We work like engineers, not a slide deck. Fixed scope. Clear definitions. Matching you can audit. Outputs your team can use without a reconciliation meeting first.
Founder-led delivery
Lucendata is led by Francisco Carvalho. The work is scoped in writing, shipped in weeks, and handed over with docs and runbooks so your team is not dependent on heroics.
What we actually build
Lucendata connects the systems that hold part of the story, cleans and matches the records that should line up, and delivers one output a team can actually use. Sometimes that output is a reconciled commercial view. Sometimes it is a duplicate-matching workflow. Sometimes it is the reporting layer behind a decision that keeps getting blocked by bad source data.
- Multi-source data pipelines for operational reporting
- Entity matching and duplicate-resolution workflows
- Reconciled views across CRM, finance, ops, and support systems
- Decision-support layers built on top of cleaner underlying data
Why the work is scoped this way
Most teams do not need a grand data programme as the first move. They need proof that one painful workflow can become reliable enough to use. That is why Lucendata starts with the Mini Proof-of-Work: a 7-day sprint around one narrow, commercially relevant use case.
If the proof is there, the next build is clearer, safer, and faster. If it is not, the client learns that before funding a larger implementation. That is a better buying process and a better delivery process.
The goal is not to make the data estate sound impressive. The goal is to make one decision path trustworthy enough that the business can use it without a reconciliation meeting first.
How we think about the build path
- 01Mini Proof-of-Work: prove one narrow use case on real data.
- 02Core build: productionise the connected inputs, matching logic, and operational output.
- 03Intelligence layer: add retrieval, scoring, or decision support once the foundation is stable.
That sequence matters. Fancy outputs on broken inputs just create polished confusion. We would rather prove one real workflow properly than promise ten vague ones badly.
Who this tends to help most
Usually mid-sized teams with multiple operational systems, recurring reporting disputes, and real pressure to move faster without guessing. Sales and finance no longer trust the same number. Duplicate customer records are creating commercial friction. Dashboards exist, but nobody believes the source data behind them. That is the territory.