After six months of building and using it in private by myself and my team, here is the tool I wished I had on every Fusion project.
If you work with Oracle Fusion Cloud for a living, you already know the thing nobody puts in the brochure: you cannot directly connect to the database.
Oracle Fusion is SaaS, Oracle owns the database, and the door is closed. That is a perfectly reasonable design decision for a multi-tenant cloud. It is also the single biggest productivity challenge on anyone whose job is to understand what is actually in the system.
Every consultant, support analyst, and IT team I know has developed the same coping ritual:
- Open BI Publisher.
- Create a data model.
- Paste a SQL query into it.
- Save it. Create a report on top of it. Save that too.
- Run the report, download the output, open it in Excel.
- Notice you need one more column. Go back to step 3.
For a one-off question, that loop takes five to ten minutes. When you are trying to debug a reconciliation break or trying to figure out why a supplier's invoices are stuck, you run it fifty times a day. The tool is fighting you the whole way.
OTBI does not help much either. Subject areas hide the physical tables behind a semantic layer, so the moment your question crosses a boundary the model did not anticipate, you are back in BI Publisher.
I built Data Collage because I got tired of that ritual. Today I am taking the wraps off it.
What Data Collage is
Data Collage is a desktop SQL workbench for Oracle Fusion Cloud. You install it on your machine, add a connection to your Fusion pod using the same login you already use every day, and then you write SQL. Real SQL, in a real editor, with results in a real grid, in a couple of seconds.
There is no database password, no VPN, no agent installed anywhere, no DBA ticket, and no middleware sitting between you and Oracle. The app supports SSO.
It looks and feels like the SQL clients you have used for years, and it even gives you super powers with a feature we call directives. More on that later!
Read-only is enforced on your machine
Data Collage validates every statement on your machine before it leaves. Only SELECT and WITH are allowed. The validator rejects DML, DDL, PL/SQL blocks, and anything hiding behind a leading comment at the boundary, with a clear message. There is one chokepoint in the code for this, and no other path around it. I built the tool for people who need to look at production data, and I wanted it to be something an IT security team could sign off on without a long conversation.
The rest of the security posture follows the same instinct:
- Your Fusion password lives in the operating system's credential vault, never in a config file or a local database.
- If your company uses SSO, Data Collage captures a Fusion token through the normal login page and keeps it in memory only. It is gone when you close the app.
- All processing happens locally. Results, history, saved queries and metadata all live on your machine.
- The only outbound destinations are your Fusion pod, the license check, and the update server.
What it actually feels like to use
The mechanism is the clever part. The reason people keep it open all day is everything on top of it.
A proper editor. Monaco, the same engine behind VS Code, with syntax highlighting, multi-tab workspaces that survive restarts, formatting on one key, snippets that expand on Tab, and a bundled Oracle function reference on hover. Type an alias and a dot, and it offers you the columns of that table.
Metadata that is already there. Data Collage ships with more than eighteen thousand Fusion tables and views, close to a million columns, with descriptions, organized by module and searchable offline. You do not need to remember whether the column is INVOICE_ID or INVOICE_NUM, or which of the four _ALL tables you actually want. You search, you find it, you drag it into the editor.
A grid you can think in. Sort, filter, hide empty columns, find duplicates on any key, take the top N by a column, see the distinct values, and copy any column as a ready-made IN (...) clause for your next query. Format a column as currency or a date once and it stays that way.
Formula columns. Add a derived column with an expression, reference other formula columns from it, and it recalculates on every run. It uses SQL null semantics, not spreadsheet semantics, because the data came from SQL.
Drill back into Fusion. Right-click a cell, pick the entity, and the actual AP invoice, journal, purchase order or expense report opens in your browser. Make a column clickable once and every row in it becomes a link. At the moment we have a subset of links. More options coming soon.
Directives in the SQL itself. A handful of -- @ comments at the top of a query let you name the tab, hide or order columns, and set formats. Oracle sees comments. Data Collage sees the recipe. Because it is just SQL, it goes through code review and version control like anything else.
Saved analyses. SQL plus formulas plus formats plus column layout, saved as a single portable file you can hand to a colleague. They open it, point it at their own connection, and see exactly what you saw.
Bind variables, history, exports. Colon-prefixed placeholders prompt for values when you run. Every execution is logged locally with timing and row counts. Export goes to CSV, Excel, JSON, or pipe-delimited HDL for the HCM loaders, with long numeric IDs kept intact instead of mangled into scientific notation.
None of these is revolutionary on its own. Together, against a system that previously offered you a report builder and a download button, they change what a working day looks like.
Cross platform. Available for both Windows and Mac.
Who it is for
I built it for people like me: Fusion functional and technical consultants who live in the data during implementations, upgrades and support. It has turned out to be just as useful on the customer side, for the finance, supply chain and HR operations analysts who own the numbers and want to check them without raising a ticket, and for support teams who get asked "why did this happen" thirty times a week.
It is not a replacement for BI Publisher or OTBI as a reporting platform. Those are for reports that get scheduled and distributed. Data Collage is for the ninety percent of questions that never needed a report in the first place.
Why it has been quiet until now
I run a small consulting practice, and Data Collage started as an internal tool for our own engagements. For most of the last six months it has been used by a handful of people on real projects and real production pods, finding real problems. I wanted it to earn its keep on our own work before asking anyone else to trust it with theirs.
That period shaped it more than any roadmap could have. Half of what is in the grid today exists because someone on a project said "I keep doing this by hand." The enforced read-only rule, the credential handling, the fact that nothing leaves your machine except the query itself: all of that came from the questions IT teams asked before they would let a new tool near their environment.
It is now in the hands of its first customers and I am comfortable saying it out loud.
Where it is today
Data Collage is available now for Windows, as a signed installer with automatic updates, and as a per-machine MSI for teams that deploy through Intune or SCCM. A Mac build is available as well. For pricing, get in touch and we will size it to your team.
The roadmap, roughly in order: a version aimed at business users where the SQL disappears entirely behind a library of certified questions, GitHub integration, Routines that let you run a series of steps, and AI-assisted SQL grounded in the bundled metadata using your own API key.
If you spend your days inside Oracle Fusion and any of the above sounded familiar, I would love for you to try it. Details, documentation and access requests are at datacollage.app, or email datacollage@newarcinc.com and tell me about your environment.
And if you have your own version of the BI Publisher ritual, I would genuinely like to hear it. That is where the next feature usually comes from.
Arun Raj is the founder of Newarc Consulting, an Oracle ERP and EPM consultancy, and the creator of Data Collage.
This article was originally published on Medium.