Snowflake Summit 2023 Keynote Summary | Snowflake Summit

0:00Hey it's Tim here, I've just spent four

0:03hours going through the Snowflake Summit

0:04keynote.

0:04I did it twice just so I could do the

0:06summary. So we're going to cover the whole

0:08entire keynote,

0:08break down what happened, what was

0:10announced, as ever let's get stuck in. Now

0:13for those of you

0:14thinking hey what happened Tim, you're

0:15Tableau Tim. Well I'm starting to cover

0:17Snowflake and

0:18this is going to be the first video really

0:20sort of pushing into that area. I've done

0:22Snowflake videos

0:23in the past but I think Snowflake Summit

0:25was a good opportunity for me to really

0:27start to get to

0:27grips with what's happening in the Snow

0:29flake space but also keep in touch and help

0:31update you

0:32about what's going on and I think

0:34conferences are always a good opportunity

0:35to do that,

0:36that's when lots of people are paying

0:37attention. So it absolutely makes sense, I

0:39'm not going to

0:40stop making Tableau content, you'll notice

0:42that I haven't stopped with that, I'm still

0:44covering 23.2

0:45but today we're covering Snowflake so let's

0:48break down the Snowflake Summit as it

0:50happened last week.

0:51The keynote started with Frank essentially

0:54doing a pretty big flex. Now there was a

0:57previous

0:57session the day before with Jensen the

1:00NVIDIA CEO and that is a pretty big flex.

1:03The main reason is

1:04is that NVIDIA hardware is pretty much hard

1:07to come by and Snowflake being able to

1:09secure a

1:10partnership with NVIDIA, essentially the

1:12leading hardware provider for AI and

1:14machine learning

1:15technologies is kind of a big deal because

1:17it really means that that partnership has

1:20some

1:20strength. NVIDIA even were on stage you

1:22know pulling out a quote like. The big deal

1:25here Sarah

1:26is that we're going to bring the world's

1:28best compute engine to the world's most

1:32valuable data.

1:33We're coming to the world's most valuable

1:36data platform, that was a really sort of

1:38important

1:39quote to kind of you know pull out of that

1:41conference and so for Frank to start off

1:44the

1:44keynote and say hey we just had NVIDIA on

1:46stage with us and that's how serious we are

1:48about the

1:49space was a pretty interesting thing but he

1:51had to immediately frame it and sort of

1:53remind people

1:54that hey AI is a buzzword and yeah you

1:55might be playing drinking games every time

1:57I mention AI

1:58but actually before AI comes a data

2:01strategy that was a pretty nice lead into

2:03essentially what the

2:04Snowflake platform has and the next sort of

2:0715 minutes we're just uh Frank essentially

2:10outlining

2:11what Snowflake is in case you've missed any

2:13of the announcements over the last few

2:14years

2:15highlighting all the different workloads

2:16and exactly how it works. So if we just go

2:18through

2:18that I've got my notes here um we started

2:20off by really positioning this concept of

2:23the data cloud

2:24this is the visualization you see on the

2:25Snowflake homepage but he had the

2:27visualization up on stage

2:28and he kind of depicted it as like a solar

2:31system with the data cloud as the sun that

2:33was the central

2:34sort of core of their entire platform and

2:37as we sort of got into this description

2:41Frank was really

2:41highlighting the USB of Snowflake this

2:44ability for companies to come onto the

2:46platform start with

2:47their basic use cases and then grow out

2:49expand do collaboration and he focused in

2:51on a company

2:52called Fiserv and he kind of showed how

2:54their node in this data cloud was sort of

2:56growing out and how

2:57they were touching base with lots of

2:58different businesses and also helping come

3:00up with new

3:01solutions inside of the platform and all

3:03this is happening within Snowflake this was

3:04sort of the

3:05whole point of this narrative look at what

3:07you can achieve if you just commit to the

3:09Snowflake

3:10platform. Now he then went on to break down

3:12the workloads very explicitly and I was

3:15actually quite

3:16surprised how many workloads they are when

3:17I started using Snowflake I'm actually a

3:19Snowflake

3:20customer first and then I'm a consultant

3:22and now I'm sort of a content creator in

3:24many ways so I've

3:25got three hats on I always sort of just saw

3:27Snowflake as just a database and I sort of

3:30knew

3:30about the peripheral sort of features and

3:32capabilities but only recently have I

3:33started

3:34to realize wow Snowflake really has a

3:36really broad remit when it comes to the

3:38data space it's not

3:39just a database there's lots of different

3:41sort of capabilities happening on the

3:43platform and so if I

3:44just go through them I've got a list here

3:46because I can't remember all of these off

3:47the top of my

3:48head. Frank started with the concept of

3:50warehouses so of course warehouses and data

3:52lakes are things

3:53that we've had for many many years I think

3:55even in the Nvidia session and the Nvidia

3:57CEO said that

3:58like the concept of a database had been

4:00sitting on a shelf for years until someone

4:03came and made

4:03a product out of it now interestingly that

4:05was Oracle and interestingly Snowflake

4:07founders came

4:08from Oracle so they're all about that

4:10journey but nonetheless we went from

4:12warehouses to data lakes

4:13and then they talked about the product

4:15called Unistore essentially their transact

4:17ional layer

4:18and that is a pretty powerful capability

4:20because later on in the keynote they do

4:22talk about some

4:23new capabilities that allows you to sort of

4:25use Unistore without compromising some of

4:27that

4:28capability that you get with a typical data

4:30warehouse and the data lake and then Frank

4:34sort of then zeroed in on the collaboration

4:35aspects he just really wants to highlight

4:37this

4:37I think this comes up as a running theme

4:39throughout the conference collaboration is

4:42something the Snowflake platform can do and

4:44it can do it well and it can do it securely

4:46and he

4:46throughout a stat here saying 70% of

4:49customers using the platform are using the

4:51collaboration

4:52features which is quite a big number that

4:54said 70% came up multiple times which made

4:56me think that

4:57it was like a generalized statistic the

4:59exact percent is probably around 70% but

5:02every time

5:02you see a number multiple times you think

5:04every time everything is 70% they all can't

5:07be exactly

5:0870% it doesn't make sense so that was

5:11interesting and then he brought up cyber

5:13security the idea

5:14that look if your data is here and

5:15everything's running here why do you have

5:17to go somewhere else

5:18to run cyber security workloads again that

5:20's another workload we've got data

5:22engineering

5:23apparently 40% of the compute capacity on

5:26Snowflake is taken up by data engineering

5:28capabilities and that makes sense data

5:31engineering is a pretty you know popular

5:33thing and you pretty

5:34much find all the data engineering

5:37platforms do sort of dovetail with Snow

5:38flake you look at dbt

5:40all their training everything dovetails

5:42with Snowflake and so that's a pretty cool

5:44capability

5:45and then you have two more ai and machine

5:47learning um again the 70% stat came up here

5:50but I didn't

5:50quite sort of understand what was being

5:53said 70% of customers have used it or 70%

5:56are using it

5:56every single day I didn't sort of get that

5:58breakdown and I I've rewatched and I can't

6:00really sort of get the clarity on that but

6:02this is one of those things you just have

6:03to sort of

6:04go at face value and trust that what they

6:05're saying is actually the case and lastly

6:08you had

6:09examples of applications apps essentially

6:11being used on the platform so this whole

6:14sort of uh

6:15let's say universe of workloads was then

6:17sort of um you know sort of brought up on

6:20screen and then

6:21around it came the governance layer this

6:23this idea that all of this happens inside

6:25of the Snowflake

6:26platform with governance and security at

6:28its core security is just a staple uh part

6:30of every

6:31database it's not really unique to Snow

6:32flake but the fact that governance is sort

6:34of baked into

6:35all of these features was a pretty big

6:37selling point and then there was a small

6:39sort of finesse

6:40of just uh you know really branding the

6:42three core parts of Snowpark which are ai

6:46machine learning

6:46and data engineering essentially those

6:48three things come together to form the

6:50product the

6:51sort of group of products called Snowpark

6:53that was again put on screen and then the

6:55very last

6:56piece was to say to essentially everyone

6:58and say look all of these workloads speak

7:01to different

7:01kinds of people whether it's developers

7:03users the business users all of those are

7:06covered by the

7:06Snowflake platform and Snowflake you know

7:08pretty much welcomes all of those so that

7:10was a really

7:11nice i think way of um you know really

7:13encapsulating what the Snowflake product is

7:16how it works but more importantly just

7:18reminding everyone in case you're new if in

7:20case you're an

7:20investor in case you don't know um that

7:23Snowflake really is turning into a platform

7:25with lots of

7:26capabilities with this sort of focus on

7:29your data and they want you to take part in

7:31that data cloud

7:32and they don't see the data cloud as just

7:34your data they see the data cloud as your

7:36data

7:37intermingled with other people's data all

7:39done in a way that you can trust and secure

7:41so that was a

7:42i think a really nice uh sort of framing of

7:44the product and then we finished off with

7:46essentially three core areas that Frank

7:48sort of wanted to highlight and it was all

7:50under the no

7:51limits banner so no limits for data

7:53essentially being able to give you what you

7:55want when it comes

7:56to managing your data without compromises

7:58no limits with applications being able to

8:00really

8:01look at the app setup as the equivalent of

8:04an apple app store which is an example

8:06Frank gave

8:07on stage and then the last one is no limits

8:09on workloads we want you to be able to do

8:11whatever

8:11work you want in Snowflake in the platform

8:14without having to worry about all the other

8:16problems and

8:17so um the these were sort of teeing up the

8:20core sort of uh components of the rest of

8:23the keynote

8:24these are what to watch out for and these

8:25are sort of the three things i like it

8:27because three things

8:28always works nicely um so that was a pretty

8:30nice uh setup and so that was that was

8:32pretty much it

8:33and i thought that was a really strong

8:36confident like um you know depiction of the

8:39product a lot

8:40of times you go to a conference and you

8:42find companies really trying not to sort of

8:45narrow

8:45their product down to say this is what they

8:47do because they want to speak to everyone

8:49because

8:49there's always someone in the room who's

8:51passionate about one area but isn't

8:52passionate about the rest

8:54so they kind of um avoid being really

8:56direct with you and saying this is what we

8:58're good at and this

8:59is what we do that wasn't the case here in

9:01my opinion as an outsider so um i thought

9:03that was

9:04pretty good the next stage we saw was

9:06actually Benoit coming up and talking about

9:09the recent

9:09acquisition of NIV uh NIV i can never say

9:12it correctly but anyway let me know how it

9:14's supposed

9:15to be said in the comments but it was uh it

9:17's a sort of an interesting one because the

9:19acquisition

9:20is still quite young it's literally only

9:22weeks uh since the acquisition closed and

9:24what was

9:24interesting here is it felt like a

9:26credentialization exercise it was very much

9:29a look here's the uh

9:30founder of NIV um he's joining snowflake

9:32these are his credentials this is what he's

9:34been working on

9:35and by the way we've been working on ml

9:37already um and we think this is how it's

9:39going to work

9:40and Benoit was very much sort of pinning

9:42this down as something that he believes in

9:45something

9:45his team is working on and something NIV

9:47can actually help a lot with and so um

9:50there was

9:50also some subtle tones to suggest that um N

9:52IV have some other capabilities beyond their

9:55sort of basic

9:56products that actually will help snowflakes

9:59internal uses of things like large language

10:01models and ai and machine learning so it's

10:04it's as much as a a purchase of the

10:06application as it is a

10:07purchase of the mindsets and the minds that

10:10can come into the snowflake universe and

10:12help them

10:13build new products so that was really good

10:14to see and it was good to see the founder

10:16actually

10:16leading that charge whenever something new

10:18like that comes up um yes the ceo is a good

10:20sort of

10:21person to sort of announce these things but

10:23actually it meant more to have the founder

10:25there

10:26to really speak to the technical and the

10:28directional sort of um you know push for

10:30snowflakes so i thought that was good and

10:32that was sort of its own little standout

10:35section it

10:35kind of felt out of place but they had to

10:37do it they had to do it early and to have

10:39it second

10:39after the ceo before the features was

10:41really sort of a statement of intent in

10:44many ways so

10:45um that was sort of the the first two bits

10:47of the keynote the next bit of the keynote

10:50was very much

10:50the product showcase and the product

10:52showcase was really really nice it was

10:54broken down into three

10:56specific areas and i'll just read them out

10:58here we had the single platform section we

11:00had the

11:00distributed deploy and monetize section and

11:04then lastly we had programmer programmer i

11:06can't say

11:07this programmability i can never say this

11:10properly anyway nonetheless um it was a

11:14really interesting

11:14sort of breakdown and so uh let's step into

11:16each one of those into a little bit more

11:18detail

11:19and uh let's really list out the features

11:21because there were lots there was a lot

11:22covered in this

11:23keynote so you'd be mistaken to think much

11:25wasn't actually announced but if you sort

11:27of go through

11:27as a list like i've written this whole

11:30thing out it's a long list of stuff so um

11:32let's go into this

11:33and really break it down so the single

11:35platform second was very much about making

11:38sure that you

11:38understand that look if you have this

11:40workload if you have this challenge inside

11:43of snowflake

11:43you shouldn't have to go elsewhere to get

11:46it done and so they really showcased a

11:48bunch of features

11:49and a bunch of demos all working within

11:51snowflake this was sort of the emphasis if

11:53you want to do

11:54this you can do it in snowflake if you want

11:55to do that you can do it in snowflake to

11:57the point where

11:58you know you could think of another

11:59solution and again the answer should be you

12:01can do it in

12:02snowflake and so the first example was this

12:05a really uh i think impressive demo of

12:08using

12:08ai machine learning to essentially process

12:11documents and they had uh as i put it up on

12:13screen they had a demo on the right hand

12:15side of a document it was a mixture of

12:17handwritten

12:18text and typed text on the left you had

12:20essentially a way to query that document

12:23using machine learning

12:24and ai with a simultaneous interface that

12:26also allows you to train and improve the

12:28quality of

12:29the model as you're querying the data so it

12:31's essentially a really smart way of pairing

12:34training

12:35with um you know out of the box capability

12:37i can't remember the exact phrase that um

12:40the presenter

12:40used but there's a term for that and

12:42essentially means that out of the box you

12:43get some immediate

12:44value but also it means once you start to

12:46train it once you start to improve the

12:48quality of the data

12:50the model is going to improve over time and

12:52as long as you sort of keep sort of feeding

12:54this

12:55this engine a little bit more context you

12:56're always then going to be able to have a

12:58more

12:59reliable sort of use case and it gave

13:01confidence scores for its ability to tag

13:03and you know

13:04recognize all the bits of information that

13:06were in each of these documents but i

13:07thought it was

13:08a really good um demo to me it's like a it

13:10's like a it's it's one of the few demos i

13:12've seen recently

13:13that i think was just out of the box

13:15convincing because you could see the

13:17document you could see

13:18what the software was doing the interface

13:20wasn't sort of ridiculously complex and

13:22then to finish

13:23it off we kind of went off to another part

13:25of the demo where we were just able to go

13:27ahead create

13:28the table query bring it into snowflake and

13:31then use those columns that you'd queried

13:34as just a

13:34simple query in a sql statement and so that

13:36was nice to see this end to end vision of

13:39here's a

13:39new feature here's how it works it's really

13:41simple it's in a familiar place it's in the

13:43platform and

13:44by the way everything the platform can do

13:46you can do it all here already as well so

13:48that was a that

13:49was really good and we then had a little

13:51bit of a focus on platform privacy

13:53essentially all the

13:54little features that snowflake has

13:56implemented to allow you to really control

13:59privacy and i think

14:00this is an important sort of placement

14:01because before you talk a lot about ai you

14:03really have

14:04to give people the confidence that you're

14:07able to help them handle the privacy and

14:09essentially

14:10compliance requirements that are required

14:12in some markets as well so that was a

14:14really good flex

14:15i'll put a sort of graphic up and screen of

14:16all the different areas that they covered

14:18but um it's

14:19pretty straightforward i think every

14:20company has a tech like this internally but

14:22it's just great to

14:23see snowflake doing it you know in a very

14:25specific way and calling it out i think it

14:26was actually

14:27mentioned on stage as well and then you had

14:30the focus on performance now this is a

14:33weird one

14:33because performance is one of those thank

14:35less things when you make something go

14:37faster when

14:37you make it fail less everyone goes like

14:39yeah well done you're doing your job and so

14:41alice and lee

14:42actually came up and was talking about the

14:43performance improvements she mentions a ton

14:45of

14:45them and she quite rightly highlighted that

14:47look um if these are working then you

14:49should be noticing

14:50faster performance and you should also be

14:52noticing things failing less often and they

14:54don't deserve

14:54a gold star for that because you know you

14:56're you're their customer they're trying to

14:58make you

14:59happy but they did also highlight that look

15:01when they make and deploy these benefits

15:03you do get a

15:03benefit as well because if things are

15:05running faster and they're failing less

15:07often then

15:07because snowflake is a usage based pricing

15:10setup you're also saving money in in other

15:12words so i

15:13think that was a really nice sort of uh

15:15point then they then went on to talk about

15:17the performance

15:18index essentially a way of really

15:19understanding what's improving inside of

15:21snowflake so you could

15:22get a better sort of feel for that and then

15:24another thing they touched on was this idea

15:27of

15:28being able to essentially control how

15:30utilization is being used through budgets

15:33so essentially being

15:34able to set a budget for a particular

15:36individual or pacific roles and then not

15:38only getting alerts

15:40when they cross over that budget but also

15:42getting alerts when they're on the trend to

15:44go over that

15:44target and that's really useful because it

15:47allows you to sort of give various teams or

15:49various

15:49people um essentially a a a walled setup

15:52that kind of gives them a bit of room to

15:54grow it can

15:55encourage a conversation if they're using

15:57up their space too quickly and i think the

15:59other nice thing

16:00about this is it maybe gives confidence for

16:02people to start thinking about buying

16:04consumption up front

16:05because they know they can budget the

16:06consumption and they can move that sort of

16:08consumption around

16:09if maybe someone's under or over and you

16:11can kind of create a small economy within a

16:13within an

16:13organization to sort of handle this as well

16:16and the last one that to be honest i don't

16:17quite fully

16:18appreciate the benefit of this yet but this

16:20warehouse utilization metric now i've

16:22always

16:23thought warehouse utilization is something

16:24that you're really only focused on when you

16:26're part

16:27of the optimization discussion and so it'd

16:29be interesting to see exactly what the

16:31warehouse

16:31utilization measures um and sort of how

16:34that's broken down so it's one of those

16:36things you have

16:36to see and i think you have to be really

16:38detailed about snowflake to really

16:40understand what it's

16:41doing as well so that was really nice and

16:43then to sort of cover all of these features

16:46off they

16:46brought on a customer called fidelity and

16:48fidelity just talked about their experience

16:50of building

16:51on the snowflake platform it's one of these

16:53things that it's just nice to have a

16:54customer speaking to

16:55exactly what you've mentioned already and

16:57making sure that you know everything you're

16:59saying can

17:00be grounded in real world use cases and

17:02customers who are fundamentally happy and

17:04want to grow their

17:05usage of the platform so that was the end

17:07of the single platform section there i've

17:09just sort of

17:10gone through some of the announcements

17:11there but i think it was a really strong

17:13section

17:13it showed people things that they're

17:15familiar with in within snowflake but also

17:17showed improvements

17:18and no doubt if you go to other sessions

17:20you'll probably get a lot more detail about

17:23this

17:23especially the what's new session i think

17:25is being called as the session to go to if

17:27you want to know

17:28how these things all work in detail so

17:30definitely go check that out but that was

17:32the end of the

17:33single platform section next we're going to

17:35go to the next section which is about

17:36deploy distributed

17:38and monetize essentially really inviting

17:40companies to come and build on the snow

17:42flake platform let's

17:43check that out so for this section snow

17:46flake really sort of changed the narrative

17:48and it's

17:49a very subtle change i mean i only noticed

17:52this on my second watch in essence at this

17:54point snowflake

17:56had been talking almost entirely to the

17:58analysts to the people using the platform

18:00and then they

18:01suddenly changed to talking to what i would

18:03call um entrepreneurs within the community

18:06to essentially

18:07say look so these are all the features now

18:10we're enabling you to come build something

18:13deploy in a

18:14secure and governed way and then monetize

18:17it so they're essentially talking to

18:19developers and

18:20opportunist people who want to come in and

18:22build on the snowflake platforms and create

18:25a business

18:25and they actually brought in a customer

18:27called cyber i think it was cyber and they

18:30talked about

18:31what they're doing how they've set things

18:33up and what they're hoping to do the apps

18:35they're hoping

18:35to to solve within the ecosystem but they

18:38're not just coming to the marketplace with

18:40apps they're

18:41coming with marketplace apps and data sort

18:43of coupled together to give you a really

18:45unique

18:46capability with their service offerings so

18:48um it was very much i sort of think of this

18:51as the kind

18:51of pitch that apple would have to do not to

18:53us as consumers but to developers to come

18:56and build apps

18:56that can then be monetized and so to speak

18:58to that they showed a whole bunch of

19:00different

19:01capabilities now i'm not a developer so to

19:03be honest i don't get the significance of

19:06half of these things so if you do know the

19:07significance if some of these are going to

19:09change your life let me know in the

19:11comments i'd really love to know more about

19:13it but

19:13they talked about the app framework

19:15essentially a framework for building apps i

19:17'll come back to

19:17apps in a second because there's one thing

19:19i just fundamentally don't disagree don't

19:21agree with

19:22sorry and i'll come to it in a bit in a

19:23short minute but let me just go through

19:25this list

19:25and they talk about app framework being in

19:27public preview so essentially a bunch of

19:29apps already

19:29available and now it's in public preview

19:31lots more people can actually start

19:33covering it and using it

19:35so that's going to be something interesting

19:37to watch over the next six months and apps

19:39are not

19:39easy to build so we're going to probably

19:41spend the next year seeing how apps are

19:44being built well how

19:45they're being built badly and sort of best

19:47practice sort of coalescing around that now

19:49alongside app

19:50frameworks they also talked about something

19:52called event billing which allows you to

19:53essentially

19:54charge users based on events rather than

19:57just the classic usage model that had

19:59previously been there

20:00so you could specify whether the app

20:03charges on a specific type of query that's

20:05run and then

20:05essentially from there you're then able to

20:08you know really actually monetize your

20:10application

20:11in a way that probably wasn't possible or

20:13feasible usable before because again you

20:16know pricing is

20:16really difficult and if you've got more

20:18flexibility to really change the way you

20:20price you can actually

20:22end up giving the customer better value but

20:25also coalesce around you know exactly where

20:27the value

20:28in the id you have is the other thing was

20:30this setup called capacity drawdown which

20:34essentially

20:34allows customers to buy capacity up front

20:37and then have apps draw down from that

20:39capacity including

20:40typical things like compute and storage so

20:43actually apps i think the feature here is

20:45that

20:46apps have the security to know that

20:47customers who do have capacity drawdown can

20:50just go ahead

20:50and start using their apps and those apps

20:52will be getting the benefit from the

20:54previous existing

20:55commitments that customers have already

20:57made rather than having to go out and sign

21:00new

21:00statements of work or new contracts or

21:02whatever just to use an app which i think

21:04was probably

21:04the issue before you know companies when

21:06they buy software they tend to be very pres

21:08criptive

21:08about what they're getting so they might

21:11get a sort of commitment-based pricing

21:13setup but that

21:14didn't account for apps or that didn't

21:15account for a specific capability so that

21:17was good to see

21:18now this brings me to an area that i was

21:20super confused by which which was they kept

21:23on showing

21:23what to me felt like dashboards but they

21:25kept on calling them apps and i just never

21:27really got any

21:28clarity as to what snowflake actually

21:31considers as an app now in some cases i saw

21:33really good

21:34examples of that because we had uh you know

21:37interfaces with forms and you know the kind

21:39of things you'd expect in an app and a way

21:41a user interface a way for someone to

21:43inject something

21:43and get something back out of value but in

21:46other cases we just saw essentially dash

21:48boards and charts

21:49so i kind of i kind of i think the jury's

21:51out on me at least to really understand

21:53what exactly is

21:54an app in the snowflake system because

21:56there's a lot of mention of applications

21:58and apps but i

21:59think i found it really hard to pass out

22:01the different tiers or the different sort

22:03of

22:04complexities or how simple or advanced they

22:06have to be and and then the final thing was

22:09we got a

22:09demo of course and the demo was really sort

22:11of just a way of piecing it all together

22:14and i think

22:14it was it was one of those uh really nice

22:17sort of demos because it really showed

22:20everything working

22:22inside of the platform again this was just

22:23a very sort of consistent theme throughout

22:25the whole

22:26event and we had the feature the example

22:28and then you had the demo just showing you

22:31it working

22:32almost live the problem with a lot of snow

22:35flake features is they go into preview for a

22:37long period

22:38of time then they come into general

22:40availability and the reason they're in

22:41preview as frank said

22:42himself was they like to test these things

22:44and make sure they work so when they were

22:46in preview

22:47you have to be given access to go and see

22:49them and you know use them but once they're

22:51in general

22:52availability then they're officially

22:53supported and they're public and everyone

22:55can start using them as

22:56long as you're in the right pricing tier to

22:59enable that specific feature okay that's

23:02the end of that

23:02section let's move on to the final section

23:05which was actually about program programm

23:07ability but

23:08really was talking to developers let's get

23:11stuck in so this last bit was really quite

23:13interesting

23:14it was mostly focused on developers

23:15essentially telling developers hey you've

23:17been asking us for

23:18a bunch of things we've listened here they

23:20all are and so we had uh capability showing

23:23us how

23:24streaming pipelines worked i think this is

23:26a really nice demo because it showed the

23:27evolution

23:28of how sort of snowflake is built on this

23:30capability over time with snowpipe as a

23:32starting

23:32point but now really going through the rest

23:35of the workflow and allowing you to build

23:37on really much

23:38more complex pipelines all the way through

23:41the entire setup and they had this sort of

23:43very brief

23:44demo of a no code sql essentially just

23:46being able to type what you wanted to

23:48happen and have

23:49snowflake come up with a sql for you and i

23:52found that demo interesting i sort of crit

23:54iqued that a

23:55little bit in my full breakdown of the

23:57whole entire summit and i kind of said well

23:59that's

24:00great but you still need someone to check

24:01it and to check it you need to know your

24:03stuff so if you

24:03know your stuff you're probably not going

24:05to use ai and so you just write the thing

24:06yourself it's

24:07kind of these one of these snake eating its

24:09own tail sort of visualization but

24:11nonetheless it was

24:12nice to see that working and a certain

24:14other competitor company has also announced

24:17something

24:18very very similar so it's always

24:19interesting how these things go down but no

24:22less i'm not surprised

24:23that snowflake announced it at their event

24:25too um the other thing was snow park

24:27container services

24:29and again for this one i just don't know

24:31enough about snowflake implementation and

24:33containers

24:34to understand where the value is now

24:36containers typically in technology allow

24:39you to run a system

24:40in a in a let's just think about it as a

24:42walled garden but a very sort of opaque

24:44wall garden so

24:45you can set up an application inside of a

24:48container and as far as the application is

24:51concerned all it sees is the operating

24:53system it doesn't know about any other

24:55application

24:56that's sitting next to it below whatever

24:58and so in in in a traditional setup you can

25:01put all of these

25:02apps into containers you put them into snow

25:04flake they're all running on snowflake but

25:06they don't

25:07have any awareness of each other which is

25:09good for security it's good for a little

25:10bit of governance

25:11and those apps don't need to know about

25:13other applications so why would they you

25:15know why would

25:15they need to see them and also from an

25:17upgrade and stability perspective if

25:20something goes wrong with

25:20one of the containers you can just stand up

25:22another one very quickly and you don't

25:24disrupt all the

25:24other systems whereas if you take your

25:26computer as an example if something goes

25:28wrong and your

25:28computer crashes as soon as you hit restart

25:30everything running on computer has to also

25:33restart

25:33that's not how containers work containers

25:35allow you to just be modular and sort of

25:37pick and choose

25:38how things are working so um i don't fully

25:40understand snowflake's implementation in

25:42snow park

25:43remember snow park is the uh you know uh

25:46data engineering ai an application sort of

25:49group of

25:50products if that makes sense so i think i

25:52have to really come back and unpick the

25:55value of the

25:55container in this specific setup but

25:57nonetheless it was great to see this as a

25:59demo we did actually

26:00see it working and then the demo went on to

26:03cover llms in a little bit more detail llms

26:07is just

26:07something that i think every company feels

26:09they have to demo on stage but nonetheless

26:12it was good

26:12to see it was good to see it working and

26:15then at the very end of this session and

26:17they essentially

26:18brought out 10 demos from uh well i think

26:21it was was it 10 demos yeah it was 10 demos

26:24from uh

26:25different companies who've all built uh

26:27apps inside of this uh new setup around

26:30containers

26:30so i think this is a really really nice

26:34really nice demo really nice setup because

26:36um you know

26:37it was kind of like a big thing to ask to

26:39have that many companies and there weren't

26:41small

26:41companies we're talking about companies

26:43such as altrix nvidia all of these

26:45companies building

26:46capabilities um for the specific one

26:48feature allowing you to do things you

26:50wouldn't think

26:51would only run in snowflake but now here

26:53they are turning up with their capabilities

26:56letting you do

26:57exactly that so i thought it was a pretty

26:59nice uh way to end the keynote leave people

27:01excited and

27:02infused to go off and talk to lots of the

27:04customers and partners as well and so that

27:07was it that's the

27:08whole keynote pretty much we blitzed

27:10through the whole entire thing um i i

27:12really enjoy the

27:13keynote my first snowflake summit and i

27:15only watched it virtually i didn't actually

27:17go um

27:17and i i found it really interesting i found

27:20it um and i compared it during my full

27:22breakdown to

27:23uh the tableau keynote and i said snowflake

27:26is obviously in a very different place it

27:28knows what

27:28its product is it knows where it's heading

27:31and all it has to do is please its existing

27:33consumers

27:34whereas tablo on the other hand when i went

27:35through that keynote i talked a lot about

27:37how it's

27:38wanting to now pivot to a group of users it

27:40's never had to serve before and so it got

27:43very

27:43different vibes as a result of that and i'm

27:45pretty excited about the direction of snow

27:47flake but um

27:48there were a few things i didn't see there

27:50was a call out for governance a lot

27:52throughout the whole

27:53entire keynote but what we didn't see was

27:55that i can explicit sort of bunch of

27:57features and by that

27:59i mean uh governance seemed to be sort of

28:01peppered in when we were going through the

28:03demo for um uh

28:05distributed deploy and monetize there was

28:07this sort of mention of observability but

28:09that's

28:09observability it's not quite the same as

28:11governance and so um maybe that's sort of

28:14my uh naivety to

28:16the product i haven't been exposed enough

28:18to know how snowflake communicates

28:19governance but um that

28:21just me to me was just a section that sort

28:23of didn't jump out even though frank

28:24mentioned it

28:25at the start i didn't feel like it jumped

28:28out from this whole uh keynote um the other

28:30thing i think

28:31was uh snowflake didn't really if you're

28:35new to databases if you're a company just

28:38bought snowflake

28:39and you went to this um summit or you saw

28:41this keynote you might feel a little bit

28:44overwhelmed

28:44because essentially snowflake might have

28:46shown you a bunch of things you've not even

28:48started to think

28:49about and um it's one of those uh sort of

28:51interesting decisions where uh in this sort

28:54of play

28:55to to show where you're heading to show the

28:57innovation you can potentially marginalize

28:59your

29:00user base and aren't going to ever use it

29:01for more than just the simple thing of a

29:03data warehouse or

29:04a data lake because that's really all some

29:06people still want right and they showed

29:08that sort of

29:08massive diagram and you had all these

29:10companies around the outside those are the

29:12companies i'm

29:13talking about the companies that don't want

29:15to collaborate can't collaborate people

29:16like me i'm

29:17a one-person band i'm a content creator i

29:19use snowflake because it's a pretty

29:21reliable database

29:22for piping my youtube data from five tran

29:24into a database and it's cost effective i

29:27can actually

29:28afford to pay for snowflake as a content

29:30creator doesn't by the way doesn't make a

29:33profit on making

29:34content so um it's really super interesting

29:38uh to sort of uh see how snowflake speaks

29:41to that

29:41customer by chasing all this innovation you

29:44could end up sort of marginalizing those

29:46smaller

29:46organizations who maybe are two or three

29:48people um maybe get a lot of value out of

29:51the platform

29:51but are never going to be running snowpark

29:53are never going to be running these ai

29:55machine learning

29:56setups they're not capital one they're not

29:58uh you know all these companies that snow

30:01flake had on

30:01stage and so that was another thing i

30:03thought of you know if you're a small

30:05customer what's snowflake

30:07doing to help you what are the soft skills

30:09that you know they can set up in the table

30:11au world i

30:12think of things on the tableau blueprint

30:14although that's not perfect i think snow

30:15flake needs an

30:16equivalent sort of here's how to deploy

30:18snowflake kind of gun i know these things

30:20exist they're

30:21white papers videos guides session

30:23conference talks all of this stuff but what

30:25i really mean

30:26is a feature a feature that really helps

30:28you walk through these steps and helps you

30:30navigate sort

30:31of the soft skills but anyway those are my

30:33thoughts i hope you've enjoyed this

30:36breakdown

30:36um if you think i've missed anything let me

30:38know in the comments it's the first time i

30:40've done a

30:40breakdown of summit so go easy on me if

30:42there's things i can improve on next time i

30:44'm always open

30:45to feedback but yeah i'm pretty excited to

30:47start covering snowflake um i'm going to be

30:50pushing on

30:50with this at least once a week to try and

30:52uh really bring up a bunch of new content

30:55on this

30:55space and lastly i think the the other

30:58thing i sort of just want to touch on is

31:00and there's

31:01going to be lots more content at this

31:02conference that i'll probably end up

31:04covering over the next

31:05two weeks um so uh it's not all going to

31:07come at once like it does with tableau i

31:09think it's going

31:09to take me more time to sort of go through

31:11it but i'll slowly maybe over the next

31:13month bring out

31:14all of the sort of breakdowns of the

31:16conference sessions that i found

31:17interesting and if you want

31:19to see those let me know in the comments

31:20below if there's any particular sessions

31:22you'd love to see

31:22covered let me also know those in the

31:24comments below thanks for watching and i'll

31:26catch you in

31:27the next one

31:29you

31:39[ Silence ]