↑
Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research Case StudyMarket Research
dichter & neira • LATAM

Ask Your Data, Securely

dichter & neira, a leading market research firm, needed answers for its clients from huge retail datasets without waiting on a technical team. SunnyData added an AI assistant on Databricks, so users ask in plain language and every client sees only its own data.

See client results
12x Faster data workflows From 2 hours down to 10 minutes.
65% Faster data checks Compared with the original manual estimate.
20 to 5 Pipelines to manage 75% fewer, now run inside Databricks.
AI Agent enabled Ask questions in plain language, even on massive retail datasets.

The Project in 80 Seconds

From static dashboards to answers in plain language

Client Challenge

Simple questions, slow answers

dichter & neira is a leading market research firm in Latin America. Its clients, major consumer brands, get product, stock and competitor data through custom dashboards.

Those dashboards were static. To dig deeper, business users had to ask a technical team, and answers took hours or even days.

The platform held the firm back too. An older data catalog blocked Databricks' newest AI tools, and every client's data had to stay strictly separate.

An analyst at a desk looks at a monitor showing a business dashboard with sales charts

The Solution

Plain-language answers, with every client's data kept apart

SunnyData upgraded the foundation first, then put an AI assistant on top.

Modernize the foundation

About 400 tables moved to Unity Catalog, unlocking Databricks' newest AI tools. SunnyData's Data Validator Accelerator checked each one, 65% faster than manual checks.

20 → 5

Simplify the pipelines

20 data pipelines became 5, now run inside Databricks.

Protect every client

Row and column-level rules make sure each brand sees only its own data, and competitors stay anonymous.

Ask in plain language

AI/BI Genie lets business users question the data directly. A custom AI agent learns each client's jargon, adds context and suggests metrics people might overlook.

What it runs on
Unity Catalog logo
Unity Catalog One place to govern data and control who sees what.
AI/BI Genie Questions in plain language, answers from live data.
Delta Lake logo
Delta Lake Reliable, fast storage for the data.
Lakeflow logo
Databricks Workflows Runs the data pipelines in one place.

Genie at Work

One question, one private answer

A user asks a question in plain language. Genie answers from live data, and competitors stay anonymous.

Follow one question through

Four stages, every client protected

QoQquarter over quarter Filtered to your data
Your brand
Brand AlphaCompetitor (anonymized)
Brand BetaCompetitor (anonymized)
Brand GammaCompetitor (anonymized)
Market share20%
Growth vs. last quarter+10%
Top productProduct A
Suggested next metric

Illustrative representation. Not client data.

1AskA question in plain language.
2UnderstandThe agent reads industry jargon and acronyms.
3ProtectThe view is limited to your data. Competitors are anonymized.
4AnswerA private result, plus a suggested next metric.
1
Ask
A question in plain language.
2
Understand
The agent reads industry jargon and acronyms.
3
Protect
The view is limited to your data. Competitors are anonymized.
4
Answer
A private result, plus a suggested next metric.

Illustrative sequence in a client's own analytics portal. Ask: the question "Which products grew fastest last quarter?" is typed into the Genie ask bar. Understand: a chip shows the agent reading the abbreviation QoQ as quarter over quarter. Protect: a "Filtered to your data" chip appears with a lock, and the three competitor names in the chart are redacted and replaced with "Competitor (anonymized)". Answer: the bar for your brand and the key figures appear, with fictional values: market share 20%, growth versus last quarter +10%, top product Product A. A "Suggested next metric" prompt appears under the chart. Not client data.

Key Benefits Achieved

Business results, before and after

MetricBeforeAfterChange
Getting an answer Request to IT or business teams, hours or days Ask directly in plain language Self-service
Data workflow time 2 hours 10 minutes 12x faster
Pipelines to manage 20 5 75% fewer
Checking ~400 tables Manual conversion and QA Automated validation 65% faster
Client data access Coarser controls on an older catalog Row and column-level rules per client More secure
AI know-how Reliant on outside consultants In-house team can build it Independent

What They Can Do Now

A data team that builds on its own

Answers without a ticket

Business users explore the data themselves, whenever they need to.

More AI, across the business

Thanks to knowledge transfer during the project, the internal team can build similar AI solutions for other business units without outside help.

Ready for what's next

The upgrade opened the door to Databricks' newest AI features, like Mosaic AI, when the firm wants them.

Ready to give your teams answers on demand, securely?

SunnyData prepares your platform, protects who sees what and puts Genie to work, so business users get answers without waiting on a technical team.

Portrait of Harrison Johnson
Harrison Johnson Account Executive, Retail & Manufacturing
Are your teams still waiting on reports to get answers?