Meet Enso Analytics, a flow based data prep tool. With Ned Harding & Sylwia Brodacka

0:00Ned, Sylvia, welcome.

0:01Welcome to the channel.

0:03Pleasure to be here.

0:05Right.

0:08Yeah, thank you.

0:09Thank you for thank you for agreeing to come on the channel actually.

0:15um people who are building products in the analytics space and um I think you're a really really good fit in in this sort of realm and so

0:23We're going to have plenty of time to get into Enso a little bit more in this video.

0:29But maybe I'll give you the floor, maybe just to introduce yourselves, who you are, and then maybe after that what Enso is.

0:37Of course.

0:38So let me start and then I will pass uh to Nat.

0:41So my name is Sylvia, I am a CEO of Anso.

0:44My background, because this is leading me to where I am right now, is

0:49From the very beginning in data analytics, it was not data analytics per se, it was material science, computational physics

0:57stuff like this so I was always dealing with the data but another part to that was helping other people to analyze data so I was starting as early as the uni

1:08uh with writing Python scripts or Jupyter notebooks notebooks to my friends who were analyzing the data and didn't have the tools actually uh to do that correctly and were blocked by it uh because there was just a

1:22standard set of things that they could do.

1:25And that has started my career with actually building the tools for them.

1:29And it is not different than what we are doing.

1:32now here because it is all about enabling people who don't have the technical task stack uh to

1:41Analyze the data, get the insights and uh use it as much as possible and get as much as possible out of it.

1:50And with that path I met at some point, Ned, who is building it longer than me, definitely, and we started working together.

1:59Amazing.

2:00Have you seen that?

2:02Uh

2:03I got my start as a hacker programmer really young.

2:08I started programming and I was eight years old, which

2:11for my age was uh exceptionally early.

2:15Um originally on a deck PDP eleven.

2:18So um

2:20Uh I doubt any of uh your listeners even have any clue what that kind of computer is, but I have no clue.

2:29It's an old mini computer.

2:30We my school had a timeshare access to it.

2:33The computer actually was three towns over

2:36Uh and we just had terminals.

2:39Uh and I just loved programming.

2:42It was the only thing I was good at.

2:43I wasn't good at school.

2:44I wasn't I was great at math, but other than math, I I

2:47really didn't do well in school.

2:49I was dyslexic and stuff.

2:50And uh but I loved programming and I I at some point said I'm gonna have a career as a

2:57uh programmer and product designer and and I said to myself, I will do anything but database work.

3:07And what have I done?

3:08Nothing but database work.

3:13That's incredible.

3:15Anyway, so yeah, sometime uh uh when I I had some experience, done a bunch of consulting and some some software jobs, I decided I needed to start my own company and I uh

3:28uh met a few people, um Dean and Libby who we started Alteryx together uh back in nineteen ninety seven.

3:37And I of course did lots of database work and then uh had no idea what analytics was, had no idea what data

3:46quality cleansing.

3:48I I'd never done any of that kind of stuff, but we did a lot of consulting for uh

3:55um fairly big companies in the the marketing space and they would send us our their data to merge with demographics and experience data and stuff like that.

4:06And

4:07We had nothing but trouble getting their data merged with these other data sources because of course their data quality wasn't good.

4:16Um

4:17Uh and so uh I got into this space out of need, just building tools for my own use and our own use.

4:27And um

4:29uh built what I think was a great tool for one user at a time.

4:35Um uh but really wanted to

4:39Um, think about it over again.

4:41What would I do differently if I was designing a product uh that I was actually designing rather than

4:49uh building just out of necessity.

4:51And so uh started thinking about working as teams and and working together uh rather than the the individualistic

5:02um uh just get a job done that I was doing before.

5:06Um but yeah that's sort of how I came into this space.

5:10It was not on purpose.

5:13But here I am.

5:15It it's actually a common thread with a lot of people I talk to in data.

5:18It's it's data almost feels to be the place that catches us when

5:23Um, the the thing we were trying to do uh doesn't work at all, we start doing what we're doing in our respective fields and actually realize people really value what we do in the data space.

5:33And so

5:34we end up here.

5:35And uh Ned, you touched on something interesting.

5:37You mentioned this concept of a timeshare for a computer.

5:40And I think to a lot of people listening, they won't even know that like, yes

5:44You used to have to I I guess timeshare in this instance is your booking time on a machine to do what you need to do because it was such a scarce resource at the time, right?

5:53I think that's sort of the concept.

5:55Yeah, exactly.

5:56Exactly.

5:57It's wild to most people.

5:58We all have personal computers.

6:01And yeah, we've we've come a long way.

6:03And so if I come back maybe and just dig into this

6:06um sort of focus around Enso.

6:08How how did the Enso team come together?

6:10If we maybe just dig into that.

6:11You uh Sylvia, you met Ned, but I'm assuming it's more than just the two of you, right?

6:16So yeah, how how did Enso fall?

6:18We have started earlier.

6:20Uh I met Nance at some point uh a couple of years ago.

6:25I was chatting with the users who were using actually tools uh from that space, including Vault Riggs.

6:31And one of the users, uh, Rafa Rafael Albert from ASDA, he was like, you should really talk to Ned Harding.

6:39He's a good guy who's doing really great job and like he is super clever that would be a good fit for the conversation.

6:46So he uh in introduced us and we had a couple of conversations at that point

6:52And one of the people we were chatting who was a heavy user of tools like this was also James Tonkerly.

7:00So James became our CTO.

7:03We have uh

7:05Team of really amazing developers from all over the world.

7:09It is not only US, it is also Europe.

7:12So we got the best talent all over the place.

7:18And uh we got also later in the career uh Adam Riley, who you may know as well.

7:25Yes, yeah, who joined us uh on the libraries, so we are building something really special.

7:30and we managed to do to get a really really amazing team.

7:35And it's a pleasure to work with Ned and guys.

7:38And they are really really

7:41Knowing the space and knowing the tools and understanding the users, which is super important.

7:47Yeah, like product intuition comes from this sort of

7:52a lived experience of, you know, what people are uh uh doing with the tools, like the pains they're still trying to solve, because even even the good tools we have today don't solve every problem.

8:02And

8:02Um yeah, I think as technology moves on, we get new opportunities to try and solve those problems.

8:07And I think, yeah, it's very clear that that team you've assembled

8:10I know that I know they've experienced a lot of the challenges of the products of old because I've read many of their blogs.

8:17I've experienced some of their work as well.

8:18You know, Adam Riley with Creep Macros, right?

8:21Like

8:21There's so much there that um it is amazing.

8:24So for you to have that in your team, I think is uh is an incredible testament.

8:27And anyone who's watching this who knows those names will know that okay, they know what they're talking about.

8:32So it's a fantastic team.

8:34Well, and it was a fantastic team before I came on board.

8:38Uh uh Tylvia had a team of developers uh mostly in Poland, but uh

8:45um elsewhere as well, um, who had developed a really cool core of an engine, but they were looking for how to turn that into a product.

8:55Right.

8:56Um so uh a lot of the design philosophy from Enso came from a a very different spot because it came out of this core that was a little bit more programming focused.

9:10Um and then morphed into a product from there.

9:16Uh and it it's really become something very interesting and unique and different.

9:21And

9:22uh approaching the space in a different way than um I think most any other product.

9:27So it it's kind of cool just doing something different.

9:31Um, and then uh I'll just say in terms of James and Adam and and and the rest of the team, uh uh that they're just some of the smartest people around and it's so fun uh working with them and uh

9:45That that's what I love doing is just being around smart people.

9:50Amazing.

9:51Um so I guess maybe this is a good opportunity to ask you both like

9:55If you could describe Enzo to my audience, my audience is um you know predominantly Tableau, but maybe let's broaden that out to say data analysts embedded within the business, embedded within analytics teams, how would you describe Enzo to them?

10:10Did I take it, Sylvia?

10:13Sure.

10:14Uh well I'll start from Tableau because Tableau is your users.

10:20Tableau is a really great way to present data um and to help people uh make decisions with their data, um give give uh reports to executives who can

10:35uh know what salesperson they need to focus on this month because they were really good or really bad or what whatever it is.

10:43The problem with Tableau is it only works well if your data is good.

10:50Um and what happens so often is people come into their tableau journey with data

10:57that has been hand entered and um is of questionable quality.

11:03And Tableau does provide some tools.

11:06for helping you with that, but if you have more than a couple data sources and more than a a couple of

11:15issues with your data, um it it tends to be very clunky uh to get your data

11:23Um, correct.

11:24Uh clean.

11:26Um so a tool like EMSO has a a full suite of all of the data quality and data prep tools.

11:35Um we can take data from most any data source, write it to most any data source, uh get it all clean, do some visualization.

11:45It's not like we don't do visualization, but we're

11:48We're not trying to be tableau.

11:51We're trying to get your data to where it's ready to present to an executive.

12:00uh not do the presentation itself.

12:03Yeah.

12:04I think you know the visual component is actually quite useful when you're doing data prep because it can be a a quick

12:10really good quick way to um I guess get a 10,000 foot view as AWS calls it right um into sort of what what what challenges you're trying to solve and sometimes it can alert you to patterns that aren't so easily recognizable um

12:24uh when you're just looking at you know tables um uh and some of that as well.

12:27So I I totally, totally connect with that.

12:30And I mean the some classic data visualizations like a scatter plot can be so important because

12:36Yeah.

12:36All of your data clusters in in one quadrant and then you have uh ten percent of your records or two percent of your records

12:45out in a completely different quadrant and and uh it does help you narrow in on those records and

12:53And in ENSO, you can uh select those records straight off the visualization and then continue your tabular-based discovery on what's going on there.

13:06And and that's really helpful because those data those records might be like your best performers, uh, they might be your worst performers, or they might just be bad data.

13:15And so seeing it on a chart is great, but then you have to dive back in uh and and know why that data is the way it is.

13:28Uh and that's where hopefully Enzo can can help in a in a very unique way.

13:34Amazing.

13:35Sophia, I think you were you were gonna add something.

13:38Yes, for me what is super important is that the people who need to have this feelless insights can go grab a tool and work with it by themselves.

13:49Or they can say to their data person who is usually preparing the first parts of the job to uh give them initial workflow and then dig into that themselves and adjust

14:01Because in many cases they are trying to explain for the third party what they want to achieve.

14:06It is really hard sometimes to explain if you don't see the data and you cannot explore it.

14:11You cannot play with that

14:13So uh this is the tool that is enabling to work between also levels, between the business and the stakeholders and the data people.

14:24Uh so everyone can achieve together the goal, which is super, super important in my in my view.

14:31Yeah.

14:31Yeah, it's uh it's a pretty lofty goal, it's a pretty lofty challenge.

14:34And I think

14:35Before we we maybe go on to a s uh screen share and um uh talk a bit more about the products, I think i if I step back a little bit, there's this um

14:45There's a lot of talk in the in the current ecosystem around semantic modeling and uh let's say semantics more generally as a concept.

14:54Um and what I wonder is where data prep comes into that.

14:58And I've always felt the view that when I talk to data engineers, they view data prep as um

15:04are somewhat different, I think, to what they do, which which sometimes can be modeling, it can sometimes be warehousing, it can sometimes be, you know, maybe pipelines.

15:14How would you how would you sort of

15:16articulate the difference between, you know, what you're doing with Enzo, the data prep side of things, versus what a data engineer does when they're, you know, talking about semantics as a as a concept.

15:33What I would say data engineers don't want to go to the visual tools.

15:40They want to write their scripts and go forward with forward with that and they are very comfy with their uh Python DVT whatever they have in under their belt.

15:52So that's one side.

15:54And uh with NSA you can achieve very similar goals, uh doing it visually uh and then unifying and uh automating that

16:05over uh the runs and over the workflows.

16:08So from my perspective, uh it's not the difference what you're doing, it's just a different approach.

16:16Because uh the philosophy is very, very, very different for data engineer than for the analyst.

16:23Yeah, exactly.

16:24And so maybe maybe if I extend that thinking

16:28your your approach is maybe closer towards the business analyst versus uh um an approach which is more towards I guess the engineer, right?

16:39Um from a data perspective.

16:42Uh in general, if if you've got engineering involved, you have a very different workflow.

16:50Uh it tends to be

16:53Somebody who understands the data, trying to write a spec, throw it over a fence, some amount of time later they get something back.

17:00It may or may not be right.

17:03Uh if um

17:07If that's what the company does, it's worthwhile, obviously, to to have those kinds of engineers involved.

17:16But the uh the time from uh

17:21seeing the data to discovering uh something you can make a decision on, uh the timeline gets very extended because of this communication gap.

17:32So

17:33Our goal is to not involve a data engineer, is to have the person who understands the data do their own work and uh

17:45Rather than uh communicating back and forth and taking months to get an answer that may or may not be right, they they can do the same thing in hours uh on their own machine.

17:58Um so it's it's just a different way of thinking of it, but the the why is the most important thing and and engineers tend to not understand the why.

18:10I mean uh

18:12I can't speak for all of them because I'm probably an engineer and I hopefully I understand the why, but uh uh it's

18:21And so is for people who understand the why of the data.

18:25They understand what the goals are.

18:28Um and it allows them to uh work together and share their results.

18:35uh with the team uh very seamlessly, very quickly.

18:42Amazing.

18:43Amazing.

18:43Right.

18:43So maybe this is a good opportunity to sort of dive into the product, if that makes sense.

18:50Awesome.

18:51So as you can see, we will start with data prep and blend.

18:56So uh where everything begins, it is as you can see very minimalistic.

19:00It is not having uh

19:03exposed money but with all the tools but it data prep and blend so we need to start with bringing data into the product so we will start with the input

19:13As you can see, we are getting something that we are calling component browser.

19:18So you can search through the components we have suggested ones, but also on the also.

19:25Feel free to ask any questions during that.

19:28Yeah, yeah.

19:30So I will choose the data read and I will bring a file that I have uh on my hard drive actually.

19:38I don't know that you don't see the file browser, but I promise it is what I opened.

19:44Yeah.

19:46So you can see the first component that is bringing the Excel file.

19:51The Excel file has two worksheets, the invoices data and CRM data.

19:58I can open any worksheet by double clicking on that and explore what is in that dataset actually.

20:08So I will

20:10Do the full screen of the visualization so we can see that we have some EIN numbers, uh, invoice men's company name, invoice date.

20:22In the dataset.

20:23Now I will exit full screen, open the second dataset, and see what is in the CRM data.

20:30We have again the EIN, EIN number and the account order.

20:35And what I want to do today is to get the summary by the salesperson that we have in the second data sheet by year and by performance.

20:48But as we can see the EIN is rather nasty, so we have some random symbols here and make this result visualization bigger

20:58So we can we need to clean that and also we didn't parse correctly the invoice state, so we need to fix that as well.

21:08So how to start with that?

21:10I will click the glass button here.

21:13And as you can see, I'm getting again a rec browser, but it is different than the first one.

21:19So it is adjusted to the data that you have uh brought to the product.

21:24uh you can um either go and click through the tools or you can do it differently I can just search.

21:33What I need to do is I need to clean uh the packs uh with the EIM so I can

21:40type clean, get the text plans tool, and now I can set up the configuration.

21:47So I need to choose the column I need to uh clean.

21:51So it's my GIM column.

21:53And what I want to remove.

21:55I can clean multiple columns or a single one.

21:58Also, I can do multiple operations in that.

22:01When I am selecting, it is the order that is preserved while doing that.

22:07Here I need just to remove the symbols.

22:11And I can see the output of the data.

22:14So the data is

22:18I will need to do the same operation on the second dataset.

22:21So what I can do is I will just copy and paste.

22:28No, it's over here and then reconnect the component.

22:32So it is just working the same way.

22:35Uh it's

22:37Exactly the same thing.

22:38Now what I will do is I will clean also the date.

22:44So

22:46We had a very weird format here.

22:50Uh so let me parse it

22:55Uh the column is the invoice date.

23:01I need to add

23:05Format it as a date.

23:07I will do my custom format.

23:10So it will be date space

23:17Comma year and now the date is correctly parsed

23:28I need to get out of this the year because I want to have the summary by the year.

23:34Yeah.

23:35So let's add another component.

23:39The component we need is set.

23:41It is

23:42Something like formula in Altrix.

23:46Yeah.

23:47So uh I will use the expression

23:51And what I want to get is the year.

23:55I'm getting suggestions.

23:57So I need to give uh the name right now, which was

24:02Invert state and close the bracket and I want that to be a Yurcorn

24:11Nice.

24:12So you're like creating a new column, right?

24:14Yes.

24:15Yeah.

24:16At the end of the data set.

24:18Uh I know because I know this data set that I have uh duplicates but I can check that uh out.

24:25So let's check for duplicates.

24:33And it is giving me a warning.

24:35Yes.

24:36It is a warning that I am comparing uh rows that have uh cells that have a floating point numbers

24:42And it might not be the greatest idea to compare compare floating point numbers, but we are fine for.

24:51Get the duplicates over a single column or all of the columns in if I'm not selecting anything, it is all of the columns

24:59Yeah.

25:00So right now I can see that I have already duplicates.

25:03I need to get rid of them.

25:05So I will remove this component but add a distinct component.

25:10That will handle that for me.

25:14And it's the same warning.

25:16I can get rid of it so it will not propagate through the workflow.

25:21Why do you not rank it?

25:23It that would be me.

25:25I feel like I need to add just for people who aren't

25:29programmers why that warning exists.

25:32Uh floating point numbers, um, even if they look the same based on how they were computed,

25:40can be different internal representations.

25:43So if you have the number three that you typed

25:48And then you have a a expression uh of five times point six

25:58uh which should be the same number, it's possible the internal representation will be very slightly different and they will uh not compare identically

26:10So the reason why Sylvia knows that she can uh get rid of the warning here is because all of those numbers have the same

26:18source.

26:19None of them are computed.

26:21But if some of them were read and some of them were computed, uh they could be different in the sixteenth decimal place.

26:29And

26:30wouldn't be found by distinct.

26:32So you'd you'd have to do a round or something like that first.

26:35Right.

26:36Right.

26:37I know I'm getting technical, but I feel like it's weird.

26:40We love the detail.

26:41Yeah, we love the detail.

26:42It's incredible.

26:43Yeah.

26:44Thank you.

26:46I will join both of the data sets together.

26:49So we'll let's go to join.

26:54And now it is asking me to provide the second table, so I will connect them together

27:01There you go.

27:02Okay.

27:02And when choosing the join kind.

27:05In answer we have always one output.

27:08So if you want to have left out a join.

27:11or um full join.

27:14You need to choose the proper join from the menu.

27:17It is not about getting two outputs and merging them them together.

27:21Yeah.

27:22On default it will try to uh join on the first column which would work here.

27:28But what I would do, and it's a good practice, is to say on what you are

27:33joining and it's because later uh with time if you're going through their workflow and looking at that after a month, you know exactly what you have done and you it is very visible step by step what was happening.

27:49So uh we have joined both of the datasets together.

27:54So now we need to crosstab.

27:57Okay

28:04So let's go with the cross top not cross joining but cross stop.

28:11We will need to group by uh account owner bridge imits here.

28:19Uh then we can group uh over multiple columns of course.

28:24Then we need to get the gear.

28:29And have some over um invoice amounts.

28:40And as you can see, we have many leading spaces in here.

28:45And we know you were calculating uh sums over

28:49uh the incomes and uh invoices so what I would do is would be formattic actually uh those columns

29:01And again we can choose one by one, but what we can do also is to choose by type.

29:08Oh

29:09What I want to get is all of the columns that are floating uh that have floats inside.

29:17Then it would be something like this

29:21We can see okay it's working, but I know we have their invoices and values are dollars, so let's put their the dollar sign in front and accept

29:32As it as you can say it is immediate change, don't have to run anything.

29:37It is always reflecting immediately

29:40Uh what was the most important for me was uh who is the best salesperson this year, let's say.

29:49So I can sort it in the visualization, but it is not permanent.

29:52If I want to get that as uh next component, I can add it from the visualization.

30:00And have it for the future, save it, send it over the email.

30:05Okay, you have done it for the one month

30:09And now you need to do it over the second one.

30:12So let's change the the file from June to July.

30:21And as you can see, our person is right now Patricia.

30:26It was Amy before.

30:28So you are getting immediate uh reward

30:32As you are changing the data.

30:36Yeah.

30:36That's amazing.

30:38What we have on the right side is the documentation of the workflow.

30:42So the presentation of the workflow can be written by the user.

30:47It is stored together with the workflow.

30:49You can write us three.

30:51What is the goal of the workflow?

30:54What are the inputs?

30:55What are the outputs?

30:56uh whatever you need to say it is together you can also comment on each component if you want to like I can add here

31:09A comment that this component is sorting four.

31:19It would be saved as well

31:22For the future.

31:24Amazing.

31:25I have to say it's really clean.

31:27I like the um the the thought process is very clear.

31:32Um I like that the

31:35The little windows you get into like preview what you're doing, it it um that is not sort of new to to I guess this kind of tool, but I think what is nice is that

31:44you know, right at the top of each of these little perspectives you've got the the the the function or the calculation or the instruction very clearly set out.

31:51Sometimes that's actually the hardest thing to get to when you're sort of working in these things.

31:55It's it's somewhat removed from the process.

31:58Um

31:58Yeah, and and then the ability to I guess the flexibility in how you work is is quite nice here.

32:05What what I'm what I what I've seen in some of the filter and

32:08sorry, the tool drop downs and the contextual awareness of the tool um means that I think it's a much easier approachable tool because you're only going to see what's actually useful.

32:19um, you know, in context of what you're actually about to do.

32:22So yeah, whenever you've hit that sort of plus icon, I've been sort of looking at like, hey, what what is it suggesting after this step?

32:28And it's it's actually quite no very logical each time.

32:31Um and yeah

32:32As you're showing here, just even the ability to compress bits of the analysis you're not interested in right now is is is very, very nice as well.

32:42I agree.

32:42I think it's very clean.

32:44The other thing I I love is there's no information hiding behind icons.

32:50Uh all all the information is is there, so it makes it um much more straightforward to do the equivalent of a code review if

33:00Yeah.

33:01Uh if I produce this data and I need to show you how I produce the data, it's very clean for you and I to read through all the steps.

33:10But um if we dive under the covers a little bit and show the code view, Sylvia, um this is something that is very unique about ENSEL is uh

33:24Sylvia actually was writing a program.

33:28Oh wow.

33:29It's all visual.

33:30It's

33:31uh all very clean and simple and you don't need to be a programmer, but at the end uh we get a uh um functional um and I mean functional as opposed to functional versus procedural

33:45uh uh programming representation of everything that was up above.

33:50Uh and what's really cool about that is you could make changes here in the code if that's how you think.

33:57Or we can easily um do version control.

34:02Uh and when you do version control, rather than seeing differences in XML or JSON or some

34:09uh abstract representation of what's going on, you can see uh the actual code changes

34:17Um that we're done.

34:19So when you go to collaborate, when you and I are working on solving one problem together, it's really easy for us to see, oh, Ned added this, oh Tim added that.

34:30Right.

34:31Uh because of that code representation.

34:34It's awesome that you're mentioning that.

34:36So here is the data catalog view because we were in the workflow, but there is a part that is a data catalog.

34:42Right.

34:43So you have your space to work with your teammates.

34:47Uh so I can go to

34:52Some workflow that we have in that space.

35:03And if I go to the random workflow, it can see the versions that were stayed.

35:09Yeah.

35:09Who was working on them all the time

35:13Let's go somewhere further.

35:15So we had Steve who is making changes.

35:17And if I go there, I can compare it with our latest version of that workflow.

35:22And see what exactly what was talking about.

35:26That we have quite the new version, the old version on the left, what has changed during the time.

35:32Who changed it?

35:34So it is really easy to to check what was happening.

35:38Also, what I can do is I can duplicate the old workflow or I can restore the previous version.

35:44So if someone and broke the workflow

35:47while you were way because they thought that they they're doing a great thing, you can always restore, go back, and also understand what has changed.

35:58That's really flexible.

35:59And I I guess this this adds to the narrative around um being able to tag I guess tag the metadata from a tool perspective, but then also just observability, being able to see this sort of

36:10fidelity um as people are collaborating.

36:14Um and I guess from a maybe if you lean into the collaboration perspective, um

36:21You know, is the idea that you there's sort of any concept around CICD, so the idea that, you know, when someone's working on the fly, um they have to finish working on it before someone else can, or is it possible for

36:35two people to work side by side?

36:37How how how does that sort of um you know look over your shoulder sort of collaboration work as it's as it as it is through the tool?

36:44Cur currently it is one person working on a workflow at a time.

36:50Uh the the technology

36:53uh under the covers that it's developed with in theory is allowing for a Google Docs style multiple people.

37:03We haven't focused on that right now because that's just not the most important thing today.

37:10So at some point in the future, uh we would hope to have that.

37:14Um as far as CI C D uh again um where we are and where we're going, uh we we don't have

37:28that uh built in um yet, but it is um the the building blocks are all there.

37:35So

37:36One of the aspects Sylvia showed reading from an Excel file, all the data can be stored in the cloud, but the other thing that can be stored in the cloud is what we call a data link.

37:48Uh and the data link is particularly cool in that um uh

37:56I only need to know where is the latest customer file.

38:01I don't need to know that it's in Oracle or in SQL Server or

38:07um Salesforce or I I don't care where the data is.

38:11I just I I read from this data link.

38:14Um

38:16Right.

38:17The future of the data link is that it's going to have uh multiple destinations.

38:24So I can have a debug data link and a release data link, and I can

38:28have the debug data link coming from a CSV file where I have a sample just loaded into the cloud and then the release uh data link going to the main customer file in a remote database somewhere.

38:41But

38:42You don't want to hit that database while you're developing your workflow.

38:47So having that uh um uh lifespan

38:53uh functionality is is definitely coming um when uh hard to say at this point um you can do it manually of course now because you can have

39:05Two separate data links.

39:08But we we will have a more automated system for that.

39:12But it's such a great way for collaboration.

39:15So a team can have a

39:18A folder in in the cloud service of data links, we don't have to host all your data.

39:24That data doesn't even need to be in the cloud.

39:27It can still be behind your firewall if you're

39:30Reading the data locally.

39:33If you're running it in what we call a hybrid environment where you're running the workflows locally, you'd be reading the data link on the cloud.

39:42The data link would have

39:44a connection to say a SQL server that SQL server is behind your firewall.

39:50So in order to utilize the c cloud collaboration, you don't have to move all your data into the cloud if your company's not ready to do that.

40:00Amazing.

40:02It's a good way to virtualize the connection almost, give it like a proxy, but then have that become an asset in itself.

40:07It's a it's a nice concept because

40:10Yeah, even just collaboration, actually that's something that's sometimes the biggest hindrance, just being able to share connections.

40:17And if you can sort of abstract that detail in this way, then that's a really good solution around that problem.

40:24And then not only are you sharing the connection, but you're sharing the metadata.

40:28So that same documentation that Sylvie was showing on the workflow, you have that same documentation panel on the data link.

40:37uh and so you can uh explain uh exactly what's going on in this data source um in in a place where the whole team can get to it.

40:51Yeah.

40:55As the next level, you are creating a data link for your team, but each of the team members have has their own secrets

41:03And you can define the secret for each person in their personal cloud space.

41:09So

41:10This is my secret is generally a a username and password or an OAuth connection or uh it's secure security credentials.

41:20It doesn't have to be security credentials, it can be any secret, but

41:24Uh that's generally what it is.

41:28Exactly.

41:29So you are creating for each user the new sequret?

41:32and that is stored in their personal personal space and then the data link in the common space is using the the secret uh from your personal space

41:45So every time you are running the workflow it is using your credentials.

41:51I see.

41:52So your IT is happy because they know it's happening and you are not sharing that access with anyone else.

41:59But uh for your team it is very convenient because they don't have to change that uh every time anywhere.

42:06Yeah.

42:07It's just

42:08It also makes the scenario easier when someone leaves a team, right?

42:12Um, because it's just a matter of, you know, handling that detail rather than

42:17Going, oh God, now we have to go and find everything that this person, you know, put their p username and password into and and sort that out.

42:26Well that and it it allows you to share data in two different ways.

42:30You can use our sharing mechanism where you can have

42:35the uh credentials embedded with the data link and uh and

42:42It's shared within the ENSO cloud system and then you can delete that user and they lose access to it.

42:48And and we do auditing and logging and all of that.

42:51But

42:52uh some IT departments might not be comfortable with the sharing being through the ENSO auditing.

42:58They want to have the auditing uh

43:03the way they've always done it.

43:04So the IT department can require that each user has their own individual access so they can

43:12still use the auditing from the data source provider as opposed to having to go to ENSO's auditing.

43:19So it's it's all about flexibility in

43:22not only how the user works, but how IT requires the user to work.

43:29Yeah.

43:30Because uh data leaks are such a a a huge thing these days.

43:36Uh it's super important for us to help IT prevent that.

43:42Yeah.

43:42It's really Wild Falls 3 concepts.

43:44And I think maybe this comes from the benefit of

43:47Um, you don't have the legacy of like a gazillion connections that you need to support.

43:53So you've had the but maybe the benefit of being able to sort of think this problem through.

43:58with a fresh start and actually build it for today's world, right?

44:01It's um it's kind of a fortunate position to be in.

44:04And and and the other thing that we're building for today's world is is a hybrid because

44:10Uh there's desktop products out there and there's cloud products out there and neither of them work in in various situations.

44:18I mean, some companies like healthcare in particular just

44:22are not willing to put their data in somebody else's cloud platform.

44:27They might have their own cloud platform and that's their business and their IT and whatnot, but

44:31They're they're because of hi HIPAA rules and whatever else, they're just not willing to move the data into somebody else's cloud platform.

44:40Um so when when and when uh Sylvia has been running here, she's actually been running that workflow locally on her laptop.

44:51at the same time as running that catalog uh remotely in the cloud.

44:57Um I see

44:59And then she could also schedule this workflow to run in the cloud every hour, every day, or whatnot.

45:07Uh um it if

45:11Her IT allowed her to uh have her data in the cloud.

45:15Um

45:18It it's really the the cloud versus desktop is up to the the user and and it's very transparent.

45:26You you almost don't realize

45:29uh when you're running in the cloud versus when you're running in the desktop.

45:33Um except I mean uh you you do, you know, but uh it it's seamless.

45:40Yeah.

45:41And I'm saying the schedule capability is also quite flexible.

45:44Um yeah, people have very regular patterns in their business and so yeah, you know, being able to

45:51break free of having to set up multiple schedules to cover all bases is kind of nice.

45:58Super nice.

45:59So maybe in terms of the uh that question, let me just dig into that a little bit more.

46:03So

46:04If I understand this correctly, the flow that we built at the beginning, that's running locally on the laptop, then you've got this data catalog component, which I understand to be

46:14um a cloud representation of the assets available to the team and I guess to the business.

46:20And so what is that interface between sort of local development and I guess moving the process to like a more resilient

46:28Maybe not more resilient, but more sort of long-lib architecture in the cloud where it can run on a schedule.

46:37Do you publish?

46:37Like what's that sort of transition like?

46:40If I am creating a project in the cloud here in my account space, it is automatically available for everyone

46:48Right.

46:48But what I can do is I can go to my local and start from scratch on my local computer with a workflow that is just on my local uh

46:59hard drive and then I can tell that I want to export that to cloud.

47:05Right.

47:05Or I can drag and drag and drop.

47:14Might be because of screen sharing uh yeah, there you go.

47:16Yeah.

47:16Sometimes when we screen share the

47:19Screen share takes ever, the drag and drop.

47:21I've had that.

47:22Exactly.

47:23And either put that in one of the folders or to the command space and upload that.

47:30I can see that it is uploading.

47:33So I uploaded the file and it is right now available to everyone.

47:39And on the cloud we have a team space and a private space.

47:43So

47:43If you want to work in the cloud but keep your data private to you, that's still an option.

47:50Uh we we don't have the enterprise

47:54uh version yet.

47:55We're just not there.

47:56So um we only have one team right now.

48:00In in the future we'll have any number of teams.

48:03uh that uh the the sharing will be automatic within a team, but then to share between teams will be a whole

48:13Different thing.

48:15Yes, yeah.

48:16That makes sense because I think people people set up that architecture for good reasons, right?

48:21So

48:21Um, you have to kind of design for those, I guess, edge cases where you put something between two teams and now you've you've potentially opened up a can of worms in terms of how those two teams work, the organization might not realize.

48:34So

48:34That's definitely a much harder um problem to solve for today.

48:38Yeah.

48:39And and w we have all the the building blocks, uh just where we are as a company, we we are uh at this point

48:49dealing with one team at a time and and it will it will quickly migrate to more

48:57If I but what if I barest things about what Sylvia just moved into the cloud, if she were to open that from the cloud, you don't need to if you don't want, but uh it would still read the local desktop file.

49:11Um so if she shared it with you and you didn't have that file, you would get a file not found error.

49:17Right.

49:18So the the workflow and the data are are separate things.

49:23Separate assets.

49:25She can load she can move the workflow to the cloud, she can move the data to the cloud, or she can move both to the cloud, depending on what the the goals are.

49:35And and we're really embracing the hybrid

49:41If you want it local, you can have it local.

49:43If you want it cloud, you can have it cloud.

49:45If you want it half and half, you can have it half and half.

49:49I think it's very unique in

49:53That that's quite um I think that's quite unique because I think people always assume that oh if you're gonna publish this asset, you will of course you take the data with it.

50:02But it's it's it's actually

50:04from a you know separation of concerns perspective, it's actually quite nice to not assume that because it then means that uh the activity of sharing the data with the asset is more deliberate and therefore you're less likely to make mistakes around

50:17sharing or security.

50:19You you you actually have to make more of a conscious um sort of approach though, which is which is nice.

50:24I like it.

50:25Yeah.

50:27If you by some reason you cannot or you don't want to schedule things on our clouds because security, there is always an option to uh schedule that on your own computer using your

50:40I don't know, Windows Task Manager, uh any other tool or script.

50:47And we are not forbidding that.

50:49You can you can do it freely and uh

50:53It's always how we can help the users to achieve their goals more than uh how to make your wife more

51:03I love it.

51:04I love it.

51:04Yeah, no, that that's um you know that some teams you have uh you know very specific constraints around that.

51:11I think if you work in finance, for example, you have to

51:14Sometimes you have requirements around running data on very specific hardware located in a very specific place.

51:21And yeah, companies use those kinds of approaches to get around those those challenges.

51:25So yeah.

51:26I appreciate the flexibility to be honest, because I think sometimes that's that's where the SaaS companies tend to get you.

51:32Like you you're you're using their products and they're like, oh

51:36Okay.

51:37You want something convenient?

51:38Well, let me let me tell you about that.

51:40And yeah, then you start to make this is where maybe the sentiment against SaaS companies has come from long term, right?

51:46Because that's almost a cultural thing in those companies.

51:50So yeah.

51:52I I I love it.

51:54In terms of then um the product, I I know we've spoken before today's video and we talked a little bit about a community version, right?

52:02So

52:02Um what what it what is what does that look like and yeah, what how how can we get people to start playing this?

52:08Because obviously my audience very interested in exploring how the product works.

52:13I'd love to get them using it.

52:14I want to make videos on it.

52:15And so to me, I'm like, right, let's go look at the community version and see what we can build so we can get more familiar with the tool.

52:22Right.

52:23You are downloading exactly the same product and from a data prep and blend uh perspective, nothing is changing, so you are getting access to the data prep and blend.

52:34Uh the difference that you that is there is that uh for the community version you're not getting the clouds capabilities so data line threads or

52:43sharing team probabilities uh licensed and so this is one part but when you are actually signing in for the community you are getting 30 days trial

52:57uh the cloud product it is a solo uh license but it's not a problem uh for with the community version uh you can use it

53:09freely even if you have a business with revenue up to a million, I agree.

53:15So for the personal use or small businesses, if you got the team capital.

53:20Or nonprofits, right?

53:21Like you have um

53:23Yeah, much yeah, but they typically wouldn't have uh a not for profit if they have that kind of revenue is probably doing something very, very again.

53:30So yeah, no, it's uh

53:32It's an incredible opportunity.

53:34And yeah, I I think it's fantastic.

53:36I I support it.

53:37I I say thank you for doing that because um if I'm brutally honest, you see day trialists, um

53:43There's often this demand to be able to be experienced at tools, but I always say that sometimes licensing doesn't enable that, right?

53:52Because

53:53You're basically asking people to take quite a big investment in a tool and get deep and deep expertise in the tool without prior having access to it.

54:01So you can almost exclusively end you do it through a job.

54:05And in those contexts, uh sometimes the data you get to work with at work is not very interesting.

54:11So you do need uh like a community version to allow people to

54:14you know, get their football data or their sports data or maybe their own data, work with something they really understand and and know intuitively really, really well.

54:23And that gets the engagement and excitement throughout the product to really test it all.

54:28I saw in a previous sort of screen you were looking at web scraping, right?

54:31As a as an idea.

54:32And so

54:33I'm like, yeah, that would be a great little project if I'm trying to go and, you know, maybe scrape game data from like a a game that I play a lot just to just get some insights.

54:42And that becomes way more fun and social than um yeah, sometimes what we're asked to do at work.

54:48And probably more than we want to dive into today, but uh Enso is particularly good at web scraping and JSON handling uh

54:58that we'll we'll save that one for another day.

55:00But backing up, I just wanted to say your your point about nonprofits is is really important.

55:05And that's actually one of the major reasons

55:07I'm involved with ENSO.

55:09I basically retired when after the Alteryx IPO.

55:13Uh, but I've been working with a bunch of nonprofits and they have the same data problems that

55:19everybody else has and they can't afford decent tools and uh they have uh sometimes more data um because of all the the donor data and whatnot and they they don't have tools.

55:32So that that was

55:34One of my goals with ENSO is to make sure that it is available for nonprofits, especially small nonprofits that can't afford uh expensive tools.

55:45Yeah.

55:46From the perspective of the person who is running the business, uh, if I'm looking at Tabla, the community has built

55:58A lot of credit, a lot of business as well for for a topic.

56:04Goodwill, yeah.

56:06Yeah.

56:07If you have users who love the product, the product will uh evolve, will grow, and uh it will be all fine.

56:17But to get those users, it is important to be first honest with them.

56:22Second, give them the ability to use the product.

56:28That's why we don't want to get rid of uh the that free access, but we want to enable more people to use it because we believe that what we are creating is really special and uh can help them.

56:41So

56:42Yeah.

56:42I I appreciate it.

56:43It's um it's uh a very rare thing.

56:47Um and yeah, I think it's super important.

56:50Excellent.

56:51Well, I think I think we've come to a natural um conclusion.

56:55Um one of the things I wanted to do just to close out is maybe just get your perspectives on

57:01Um w what you see as the big challenges, I guess, maybe for ENSO or maybe for analytics in general.

57:07Um and then uh

57:10what if you have any questions for me, I'll also happy to happy happy to answer those.

57:14And then obviously we we talked a little bit about this before, but what I'm planning to do is over the course of

57:20the next month.

57:20I'll be using Ansa a bit more on the channel.

57:22As people are watching this, it should be March-ish.

57:25We're recording this in February.

57:27So yeah, um look up for more content on that.

57:29But maybe back to you.

57:30Yeah.

57:30What are the big challenges from an

57:32From an analytics or an ENTER perspective that you see sort of coming up in the next year or two?

57:38Um

57:40It it's just a matter of uh getting the the product out to people, you know?

57:49I mean uh

57:51ENSO has been a lot of work to get to this point and it's in a really great state.

57:57And uh so the challenge for ENSO is just introducing it to the world.

58:02Uh the challenge for for people is there's just such a variety of tools out there and

58:11uh knowing what tool is the right tool for the right uh uh company workflow is is is very hard these days.

58:21Um

58:22So um we're really thankful to you, Tim, for uh being able to share what we do with with you and your audience and um

58:32Uh hopefully that um we'll solve some problems for both us and and your audience making that introduction.

58:38Uh

58:40But I I I can't tell you how many times I see people using the wrong tool for the job.

58:44And that's not to say that ENSO is always the right tool for the job.

58:48It might be the wrong tool for a lot of jobs, but

58:51Yeah.

58:51Uh just getting people to to the right solutions is is a challenge and and amazing.

59:04Good.

59:05All right.

59:05I think uh yeah, do you have any questions for me?

59:08Maybe maybe you came with a burning question.

59:10You don't need to uh ask one if you don't want to, but I always I always offer the uh the opportunity

59:17Well you you go by uh tableau Tim.

59:20Uh obviously you're you're uh branching out and looking at other aspects.

59:28Uh what what do you see as the your biggest goals for the next year or two in terms of where where you are going?

59:39Yeah, it's a great question.

59:41It's actually something I've spent probably longer than the last year thinking about just because of, you know, what's been happening in this space.

59:48Um

59:49I've written about this a little bit in in the last couple of weeks, but if I if I step back, I think one of the

59:58One of the challenges with analytics in general is that I think the the way we use the tools, the workflow is that underpin the tools that we choose.

60:08um they're being challenged.

60:10I don't think they're changing.

60:11I think they're just being challenged.

60:12And so you've got new new ways of thinking about the same workflows, new ways of

60:17uh conceptualizing the same workplace that we've we've had in analytics for ages.

60:21Like semantics is a big thing at the moment, but it's existed for a long time.

60:24It's just rearing its head in a different context.

60:27And I think the the the the the the challenge for me is at least for the last five years I've ridden um I've ridden this wave of teaching people about data through the lens of one tool

60:40And I think what that's led to is um I've I've almost co-opted and adopted a philosophy around data.

60:49that is very strongly aligned with uh a a product's own philosophy, if that makes sense.

60:54So the way I think about my workflows, the way I think about the way I do things.

60:58naturally just aligns itself with like the tabloid ecosystem because that's what I've been spending a lot of time covering.

61:04But as these challenges are coming in, I actually naturally feel that sort of perspective being challenged and say

61:11one of the things I've set out to do is to, you know, just explore a bigger range of tools.

61:15And at least to my audience, I'll always be Tableau Tim because that's who they know me as.

61:20But

61:21I think I've also maybe gained their trust over those five years that I that I know enough about the field and the topics to be able to guide them through

61:29what change might look like for someone who happens to have had the same genny as I have, someone who's spent a long time having a philosophy that's aligned with the BI tool, whether it's Tableau, whether it's Power BI, whatever it is, right?

61:42And then how do we teach those people not to fall into those same same traps?

61:46So I'm really exploring new tools, I'm really exploring new ways of working.

61:50And I started a newsletter for the very fact that I wanted to challenge the way that I think in the form of writing.

61:56Uh whereas with video, it's you know, you you you tune in to listen to my perspective.

62:00So it's

62:01It doesn't really challenge me and I just get to say what I want.

62:04Whereas with writing, people can critique it in a different way.

62:07So those I think are the two big things, challenging my workflow and then bringing my audience with me and then challenging myself to learn um, you know

62:15Uh the the industry almost all over again because I think yeah I've been aligned to philosophy for a long time.

62:21Oh, I can't wait to follow your journey.

62:23That sounds fun.

62:24Yeah, yeah, it's a big challenge as well.

62:26You the the the social media platforms don't help with the way they um they they sort of handle audiences and stuff.

62:33You don't really own the audience per se when you're um you know working on one of these platforms like YouTube or LinkedIn.

62:39you're borrowing the audience just for a moment.

62:41And if you if you do the bidding of the platform, they'll push your message.

62:45But if you don't, don't show your punter to anyone.

62:50So yeah, there is that.

62:52Very true.

62:52Um it's a big challenge.

62:55But there is a huge opportunity also to get to all of those people who are saying, okay, I have learned SQL, I have learned um basically

63:04I have learned so many tools and how do I get to be a data analyst?

63:08Because I still don't know.

63:10So learn from you because you have the perspective as well.

63:15Yeah, no, absolutely.

63:16I think um yeah, there there's maybe a bit of conversions happening in the industry.

63:22Uh I don't think it's because uh you know AI is really not yet truly challenging, I think

63:27our roles, but I think it is is challenging the way we approach work, the way we think about work, because there is a sort of

63:34uh openness to explore new ways of thinking again, even if we end up realizing that what we did before was perfectly fine.

63:41We just need better tools or, you know, m better ways of thinking.

63:44So yeah, there there's there is a huge opportunity in the data niche.

63:48And I think um

63:49I'm always of the view that uh even if AI becomes a big thing in the industry, it's only gonna increase the amount of opportunity to work with data, not decrease it.

63:57And so we're still gonna need tools, we're still gonna need people.

64:01And I think it's a mistake to assume that, you know, AI will will replace the roles that we've played in the businesses

64:09It might replace some of the workflows we execute, but I don't think it'll replace what we do in the business and how we understand data.

64:15So anyhow, I'm I'll fly that flag.

64:18As long as I can agree with you.

64:20Until I'm wrong.

64:22Until I'm wrong then then we can have another discussion.

64:27Amazing.

64:28Okay.

64:28Um, yeah, we'll call it there.

64:30Thank you both of you for coming on the channel.

64:32I I really appreciate this.

64:33Um, you know, I've followed Ansip for quite a bit of time.

64:36I've I've used it in in a word context and also just in there

64:39in a social context.

64:40So yeah, over the coming uh weeks we're going to be using it more on the channel.

64:43And yeah, hopefully everyone watching starts to play around with it.

64:46And yeah, I know you've got product updates as well.

64:49So I'll be keeping an eye on those and excited to see what you built.

64:52Thank you very much.

64:53Thank you, Brick, and see you on maybe our community plan platform.

64:58Yes.

64:59All right, perfect.

65:00Take care.