The World Bank Replaced Tableau With Software It Owns
A Tableau alternative at World Bank scale: workflows too specific to buy, dashboards the team built and owns instead, deployed in the Bank's own cloud.
8
Hours to rebuild an app with the AI App Builder that originally took two weeks
100+
Internal World Bank staff using the tools the team built
1,000,000+
Rows served live in the dashboards that replaced the old reporting tools
- Founded
- Washington D.C., 1944
- Website
- worldbank.org
Our work doesn't fit cleanly into any off-the-shelf product. Our users pull data from too many systems, with workflows too specific to buy. So we build the platform ourselves with a lot of custom development to fit how they actually work.
Meet the World Bank, a global development institution founded in Washington D.C. in 1944, supporting projects across 189 member countries and managing some of the most extensive development data in the world.
Inside the Bank, an internal engineering team uses Reflex to build the kinds of tools staff rely on every day: chatbot-style interfaces that let internal audiences query institutional knowledge, profiling tools that help review suppliers involved in projects financed by the World Bank, and data-review dashboards that pull from many internal systems. Because the team works with strictly confidential data, those tools do not fit cleanly into off-the-shelf SaaS, which is what pushed the team toward Reflex.
The Tableau dashboards the Bank used to buy
Historically, analysts and researchers across the World Bank relied on what most organizations buy for this work: Tableau licenses for reporting, with manually built dashboards filling the gaps.
Tableau handled static reporting, but the work did not fit the license. It locked analysts into a fragmented loop: prepare datasets offline, upload them, share dashboards by hand. Anything genuinely interactive required escalation to specialized engineering teams, creating multi-month delays for tools the team's own engineers could have written if the front-end gap had not been there. Teams comparing Tableau alternatives will recognize the shape of the problem: the product works as designed, and the work does not fit the design.
Buying a replacement was no faster. At the World Bank, every new piece of software carries its own price of admission:
When we go through the security process, it will take 6 to 7 months just to get the accreditation and to get the purchase order and everything. World Bank's accreditation process is very rigorous and very lengthy.
What changed at the Bank, side by side:
| The dashboards the Bank bought | The software the Bank built | |
|---|---|---|
| Workflow fit | The process adapts to the product | Built around how the team works |
| Where the data lives | Uploaded into the tool | Stays in the Bank's own cloud |
| Cost shape | Per-seat licenses, renewed every year | An asset the Bank owns |
| Time to a new tool | Months of escalation and accreditation | Hours with the AI App Builder |
| Who builds it | Specialized engineering teams | The team that understands the work |
For how to run this decision on your own tools, see our build vs. buy framework for internal software. For the build side of the same decision at hardware scale, see how Dell built the platform it could not buy.
The cost of building the conventional way
Everything behind the tools already existed. The services that hold the data and do the heavy analysis were the team's own, running inside the Bank's cloud and already cleared by its security model. What was missing was the interface, and building one the conventional way meant a second technology stack to maintain, new dependencies to take through internal procurement, and a skill set the team did not have and was not going to hire.
Continuing to escalate prototypes into ground-up rebuilds was a structural tax on the team.
What the team built instead
The team adopted Reflex for two distinct workloads. In both cases, the first version was the production version, built on systems the Bank already ran.
Chatbot-style interface for internal staff
Pooja built an internal chatbot-style interface that lets World Bank staff query a repository of internal documents in natural language.
We give the AI access to a variety of different types of internal documents, and it answers users' questions directly from them.
The app runs inside the World Bank's own cloud environment, structured to fit the team's security model. Over 100 internal staff use the deployed tool.
The first version of the chat assistant, an earlier prototype the team had built, was rebuilt in Reflex in a single working day using the Reflex AI App Builder:
With the AI App Builder, I converted our entire app to Reflex in 8 to 9 hours, every feature ready for production, in a single working day.
The app reads from a database holding millions of records, writes AI-generated summaries, and produces finished reports, exported to PDF and Excel with relationship graphs included.
Live data-review dashboards
In parallel, Thad has been building a data-review interface where users ask questions in plain language and the answer appears as a live, filterable table. It serves tables of more than 1,000,000 rows at once and supports the advanced filtering the team's previous reporting tools could not.
Why the World Bank chose to build
The team evaluated several paths to a production tool and chose Reflex because the first version is the production version: no prototype thrown away, no rebuild, no structural delay between an idea and a tool staff can use.
It also meant shipping polished applications without staffing a front-end specialist, and reusing everything the Bank already ran: the same data, the same cloud, the same security model, with no parallel stack to maintain. The people who understood the work built the tools, and the Bank owns the result.
How it was built
This section is written for your technical team. In short: nothing the Bank already ran needed to be replaced.
View the technical details
The team's backend already ran on Python: FastAPI services, SQLModel-backed data layers, and Databricks for the heavy data work, inside an Azure-hosted stack tied into the Bank's security model. Reflex applications sit directly on top of that stack via the Reflex Enterprise Package. Across both workloads, the team was able to:
- Reuse their existing Python backend services (FastAPI, SQLModel, Databricks) without modification
- Build full browser UIs for chat, AG Grid tables, document search, and report generation in pure Python
- Deploy into the team's Azure App Service Environment with split front-end and back-end services to fit existing security requirements
- Generate AI summaries, PDF and Excel exports, and embedded relationship graphs from inside the Reflex app
- Serve datasets of more than 1,000,000 rows via AG Grid's server-side row model, with multi-AND/multi-OR filtering the team's prior MUI-based tooling did not support, connected live to existing FastAPI endpoints through Reflex's
api_transformer
Pooja has a ton of experience with ML models and Python, but no experience with JavaScript. And so it just really flattens the possibility space so that she doesn't have to worry about the front-end.
Reflex gave us the flexibility of React without having to actually write React. That's huge for teams like ours.
From pilot to standard practice
Pooja's chatbot-style interface is deployed in the World Bank's Azure environment with 100+ internal staff actively using it. Thad's dashboards are running on real internal data and are being expanded to replace the team's older reporting tools as well, the second set of dashboards Reflex has replaced at the Bank.
She was able to get two additional apps up into production, so into usage. So it's going very well.
What started as one team's decision to replace bought dashboards with software it owns is becoming the standard way the team ships internal tools at the World Bank.
We should have them go straight into Reflex and deliver their initial product with something like Reflex.
Frequently asked questions about Tableau alternatives
What is a good alternative to Tableau for internal dashboards?
That depends on why Tableau is not fitting. When the limits are workflow fit, data confidentiality, or per-seat licensing, the real alternative to one BI license is not another BI license: it is a dashboard your team builds and owns. The World Bank replaced Tableau dashboards with applications its own engineers built in Reflex, connected directly to internal systems and deployed in the Bank's own Azure environment.
Should we choose Power BI or Tableau?
If a standard BI product fits how your teams work, either can serve you well. The question to ask first is whether you are comparing two licenses when the work is too specific to buy. The World Bank concluded its workflows did not fit off-the-shelf products and built its dashboards instead, as software the team owns and extends.
How does the cost of Tableau compare to building your own dashboards?
BI platforms are licensed per user, so the cost grows with every viewer and recurs every year. Software you build is an asset you own, and AI-assisted development has changed what building costs: at the World Bank, an internal app that originally took two weeks to build was rebuilt as a production application in 8 to 9 hours with the Reflex AI App Builder.
What are the limitations of Tableau for interactive internal tools?
In the World Bank team's experience, Tableau handled static reporting but forced a fragmented loop: preparing datasets offline, uploading them, and manually sharing dashboards. Anything genuinely interactive meant escalating to specialized engineering teams, with delays of months. Those limits are what pushed the team to build instead.
Can a team without frontend engineers build production dashboards?
Yes. The World Bank team ships production applications with the Reflex AI App Builder, including a chat-driven document interface and live data-review dashboards, deployed in the Bank's own cloud environment without staffing a frontend specialist.
See what other teams built
See what your team could build.
Tell us about your workflow. See how Reflex can help you turn it into an application.