Apache Superset, built and run by people who commit to the project
Migrating off licensed BI, starting analytics from scratch, building the charts Superset lacks, embedding dashboards inside your product, or running Superset on Kubernetes. We upstream fixes to Apache Superset and we run it in production ourselves.
- Superset licence
- $0, forever
- Dashboards after tuning
- 5x faster
- One logistics customer
- 15s to under 3s
- Fixes we write
- Upstreamed to the project
- Median to first production artefact
- Six weeks
Find your situation
Most Superset conversations start in one of six places, and the right first move is different in each.
Coming off Tableau, Looker or Power BI
The per-seat bill is usually what starts it. We rebuild the dashboards that matter to parity, then the licences get switched off rather than quietly renewed.
Read the migration case →StartingNew to analytics entirely
No BI tool yet. We stand Superset up on your warehouse and build the first dashboards that carry decisions, which is a smaller list than most teams expect.
See data engineering →ChartsThe chart you need is not in Superset
Waterfall, mekko, whale and advanced tabular, the views finance and FP&A teams ask for by name. We write the plugin and upstream the fix to the project.
Ask about a plugin →EmbeddingDashboards inside your own product
White-label theming, row-level security so each tenant sees only their data, SSO and guest tokens. Your customers see your product, not someone else's tool.
See embedded work →ScalingK8sRunning it properly in production
Superset on Kubernetes with high availability, caching, performance tuning, security hardening and monitoring, run for you with response times in writing.
Ask about managed support →ProofSuperset inside our own product
We build Glimvia, our own analytics product, on Apache Superset in production. The patterns we hand you are ones we already live with.
See Glimvia →How a slow dashboard gets fast
Performance is not a tuning pass at the end, it is the reason anyone opens the dashboard twice.
- Profile
Find the queries that actually hurt
Not the ones people complain about, the ones the database spends its time on. These are frequently the same query fired by several charts on one page.
- Cache
Put the right caching in the right place
Caching everything is as unhelpful as caching nothing. What gets cached, and for how long, follows how fresh the number needs to be.
- Reshape
Change the schema, not just the SQL
Some questions are slow because the tables were never modelled to answer them. Pre-aggregated tables do more for load time than any query rewrite.
- Pair
Add ClickHouse where it earns its place
For heavy scans we often put ClickHouse behind Superset. It is a choice made on evidence from the profiling step, not a default we ship to everyone.
- Adopt
Watch whether people come back
The real test is not the benchmark. At one logistics customer the whole ops floor started using the dashboard once the wait fell below three seconds.
What that did to one dashboard
A logistics customer, before and after ClickHouse and pre-aggregation.
What teams worry about, and what happens instead
The five objections that come up before a migration, each answered with something we have already done.
- The licence bill grows with every viewerSuperset costs $0 in licence at 10 viewers or 1,000
- Charts our team depends on are not in SupersetWe build the plugin and upstream the fix
- Different roles must see different dataThree separate role-based views at one healthcare customer
- Somebody has to run it in productionKubernetes, monitoring and response times in writing
- Dashboards are too slow for anyone to use15 seconds to under 3 at one logistics customer
What contributing upstream actually changes
The difference shows up when something in Superset is wrong and you need it fixed rather than worked around.
We ship code into Superset itself
Horizontal waterfalls, axis-label wrapping, cross-filtering and legend-colour fixes have gone upstream from us into the project.
Your build sits on the project, not a fork
Fixes land in Superset and come back to you with the next release. Nothing we write for you turns into a private branch somebody has to maintain forever.
Missing charts become plugins
Waterfall, mekko, whale and advanced tabular are the ones asked for most. They get built as plugins and then offered back to the project.
We run Superset ourselves, in production
Our own analytics product, Glimvia, runs on Apache Superset. What we recommend has already been through our own on-call rota.
The three numbers this page keeps coming back to
One about cost, one about speed, one about how long it takes to have something real.
How the engagement runs
The same shape as every other engagement here, applied to Superset.
- Week 1
Fixed-fee audit
Two calls and a one-pager you keep either way. What you run today, what the licences cost, and which dashboards genuinely carry decisions.
- Weeks 2-6
Build
A pod of three. Superset stood up on your warehouse, the dashboards that matter rebuilt to parity, plugins written where Superset lacks the chart.
- Week 7 onward
Operate
Production on Kubernetes with high availability, caching, performance tuning, security hardening and monitoring, run for you with response times in writing.
- Median
Six weeks to the first production-grade artefact
Across the 50+ projects we have delivered for 20+ clients since starting in Pune in 2023. It is a median, not a promise for your case.
Asking Superset questions in plain language
Text-to-SQL on Claude, so people who do not write SQL can ask and get a chart back.
- Production path
Sonnet
The default for questions people act on. Better at holding your schema and the definitions your business agreed when a question is phrased loosely.
- Cost-sensitive long tail
Haiku
For the high volume of simple, repetitive questions where a larger model buys you nothing but a bigger bill.
The 50th viewer and the 500th viewer
Per-seat BI charges for every dot on the right, and Superset charges for none of them.
Both cost $0 in licence.
When Superset is the wrong answer
There are four cases where we would tell you not to do this, and week one is the cheapest time to hear it.
Your viewer count is small
The economics of Superset come from viewers costing nothing to add. With a handful of seats, the licence was never the expensive part, and a migration will not pay itself back.
Nobody is going to own the platform
Open source with no licence still needs somebody running it in production, either us or your team. If neither is funded, a licensed tool with a support line is the safer purchase.
The dashboards are not the problem
If the numbers are wrong or arrive late, a new BI tool renders the same problem in a different font. That is a data engineering job first.
You need chart parity by next week
Custom plugins take real time to build and test. If a hard date is what your migration hangs on, start the conversation early or do not start it yet.
Receipts from real Superset work
Three engagements, three different reasons Superset was the right call.
The questions buyers actually ask
Five that come up in nearly every first call about Superset.
Can Superset really replace Tableau, Looker or Power BI?
For the large majority of dashboards, yes. Where it cannot, the gap is usually a specific chart type rather than the platform. We build those as plugins, waterfall, mekko, whale and advanced tabular among them, and upstream the fix, so capability is not what you give up.
What does Superset actually cost?
The licence is $0 because it is open source. You pay for hosting and for our build, once. Adding the 50th or the 500th viewer costs nothing in licence fees, and that is precisely where per-seat BI quietly drains a budget.
Can you deploy and run it on Kubernetes?
Yes. Production deployments on Kubernetes with high availability, caching, performance tuning, security hardening and monitoring, run for you with response times agreed in writing. If you would rather your own team ran it, we will hand it over set up that way instead.
Can you embed Superset in our product?
Yes. White-label theming so it looks like your product, row-level security so each tenant sees only their own data, SSO and guest tokens. There is more detail on the embedded analytics page.
Are you Superset experts or Superset resellers?
Contributors. We have upstreamed fixes into Apache Superset, including horizontal waterfalls, axis-label wrapping, cross-filtering and legend-colour fixes, and we run Superset in production inside our own product, Glimvia. On the AI side we are an Anthropic Build Partner and joined the Claude Partner Network at launch.
The rest of the practice
These are genuinely different jobs with different ways of failing. Most engagements start in one of them.
- Data engineeringJobs that finish, migrations that reconcile, and storage that stops paying for cold data.
- Data integration and governanceOne set of records the finance team and the operations team both accept.
- Data platforms and modernisationOff the platform you outgrew, without a twelve-month freeze on new reporting.
- Command centresThe one screen an operations floor runs the day from, not another dashboard.
- AI audit and roadmapOne week, and you know which two or three AI initiatives are worth building.
- AI evaluationA graded eval set, because an AI system cannot simply pass or fail a test suite.
- AI governanceGovernance you can defend in a board meeting and audit on demand.
- AI agentsOne reasoning agent on Claude, on your live data, with evidence and a kill switch.
- Data agentsAgents that watch the data, flag what moved, and explain what is driving it.
- AI use case, guaranteedOne production agent in eight weeks, or you do not pay for the build.
- Applications and automationThe system your team works in all day, built or replaced in slices.
- Application modernisationThe system nobody wants to touch, replaced a slice at a time rather than rewritten.
- System integrationSystems that stop disagreeing about the same customer, order and item.
- Process digitisationThe process that still runs on paper, WhatsApp and one shared spreadsheet.
Start with the audit and find out whether Superset is your answer
Two calls, a fixed fee, and a one-pager you keep either way: what your BI estate costs to run, which dashboards carry decisions, and whether Superset pays for itself here. If it does not, we will say so in week one.