# How Brick Traded Manual Data Ingestion for a Data-Driven Operation > Brick centralized data ingestion with Erathos and now maintains its entire data infrastructure with a single analyst, without needing a dedicated pipeline engineering team. Source: https://www.erathos.com/en/clientes/brick-software Em português: https://www.erathos.com/clientes/brick-software ## About - Brick - Sector: SaaS / Insurtech - Location: Curitiba, Paraná - Employees: 25 - Erathos customer since: October/2024 - Data stack: Erathos (data ingestion), Google BigQuery (data warehouse), Metabase (data visualization) - Connected services: HubSpot, Conta Azul, Linear, GitHub, PostHog (beta) Brick is the AI-native operating system for risk decisions at insurance companies, giving business teams the autonomy to create, adjust, and audit underwriting, claims, and fraud-prevention workflows without depending on IT. Brick is the AI-native operating system for risk decisions at insurance companies, used by underwriting, claims, and fraud prevention teams. Before Erathos, data ingestion was maintained by hand by the company's co-founder, with Airflow orchestrating extractions and DBT running on top with no guarantee the data was actually up to date. Any schema change in an API would break the pipeline, and fixing it ate into real time that should have gone toward building product. After centralizing ingestion with Erathos, Brick now maintains its entire data infrastructure with a single analyst, and cut the time to bring a new source online from months to days, while supporting product, operations, and finance decisions with reliable data. ## The results - **30×** — faster to make new data available - **97%** — less time to generate the Investor Report - **78× ROI** — R$783K in net savings over 21 months ## The challenge ## The challenge Before Erathos, Brick ran its own data ingestion with Airflow triggering routines that dumped data into Cloud Storage and then into a delta table in BigQuery, with DBT running on top regardless of whether the data was actually up to date. Co-founder Carlos Schwabe describes the result as a basic ingestion process, built with the everyday tools of any engineering team: the customers table would update, the sales table wouldn't, and the final report would show a sale tied to a customer that didn't exist. The biggest cost wasn't running the pipeline — it was maintaining it. Each data source (HubSpot, Conta Azul, Zendesk, the production database) had a different API, some documented, some not, and every scope change — like needing call data instead of just contacts — forced the team to reopen the API docs and rebuild the cursor logic from scratch. About twice a week, Carlos had to drop what he was doing to fix something that broke and make sure the numbers lined up again. On the finance side, the problem showed up as delayed decisions. CFO Thiago Paz remembers a night the team stayed at the office until 10pm trying to calculate churn rate, because the data was scattered across the ERP, the production database, and the CRM, with no way to cross-reference it without manually building everything in a spreadsheet. At its core, Brick didn't have a tooling problem — it had a time problem: every hour spent maintaining data pipelines was an hour not spent on product and strategic decisions, right when the company most needed speed. ## The solution ## The solution Brick found Erathos through a referral. Carlos wanted to solve the problem quickly and within budget, without time to deeply evaluate many alternatives. As soon as the first connection was set up, the difference was clear: instead of mapping endpoints and maintaining their own scripts, they just plugged in the connector and the data arrived up to date, while Carlos kept maintaining, on his own, whichever sources Erathos didn't cover yet at the time. That transition period made it clear to the team how much simpler and more reliable ready-to-use, plug-and-play connectors were compared to maintaining their own pipelines line by line. With every new source Erathos added to its catalog, Brick eliminated one more front of internal maintenance, to the point where, according to Thiago, there's no system used by the company today without data available via Erathos. ## The results ## The results - Time to bring a new data source online dropped from **2–3 months to 2–3 days** with Erathos, according to Bruno Rosso. - The monthly Investor Report stopped taking a whole morning of work and now takes **7 minutes**. - Monthly financial close dropped from a day and a half to a single morning. - Calculating metrics like churn, which once required a full night of manual work, is now available to query and update at any time, with no manual data-gathering required. - Co-founder Carlos went from spending 4–5 hours a week maintaining pipelines to spending zero. - The company's entire data infrastructure is now maintained with a **much leaner investment**, supported by a single data analyst. Beyond the time savings, these numbers reflect what Brick set out to do from the start: build a genuinely data-driven culture, where product, operations, and finance decisions are based on reliable data accessible to the whole team — not just one technical person. ## Return on investment (ROI) ## Return on investment (ROI) Brick's team estimated that, without Erathos, they would have needed to build an internal team of 2–3 people — data engineers and data analysts — to sustain the same operation. Using average market salaries in Brazil (R$8,000 for a data engineer and R$5,000 for a data analyst, plus mandatory Brazilian labor charges), the estimated cost of a dedicated team of 2 engineers and 1 analyst would have been R$793,800 over the period. _Cost avoided over the last 21 months (2 data engineers + 1 data analyst, average market salaries + mandatory labor charges): R$783,751_ **R$ 783,751 in net savings, approximately 78x the amount invested in Erathos** ## What's next ## What's next Brick continues to explore the potential of the data infrastructure built with Erathos. On the connector side, Conta Simples and PostHog are in beta and should be tested by the team soon. Bruno also signaled interest in going deeper on the Pro plan, with custom job retries and custom backfill for the larger tables, plus turning on failure alerts via email or Slack — a feature he didn't know was available. > If I didn't have this infrastructure, I'd need two or three people to do what Erathos does today. I was able to generate real value from my data with a much leaner setup than I thought I'd need. That's the core of building a data-driven culture: the team only uses data when they can trust it. > > — Thiago Paz, CFO > We used to have a lot of rework, and now we don't. It's more efficient to work this way. If it's not your core business, hire Erathos. Spend your time modeling your own domain, not someone else's. > > — Carlos Schwabe, Co-Founder & CDO > Every new source implementation, if I had to do it myself, would take about two or three months. With Erathos, it's done in a few hours. I don't need the skills of a data engineer to generate real value from the company's data. > > — Bruno Rosso, Data Analyst