Man­age Data Where It Lives With Dat­ab­ricks Apps

As organ­isa­tions scale their data estates, more and more of the day-to-day work shifts from build­ing new pipelines to man­aging, main­tain­ing, and gov­ern­ing the ones that already exist. Someone has to check that yes­ter­day’s load ran cleanly, that a schema change did­n’t silently break a down­stream report. Dat­ab­ricks Apps close that gap. They address this by provid­ing light­weight, gov­erned applic­a­tions that run dir­ectly inside the Dat­ab­ricks work­space. They turn data products and mod­els into inter­act­ive tools without export­ing data to external sys­tems while also keep­ing gov­ernance guard­rails at a high level.

Dat­ab­ricks apps are designed for oper­a­tional work­flows: mon­it­or­ing pipeline health, val­id­at­ing data qual­ity, hand­ling schema changes, coordin­at­ing approvals, and provid­ing safe inter­faces for busi­ness users to inter­act with data. They are espe­cially use­ful for self‑service tasks, such as cur­ated access and approval flows tied to gov­ernance con­trols. These apps provide a gov­erned, in-work­space inter­face to per­form  oper­a­tional tasks. Teams can also use Dat­ab­ricks Apps to mon­itor pipeline health, run self-ser­vice data qual­ity checks, detect anom­alies, track cost and usage, sup­port metadata stew­ard­ship, and give busi­ness users safe, cur­ated access to data, all while pre­serving the secur­ity, lin­eage, and per­mis­sions the plat­form already provides.

Table of contents

Estim­ated read­ing time: 5 minutes

What Dat­ab­ricks Apps Are and Why They Matter

Dat­ab­ricks Apps are web applic­a­tions that are deployed dir­ectly to the Dat­ab­ricks envir­on­ment. They allow teams to trans­form repeat­able data work­flows like gov­ernance decisions, qual­ity gates, and oper­a­tional tasks into user‑friendly inter­faces that min­im­ise manual inter­ven­tions and reduce risk. Instead of ad‑hoc note­books or SQL quer­ies, you can provide struc­tured “front doors” for busi­ness users, data stew­ards, and engin­eers. A way to do con­trolled approval flows, gov­erned data edits, and self‑service oper­a­tions that fully respect Unity Cata­log per­mis­sions and auditability.

Dat­ab­ricks pos­i­tions Apps as a way to quickly build secure data and AI applic­a­tions using famil­iar Python frame­works like Dash while keep­ing deploy­ment and main­ten­ance over­head min­imal. For Dat­ab­ricks, they are the nat­ural next step towards an end-to-end data plat­form that cov­ers every step from data inges­tion over trans­form­a­tions, data qual­ity and gov­ernance up to BI and operations.

What These Apps Look Like in Practice 

Apps serve as con­trolled, gov­erned inter­faces for pro­cesses that would oth­er­wise require dir­ect data access or informal approvals. Com­mon examples include: 

  • Approval work­flows for gov­ernance decisions: 
    • Self-ser­vice access con­trol portal
    • PII (Per­son­ally Iden­ti­fi­able Inform­a­tion) data clas­si­fic­a­tion, tag­ging and anonymisa­tion based on engin­eer­ing, busi­ness and GDPR requirements
  • Qual­ity gate dash­boards that val­id­ate data before promotion
  • Self‑service oper­a­tional tools enabling busi­ness users to inter­act with cur­ated pro­cesses without using note­books or SQL 
  • Gov­ernance toolkits like Ontos, a data con­tract and data product gov­ernance plat­form built and main­tained by Dat­ab­ricks, which fol­lows the open source ODPS and ODCS stand­ards to unify how busi­nesses buy, man­age, and scale data products.

A Key Archi­tec­tural Detail: The Read/Write Access Pat­tern 

Effect­ive gov­ernance hinges on how the app inter­acts with data. Dat­ab­ricks Apps allow to imple­ment a hybrid access pattern:

Both the access to the app and the read oper­a­tions inside the app occur “on behalf of” the authen­tic­ated user lever­aging their indi­vidual access. Unity Cata­log per­mis­sions apply auto­mat­ic­ally, ensur­ing data access is con­sist­ent and trace­able. This also ensures that data mask­ing and row level secur­ity rules work con­sist­ently accross the platform.

Writes are per­formed via a ded­ic­ated ser­vice prin­cipal, allow­ing updates only through the app’s val­id­ated busi­ness logic, while record­ing who ini­ti­ated the change for full auditability. 

A few Examples

Let’s explore some examples of how Dat­ab­ricks Apps can enhance com­mon data oper­a­tion pat­terns that would oth­er­wise require elev­ated priv­ileges and addi­tional audit trails.

Manual Data Adjust­ments with Approval Process

Wherever approval pro­cesses or manual adjust­ments of data are needed, apps can provide a plat­form that provides an inter­face for non-tech­nical users.

In this example, an app is provided that exposes the func­tion­al­ity to manu­ally adjust KPI factors applied dur­ing data pro­cessing. These adjust­ment factors are oth­er­wise provided by CSV or Excel. And while manual file trans­fer works, it is error-prone, espe­cially due to loc­al­isa­tion and user error.

1. Request tab 
The requester selects an attrib­ute, loads ref­er­ence data into a grid, edits only per­mit­ted fields, pre­views changes, and sub­mits a “Change Request”. The sys­tem records all rel­ev­ant metadata (requested_by, requested_on, com­ments) asso­ci­ated with the request. 
Only columns con­figured in a YAML con­fig are opened for changes. If changes are reques­ted and approved suc­cess­fully they flow back into the ETL pipelines based on their ori­ginal keys. The changable columns are high­lighted in the UI.

2. Bulk Upload tab
The user can upload a full CSV or Excel file to manu­ally pop­u­late the plat­form with data. The data struc­ture and com­pli­ance with the expec­ted schema are checked. Within the same page the user instantly receives feed­back on any issues.

3. My Requests tab 
Requesters can track each sub­mis­sion, with status derived from approval metadata. 

4. Approve tab 
Approvers view only requests sub­mit­ted by oth­ers, enfor­cing the four‑eyes prin­ciple. Approvals cap­ture approver metadata and may trig­ger down­stream pro­cessing or noti­fic­a­tions. Cur­rent and pro­posed entries are dis­played next to each other for vis­ib­il­ity.
Approvers can select indi­vidual rows to approve. On approval repro­cessing of the down­stream pipeline starts automatically.

PII Data Clas­si­fic­a­tion, Tag­ging and Anonymisa­tion App

Strong data gov­ernance plays a key role in GDPR and PII and ties in closely with a busi­ness’ data strategy. Espe­cially the dis­cov­ery and clas­si­fic­a­tion of such data are tedi­ous and often a manual process.

This example is another approval work­flow app. It stand­ard­ises how organ­isa­tions identify, val­id­ate, and approve sens­it­ive data in Dat­ab­ricks Unity Cata­log. The app ensures that data is tagged based on com­pli­ance needs before mask­ing, hash­ing, and related rules are applied. Admin­is­trat­ors con­fig­ure approvers on a pro­ject level while review­ers pro­cess assigned approval cases via a guided work­flow. The work­flow cap­tures table con­text and enforces column-level clas­si­fic­a­tions, cat­egor­ies, and anonymisa­tion set­tings. Built-in guard­rails pre­vent invalid classification/category com­bin­a­tions and sup­port dis­cov­ery of miss­ing tables or columns. Final­iz­ing the pro­cess applies gov­erned tags to tables and columns. A gov­erned Tag is an account-level tag, set up by admin­is­trat­ors. They can enforce policies or cus­tom rules for attrib­ute based access con­trol or flag data based on GDPR.

Dat­ab­ricks Labs’ Ontos

Ontos is a Dat­ab­ricks Labs pro­ject that adds a busi­ness gov­ernance and semantic layer on top of Unity Cata­log. It’s goal is to help organ­iz­a­tions move bey­ond tech­nical metadata (tables, schemas, lin­eage, per­mis­sions) and man­age the busi­ness mean­ing, own­er­ship, qual­ity, and life­cycle of data products.

Built in React, dir­ectly integ­rated with Dat­ab­ricks Unity Cata­log and Dat­ab­ricks Lake­base as a fast stor­age layer it offers a stack of data gov­ernance functionality:

  • AI-assisted data discovery
  • Data product management
  • Busi­ness glossaries
  • Data con­tracts
  • Semantic/ontology man­age­ment
  • Stew­ard­ship workflows
  • Data mar­ket­place capabilities
Dat­ab­ricks Labs‘ Ontos – Gov­ernance Suite – Land­ing Page

Bot­tom Line 

Dat­ab­ricks Apps extend the Lake­house bey­ond ana­lyt­ics by bring­ing gov­erned oper­a­tional pro­cesses dir­ectly into the Dat­ab­ricks plat­form. They enable organ­iz­a­tions to build secure, user-friendly inter­faces for approvals, data stew­ard­ship, qual­ity man­age­ment, access requests, and gov­ernance work­flows without mov­ing data out­side the plat­form. By com­bin­ing user-based read access with con­trolled write oper­a­tions, Dat­ab­ricks Apps help enforce com­pli­ance, improve audit­ab­il­ity, and reduce manual effort while pre­serving Unity Cata­log secur­ity, lin­eage, and gov­ernance stand­ards. Whether sup­port­ing approval pro­cesses, PII clas­si­fic­a­tion, or broader data product gov­ernance through solu­tions such as Ontos, Dat­ab­ricks Apps provide a prac­tical way to turn gov­ernance from a manual bur­den into an integ­rated, scal­able oper­at­ing model.