6 ms·

Show HN: Tablum.io – No-Code Self-Service BI/Data Analytics Tool

42 points by instad 5y ago

mattm 5y ago
Good luck. I previously worked for a BI company. It's an extremely crowded space.
- instad 5y ago
  Thanks! Yes, it’s very competitive environment. So we’re trying to stand out as a data analytics tool for developers.
mercurialsolo 5y ago
The interface lacks punch. While the idea of making a self-serve BI tool which can plugin the gap of bringing better understanding around data is super-cool, the execution here for sure needs work. Tried it out and it's cool for someone to hack with right now. Early launch.
- instad 5y ago
  Thanks. We've just recently announced the very first version, and looking forward to any feedback and feature requests. Will do our best to make interface punchy!
tmtwhn 5y ago
No information on pricing and social only logins? I was interested in this product, but no thanks. At the very least if you're not ready to commit you could provide "beta" pricing for a year until you get that figured out.
- instad 5y ago
  Fixed. Added pricing and email registration. It would be awesome if you could take a look and give some feedback. Thanks!
- instad 5y ago
  There's no paid plans at the moment. Just try it out for free with full features. Just curious, why the social logins are bad?
  - garciasn 5y ago
    Because they allow third parties to build bigger ID graphs and audiences for ad targeting. By all means, have them for convenience for those users who want to use them, but for the rest of us who know better, we’d appreciate the option to provide you our burner/one-time email addresses.
    - instad 5y ago
      We're going to extend the way to login for the cloud version. Currently, email/password auth is available only in self-hosted version (docker image installed on your VPS). Feel free to reach out to me info@tablum.io, if you'd like to test it. Thanks.
      - tmtwhn 5y ago
        Glad to hear this. Marketing that you'll offer a self hosted version and that email logins are coming soon would go a long way to improving trust.
  - hungryforcodes 5y ago
    > There's no paid plans at the moment. Just try it out for free with full features. OK. That's compelling enough. > Just curious, why the social logins are bad? How uniquely identifying are they to be honest?
  - abakker 5y ago
    because this is a tool for doing work? Nobody (or very few) HN people are going to want to do BI with their facebook accounts, if we even have FB accounts.
    - instad 5y ago
- runako 5y ago
  +1. Given that prices for products in this space range over several orders of magnitude, I wouldn't spend time evaluating without knowing whether my budget is even in the same ballpark. Would I expect to spend $1k annually? $10k? $100k? More?
  - instad 5y ago
    Fixed. Added pricing. Thanks!
  - instad 5y ago
    Got you. We will add the pricing soon just for the sake of right budgeting.
roystonvassey 5y ago
Cool stuff. Just curious- how does it stack against Metabase (https://www.metabase.com/product/ https://www.metabase.com/product/) for example? I used Metabase a few years back and was really blown away with the ease of use.
- marcinzm 5y ago
  This seems more of an end-to-end system including etl job scheduling, data loading and local storage.
  - instad 5y ago
    Exactly!
- eatonphil 5y ago
  Metabase is primarily for sql and this system seems to support more than just sql.
  - vladsanchez 5y ago
    Agreed! Suddenly I thought of Metabase, Looker or even Superset.
  - roystonvassey 5y ago
    Ah right, thanks!
- instad 5y ago
  Thanks! Well, in general they both are self-service BIs, and have common features. Yet, TABLUM.IO is focused more on faster data extraction from files e.g. you can just drag and drop JSON, XML or CSV file to load data from it. Or specify an URL for data loading from remote host. Then it is stored in a local storage as a "view" (similar to sql table). And you can work with it like it is an SQL table using sqlite syntax: JOIN with other loaded data, filter, clean up, export or generate charts. Of cause, it has connectors to popular DBs such as postgres, mysql and clickhouse. Ultimately, we're working towards a faster data analysis so you won't spend too much time setting up connectors to files and databases. Just get the results as fast as possible. Same with data analysis: it has embedded support for aggregate - and window-functions. You can get "MOVING AVERAGE" just clicking the table columns (no keystrokes required). Dashboards are not yet implemented, but we're working on it. Thanks!
prepend 5y ago
The dream of no code lives on. I’ve tried many no code tools over the decades. As an individual and as a group. Ease of use is really important. But I think easy for Python programmers is better than easy for people to click. Logic is logic, so whether you program in code or via some wack gui, you still need programmers. Every time it’s come down to regretting committing to a closed, proprietary “programming language” that is the no code tool. The exception is for simple stuff. But with data projects simple projects usually get complicated fast. So Id rather use a tool that scales well rather than one that starts easy but then is very hard to scale.
- instad 5y ago
  I would call TABLUM.IO - a low-code tool. Power SQL user can type any complex SQL query in the SQL console. Whereas the beginners won't spend much time doing simple "SELECT" or calculating the average, they can just click on the header of the table and select "stats" to see the result. I agree that no-code is limited to some general scenario, but there're still a lot of them. Thus, we decided to leave an opportunity to choose which way to go: use mouse for simple / general actions or use SQL console by power users to archive some complex results.
- narush 5y ago
  Totally agree with this comment. This sort of "no-lock-in" thinking is exactly what inspired Mito [1] (I'm a co-founder, FYI). Mito is low-code data analysis tool inspired by spreadsheets. When you edit the spreadsheet, Mito generates Python code directly into your Jupyter lab notebook. The benefit here is that there's no lock in. If Mito support the action the user wants to take, they can use Mito for a quick point-and-click experience. If not, they can just write Python code as they normally do! [1] https://trymito.io/hn https://trymito.io/hn
- hermitcrab 5y ago
  No-code BI and ETL tools are intended mainly as a replacement for Excel, not a replacement for Python. If you are an accomplished Python programmer, you might code a solution or (e.g. if you want to do something ad hoc quickly) use a no-code tool. If you aren't a programmer, your choice is typically: learn Python, use Excel, wait for a programmer to do it for you or use a no-code tool. A no-code tool will often be by far the best solution.
hermitcrab 5y ago
Feedback - the music on the demo video is horrible (IMHO).
- instad 5y ago
  Thanks for watching.
- Bishonen88 5y ago
  Not only the music, but the whole video was lacklustre. The effects used were not really fitting the whole vibe of a BI tool (personal opinion). Contrast this with a video like this one: https://www.youtube.com/watch?v=WuIS-m0B2cY https://www.youtube.com/watch?v=WuIS-m0B2cY After watching the above, I'm more curious to check out the real product.
rexreed 5y ago
Not to be flippant, but in what way is Excel not a no-code / self-service BI / Data Analytics tool? Or where Google Sheets is heading? Serious question. The average non-technical user goes first to Excel or Excel-like things.
- Frost1x 5y ago
  Excel doesn't have a lot of visual based development features aside from the tabular structure and graphical data and function reference. People still have to understand function calls and conditionals which aren't represented in what I would call verging on "no-code" ideals. Excel applications also tend to start rather innocently where something rather simple is programmed in the cellular structure. People quickly feel empowered and begin adding more and more dynamic aspects. At some point spreadsheet based applications turn more into some unmaintainble monstrosity that have all sorts of business and application logic embedded in them that need to get handed elsewhere and someone else familiar with "code" or software design has to translate everything to. I've been involved in more of this than I care to admit for business applications and it's usually a nightmare because the applications become more critical to the business than anyone but the spreadsheet maintainer is aware of. Suddenly something new and time critical is needed or an error is discovered and the person who created it largely can't remember everything they created and why, so it turns into an software archeological endeavor. I've seen notebooks with dozens of sheets containing all sorts of embedded data structures on different sheet with all sorts of relationships inside and outside of sheets and nightmare chains of conditionals added in that I've had to sift through. Sometimes the user then discovers macros or VBA and it becomes a dice roll if things are better or worse (usually better because at least you can quickly follow all the embedded relationships even if they're not well structured unlike logic embedded in cells with all sorts of spatial references you need 10 copies open to trace and debug). Excel is hardly "no code" it just has a nice beginner learning curve that gets people into dangerous situations before they know they are. I'm not entirely sure about the direction of Google Sheets as I haven't used it in years.
- eatonphil 5y ago
  My workflow for the longest time has been: use a SQL GUI to run SQL queries but then dump that to CSV so I could load in Excel and make graphics I could embed in a presentation or sent in an email. I don't think Excel or Google Sheets are the right interface for diverse data sources that have different query styles and different connection types. Visually the space in each Cell is just pretty constrained. I am building an open source tool [0] for these kinds of workflows where you need to combine queries, scripting in Python/JavaScript/etc, graphing, dashboards, email exports in one place. It's very similar to tablum.io (this post) but instead of doing any ETL it's just a smart client for joining data client-side. [0] https://github.com/multiprocessio/datastation https://github.com/multiprocessio/datastation
TekMol 5y ago
The site wants me to log in via Google or Facebook. No, thanks.
- instad 5y ago
  We will enable email registration and authorization soon. Stay tuned. And thanks for commenting.
- instad 5y ago
  Fixed. Added email registration. Could you please test it now?
crad 5y ago
"Try for Free" with no info about what I would pay... no thanks.
- instad 5y ago
  Fixed. Added pricing on website: https://tablum.io/pricing https://tablum.io/pricing
- instad 5y ago
  This is the early stage product launch, the product is completely free for now. So no worries regarding the pricing and charged, you will not be asked about credit cards. Feel free to test it. We do it mostly to collect feedback on the missing features and workflow. Thanks!
asavinov 5y ago
Most of the self-service or no-code BI, ETL, data wrangling tools are am aware of (like airtable, fieldbook, rowshare, Power BI etc.) were thought of as a replacement for Excel: working with tables should be as easily as working with spreadsheets. This problem can be solved when defining columns within one table: ColumnA = ColumnB + ColumnC, ColumnD = ColumnA * ColumnE we get a graph of column computations similar to the graph of cell dependencies in spreadsheets. Yet, the main problem is in working multiple tables: how can we define a column in one table in terms of columns in other tables? For example: Table1::ColumnA = FUNCTION(Table2::ColumnB, Table3::ColumnC) Different systems provide different answers to this question but all of them are highly specific and rather limited. Why it is difficult to define new columns in terms of other columns in other tables? Short answer is that working with columns is not the relational approach. The relational model is working with sets (rows of tables) and not with columns. One generic approach to working with columns in multiple tables is provided in the concept-oriented model of data which treats mathematical functions as first-class elements of the model. Previously it was implemented in a data wrangling tool called Data Commander. But them I decided to implement this model in the Prosto data processing toolkit which is an alternative to map-reduce and SQL: https://github.com/asavinov/prosto https://github.com/asavinov/prosto It defines data transformations as operations with columns in multiple tables. Since we use mathematical functions, no joins and no groupby operations are needed and this significantly simplifies and makes more natural the task of data transformations. Moreover, now it provides Column-SQL which makes it even easier to define new columns in terms of other columns: https://github.com/asavinov/prosto/blob/master/notebooks/column-sql.ipynb https://github.com/asavinov/prosto/blob/master/notebooks/col...
asavinov 5y ago
Similar products: - https://rows.com/ https://rows.com/ - https://www.rowshare.com/ https://www.rowshare.com/ - https://www.airtable.com/ https://www.airtable.com/ - fieldbook - Power BI
- instad 5y ago
  Not exactly. Timing is the issue when it comes to mixed type of data sources. try to load data via the URL (e.g. the results of API invocation or rss feed), then merge it with the data from postgres sql table or data from CSV file. It becomes challenging with the listed tools. Whereas TABLUM.IO solves the issue and make it organic and fast.

42 points by instad 5y ago

mattm 5y ago

Good luck. I previously worked for a BI company. It's an extremely crowded space.

instad 5y ago
Thanks! Yes, it’s very competitive environment. So we’re trying to stand out as a data analytics tool for developers.

mercurialsolo 5y ago

The interface lacks punch. While the idea of making a self-serve BI tool which can plugin the gap of bringing better understanding around data is super-cool, the execution here for sure needs work. Tried it out and it's cool for someone to hack with right now. Early launch.

instad 5y ago
Thanks. We've just recently announced the very first version, and looking forward to any feedback and feature requests. Will do our best to make interface punchy!

tmtwhn 5y ago

No information on pricing and social only logins? I was interested in this product, but no thanks. At the very least if you're not ready to commit you could provide "beta" pricing for a year until you get that figured out.

instad 5y ago
Fixed. Added pricing and email registration. It would be awesome if you could take a look and give some feedback. Thanks!
instad 5y ago
There's no paid plans at the moment. Just try it out for free with full features. Just curious, why the social logins are bad?
- garciasn 5y ago
  Because they allow third parties to build bigger ID graphs and audiences for ad targeting. By all means, have them for convenience for those users who want to use them, but for the rest of us who know better, we’d appreciate the option to provide you our burner/one-time email addresses.
  - instad 5y ago
    We're going to extend the way to login for the cloud version. Currently, email/password auth is available only in self-hosted version (docker image installed on your VPS). Feel free to reach out to me info@tablum.io, if you'd like to test it. Thanks.
    - tmtwhn 5y ago
      Glad to hear this. Marketing that you'll offer a self hosted version and that email logins are coming soon would go a long way to improving trust.
- hungryforcodes 5y ago
  > There's no paid plans at the moment. Just try it out for free with full features. OK. That's compelling enough. > Just curious, why the social logins are bad? How uniquely identifying are they to be honest?
- abakker 5y ago
  because this is a tool for doing work? Nobody (or very few) HN people are going to want to do BI with their facebook accounts, if we even have FB accounts.
  - instad 5y ago
runako 5y ago
+1. Given that prices for products in this space range over several orders of magnitude, I wouldn't spend time evaluating without knowing whether my budget is even in the same ballpark. Would I expect to spend $1k annually? $10k? $100k? More?
- instad 5y ago
  Fixed. Added pricing. Thanks!
- instad 5y ago
  Got you. We will add the pricing soon just for the sake of right budgeting.

roystonvassey 5y ago

Cool stuff. Just curious- how does it stack against Metabase (https://www.metabase.com/product/ https://www.metabase.com/product/) for example? I used Metabase a few years back and was really blown away with the ease of use.

marcinzm 5y ago
This seems more of an end-to-end system including etl job scheduling, data loading and local storage.
- instad 5y ago
  Exactly!
eatonphil 5y ago
Metabase is primarily for sql and this system seems to support more than just sql.
- vladsanchez 5y ago
  Agreed! Suddenly I thought of Metabase, Looker or even Superset.
- roystonvassey 5y ago
  Ah right, thanks!
instad 5y ago
Thanks! Well, in general they both are self-service BIs, and have common features. Yet, TABLUM.IO is focused more on faster data extraction from files e.g. you can just drag and drop JSON, XML or CSV file to load data from it. Or specify an URL for data loading from remote host. Then it is stored in a local storage as a "view" (similar to sql table). And you can work with it like it is an SQL table using sqlite syntax: JOIN with other loaded data, filter, clean up, export or generate charts. Of cause, it has connectors to popular DBs such as postgres, mysql and clickhouse. Ultimately, we're working towards a faster data analysis so you won't spend too much time setting up connectors to files and databases. Just get the results as fast as possible. Same with data analysis: it has embedded support for aggregate - and window-functions. You can get "MOVING AVERAGE" just clicking the table columns (no keystrokes required). Dashboards are not yet implemented, but we're working on it. Thanks!

prepend 5y ago

The dream of no code lives on. I’ve tried many no code tools over the decades. As an individual and as a group. Ease of use is really important. But I think easy for Python programmers is better than easy for people to click. Logic is logic, so whether you program in code or via some wack gui, you still need programmers. Every time it’s come down to regretting committing to a closed, proprietary “programming language” that is the no code tool. The exception is for simple stuff. But with data projects simple projects usually get complicated fast. So Id rather use a tool that scales well rather than one that starts easy but then is very hard to scale.

instad 5y ago
I would call TABLUM.IO - a low-code tool. Power SQL user can type any complex SQL query in the SQL console. Whereas the beginners won't spend much time doing simple "SELECT" or calculating the average, they can just click on the header of the table and select "stats" to see the result. I agree that no-code is limited to some general scenario, but there're still a lot of them. Thus, we decided to leave an opportunity to choose which way to go: use mouse for simple / general actions or use SQL console by power users to archive some complex results.
narush 5y ago
Totally agree with this comment. This sort of "no-lock-in" thinking is exactly what inspired Mito [1] (I'm a co-founder, FYI). Mito is low-code data analysis tool inspired by spreadsheets. When you edit the spreadsheet, Mito generates Python code directly into your Jupyter lab notebook. The benefit here is that there's no lock in. If Mito support the action the user wants to take, they can use Mito for a quick point-and-click experience. If not, they can just write Python code as they normally do! [1] https://trymito.io/hn https://trymito.io/hn
hermitcrab 5y ago
No-code BI and ETL tools are intended mainly as a replacement for Excel, not a replacement for Python. If you are an accomplished Python programmer, you might code a solution or (e.g. if you want to do something ad hoc quickly) use a no-code tool. If you aren't a programmer, your choice is typically: learn Python, use Excel, wait for a programmer to do it for you or use a no-code tool. A no-code tool will often be by far the best solution.

hermitcrab 5y ago

Feedback - the music on the demo video is horrible (IMHO).

instad 5y ago
Thanks for watching.
Bishonen88 5y ago
Not only the music, but the whole video was lacklustre. The effects used were not really fitting the whole vibe of a BI tool (personal opinion). Contrast this with a video like this one: https://www.youtube.com/watch?v=WuIS-m0B2cY https://www.youtube.com/watch?v=WuIS-m0B2cY After watching the above, I'm more curious to check out the real product.

rexreed 5y ago

Not to be flippant, but in what way is Excel not a no-code / self-service BI / Data Analytics tool? Or where Google Sheets is heading? Serious question. The average non-technical user goes first to Excel or Excel-like things.

Frost1x 5y ago
Excel doesn't have a lot of visual based development features aside from the tabular structure and graphical data and function reference. People still have to understand function calls and conditionals which aren't represented in what I would call verging on "no-code" ideals. Excel applications also tend to start rather innocently where something rather simple is programmed in the cellular structure. People quickly feel empowered and begin adding more and more dynamic aspects. At some point spreadsheet based applications turn more into some unmaintainble monstrosity that have all sorts of business and application logic embedded in them that need to get handed elsewhere and someone else familiar with "code" or software design has to translate everything to. I've been involved in more of this than I care to admit for business applications and it's usually a nightmare because the applications become more critical to the business than anyone but the spreadsheet maintainer is aware of. Suddenly something new and time critical is needed or an error is discovered and the person who created it largely can't remember everything they created and why, so it turns into an software archeological endeavor. I've seen notebooks with dozens of sheets containing all sorts of embedded data structures on different sheet with all sorts of relationships inside and outside of sheets and nightmare chains of conditionals added in that I've had to sift through. Sometimes the user then discovers macros or VBA and it becomes a dice roll if things are better or worse (usually better because at least you can quickly follow all the embedded relationships even if they're not well structured unlike logic embedded in cells with all sorts of spatial references you need 10 copies open to trace and debug). Excel is hardly "no code" it just has a nice beginner learning curve that gets people into dangerous situations before they know they are. I'm not entirely sure about the direction of Google Sheets as I haven't used it in years.
eatonphil 5y ago
My workflow for the longest time has been: use a SQL GUI to run SQL queries but then dump that to CSV so I could load in Excel and make graphics I could embed in a presentation or sent in an email. I don't think Excel or Google Sheets are the right interface for diverse data sources that have different query styles and different connection types. Visually the space in each Cell is just pretty constrained. I am building an open source tool [0] for these kinds of workflows where you need to combine queries, scripting in Python/JavaScript/etc, graphing, dashboards, email exports in one place. It's very similar to tablum.io (this post) but instead of doing any ETL it's just a smart client for joining data client-side. [0] https://github.com/multiprocessio/datastation https://github.com/multiprocessio/datastation

TekMol 5y ago

The site wants me to log in via Google or Facebook. No, thanks.

instad 5y ago
We will enable email registration and authorization soon. Stay tuned. And thanks for commenting.
instad 5y ago
Fixed. Added email registration. Could you please test it now?

crad 5y ago

"Try for Free" with no info about what I would pay... no thanks.

instad 5y ago
Fixed. Added pricing on website: https://tablum.io/pricing https://tablum.io/pricing
instad 5y ago
This is the early stage product launch, the product is completely free for now. So no worries regarding the pricing and charged, you will not be asked about credit cards. Feel free to test it. We do it mostly to collect feedback on the missing features and workflow. Thanks!

asavinov 5y ago

Most of the self-service or no-code BI, ETL, data wrangling tools are am aware of (like airtable, fieldbook, rowshare, Power BI etc.) were thought of as a replacement for Excel: working with tables should be as easily as working with spreadsheets. This problem can be solved when defining columns within one table: ColumnA = ColumnB + ColumnC, ColumnD = ColumnA * ColumnE we get a graph of column computations similar to the graph of cell dependencies in spreadsheets. Yet, the main problem is in working multiple tables: how can we define a column in one table in terms of columns in other tables? For example: Table1::ColumnA = FUNCTION(Table2::ColumnB, Table3::ColumnC) Different systems provide different answers to this question but all of them are highly specific and rather limited. Why it is difficult to define new columns in terms of other columns in other tables? Short answer is that working with columns is not the relational approach. The relational model is working with sets (rows of tables) and not with columns. One generic approach to working with columns in multiple tables is provided in the concept-oriented model of data which treats mathematical functions as first-class elements of the model. Previously it was implemented in a data wrangling tool called Data Commander. But them I decided to implement this model in the Prosto data processing toolkit which is an alternative to map-reduce and SQL: https://github.com/asavinov/prosto https://github.com/asavinov/prosto It defines data transformations as operations with columns in multiple tables. Since we use mathematical functions, no joins and no groupby operations are needed and this significantly simplifies and makes more natural the task of data transformations. Moreover, now it provides Column-SQL which makes it even easier to define new columns in terms of other columns: https://github.com/asavinov/prosto/blob/master/notebooks/column-sql.ipynb https://github.com/asavinov/prosto/blob/master/notebooks/col...

asavinov 5y ago

Similar products: - https://rows.com/ https://rows.com/ - https://www.rowshare.com/ https://www.rowshare.com/ - https://www.airtable.com/ https://www.airtable.com/ - fieldbook - Power BI

instad 5y ago
Not exactly. Timing is the issue when it comes to mixed type of data sources. try to load data via the URL (e.g. the results of API invocation or rss feed), then merge it with the data from postgres sql table or data from CSV file. It becomes challenging with the listed tools. Whereas TABLUM.IO solves the issue and make it organic and fast.