you
know,
decentralized,
Ai
and
ml
could
go
as
well.
A
A
Yeah,
so
currently
the
public
background
network
is
free.
We
don't
have
a
pricing
model
on
it.
So
we're
running
this
as
a
like
public
good
at
the
moment
in
future.
We're
developing
the
game
theory
around
that
to
enable
you
know
other
nodes
joining
the
network
like
having
untrusted
nodes,
be
on
the
network
and
get
paid
for
jobs
that
get
run
also,
you
know,
obviously,
then
we'll
have
a
pricing
model.
I.
A
Think
probably
the
best
way
to
go
is
everything's
in
wasm
code,
then
new
jobs
will
have
to
be
wasn't
code,
so
you
can
calculate
that
kind
of
gas
that
it
would
cost
to
run
that
job,
but
yeah,
that's
that's
kind
of
where
we're
heading
in
the
future
and
what
we're
planning
out
right
now.
So
that's
definitely
what
we
we
want
to
do
with
the
network.
A
Yeah,
so
it's
an
edge
compute
Network,
so
we
currently
can't
fetch
from
filecoin,
though
that's
part
of
our
roadmap
in
the
future.
We
want
to
investigate
how
easy
that
is
to
do
and
try
and
enable
that
in
one
way
or
the
other,
whether
that's
just
through
current
tools
we
already
have
or
whether
that's
natively,
which
you
know
would
be
amazing,
but
at
the
moment
everything's
comes
out
of
ipfs
for
the
web
free
slide.
A
Otherwise
you
can
actually
run
on
AWS
buckets
and
things
like
that,
if
you're
not
that
you
don't
really
care
about
where
it
is
but
yeah,
definitely
it
runs
on
ipfs
as
well.
At
the
moment,
egress
I
couldn't
tell
you,
but
probably
one
of
our
Engineers
can
who
I
can
find
for
you
later
or
there's
one
right.
There
actually.