Every source becomes a table
Databases, spreadsheets and systems become queryable tables — ready for queries, metrics and agents.
Synapse Data
Bluey reads your documentsand computes
A PDF and a spreadsheet do not answer the same way, which is why they take different roads in here: text becomes a passage with the page in view, and a table becomes a column you can sum, filter and join. You connect the source, the catalog builds itself, and a question in plain language turns into a real query with the SQL on screen.
- A spreadsheet never becomes loose text
- Every number arrives with the SQL that produced it
- You only see what you are allowed to see
From table to answered question, in one short video.
The catalog with each table's origin, the question the AI turns into a query, the result you check, and the certified metric everyone uses.
Read what the video shows
- Catalog & Data Lake: your business tables, with the source in view. Find the right table, check where it comes from and ask in your own words. Synapse writes the SQL.
- Your business numbers, one question away. In your words. The number comes from the table.
- Each table in the catalog, source in plain view. Rows, columns, source and the Certified badge.
- Columns, preview and history before you run a query. Data preview without spending a query; every sync on record.
- In your words, the SQL is ready. Synapse writes the SQL. The math comes from the table.
- The number and the query behind it. SELECT only: nothing changes in your systems.
- Save the result as a table. Personal, just for you. Corporate, for the team.
- One number, one definition. Certified metric, with the synonyms your team uses.
- Same definition, whatever you ask. Synapse answers with the certified metric.
- Fewer shadow spreadsheets. More verified numbers. Meet Synapse.
The same number, two ways.
Drag the line and see the difference between rebuilding the spreadsheet and simply asking.
The sources are many. The place where everything meets is one.
Databases
Documents and spreadsheets
Systems and APIs
Agents at work
Support
Analysis and decisions
A digital nervous system: you connect what your company already uses — through a ready-made connector, when one exists, and through the API, when one does not. In both directions: Synapse queries your company’s system, and your company’s system calls Synapse with a signed call.
The Data Lake is where your numbers come to live
When Bluey answers with a number from your company, it did not come from a similarity search or from a file someone attached a minute ago. It came from here.
- How it fills upYou connect the system you already use, through a ready-made connector or an API, and its tables start arriving on their own. The source stays in charge of the source, because synced tables are read and never changed in the system they came from.
- What it holds, and in what shapeA table stays a table, with its columns, its types and its rows. No spreadsheet is turned into loose text, and that is what keeps sums, exact filters and joins working when a question arrives.
- This is what gives a number its contextWithout the Data Lake the AI talks about the world, with market averages and general rules. With it the AI talks about your operation, using your definition of revenue, your regional split and your closing calendar.
- You do not visit it, you use itDay to day the Data Lake is not a screen you open. It holds up two that you do open, the Data Catalog to find the right table and the Data Lab to ask it a question in your own words.
Your sources become tables the company queries — with a count you can audit.
The Data module connects the company’s sources, lets anyone ask in plain language and returns the number with the page that proves it — certified metrics, permissions and the SQL in plain sight.
Ask in plain language
Your way, the SQL comes ready: Synapse writes the query and the number comes from the table, not from a guess.
The number with its source
The answer comes with the page that proves it: "R$ 480 per night" — and where it came from, in one click.
Certified metric
One number, one definition — a certified metric, with the synonyms the team uses.
Save as a table
Save the result as a table: Personal, only you see it; Corporate, the team uses it.
Permissions and governance
Every source and table with the permission it already has: the search returns only what your account can reach.
From the raw source to the number you take into the meeting.
Each part of the layer hands over to the next: the source you connect today becomes a table in the catalog, then a question in plain language, then a certified metric.
You connect, and the source becomes a table
The connection is always one of two things: a ready-made connector or your own API. You pick the account, pick what to sync and how often, and every sync materializes a dataset that agents and the console can query.
- Account connected, object chosen, sync scheduled
- From every fifteen minutes to once a day, or on demand
- Credentials encrypted and used only to sync your source
The file type picks the track
A text document goes in to be read and comes back as a passage with its page. A spreadsheet goes in to be queried and comes back as a column you can add up. Two different mechanisms because they are two different questions, and mixing them is what turns a number into a guess.
- PDF, DOCX, TXT and their kin go through meaning-based search
- CSV, XLS and XLSX go down the columnar track, never vectorized
- Search crosses both tracks and ranks everything by relevance
A question in plain language becomes a query
You write the question the way you would write it to an analyst, Bluey generates the SQL and the database runs it. The result comes back as a table and a chart, and the query that ran stays on screen so you can check the column and the filter before you use the number.
- Ask across the whole lake or focus on a single table
- Read-only on the source, with limits and an audit record
- The chart comes from the result and goes onto a dashboard
One number, one definition
Gross revenue and billing are usually the same thing said two ways, and that is exactly where two screens start to disagree. A certified metric stores the definition once, together with the words the team actually uses, and everyone starts getting the same number.
- The definition lives in certified SQL, not in someone's head
- Synonyms so each person asks with the word they use
- Lake layers keep the raw data apart from the certified data
What is yours stays yours
Not every analysis belongs to the company. What you upload and the tables you save as personal in the Data Lab live in a space only you can see, and when the analysis matures you publish it to a schema your team shares.
- A personal table invisible even to the workspace administrator
- Publishing is a choice you make at the moment you save
- Favorites and recents to get back to what you were looking at
Some of the sources you can already connect
The Data Lake does not start empty and does not start with a project. You connect what the company already has, and each connection becomes a queryable table on the other side.
Spreadsheets and files
- Google Sheets
- Google Drive
- SharePoint
- Amazon S3
Databases
- PostgreSQL
- MySQL
- MongoDB
- SQL Server
- Oracle
Sales and commerce
- HubSpot
- Shopify
- VTEX
- WooCommerce
- Stripe
ERP and finance
- TOTVS
- Omie
- Conta Azul
- Bling
- Senior
- QuickBooks
Marketing
- Google Ads
- Meta Ads
- Mailchimp
- Google Analytics
Everyday work
- Slack
- Notion
- Jira
- Trello
- Asana
- Zendesk
- Gmail
- Airtable
- Your own API
Either way, the synced table is read and never changed at the source.
From the first source to the first number, with no data project.
- 1
Connect the first source
A ready-made connector or your own API, with the sync already scheduled at whatever frequency fits.
- 2
Let the catalog build itself
Every table arrives with its columns, the rows it holds and the trust layer it lives in.
- 3
Ask in plain language
Bluey writes the SQL, the table returns the number and the query stays on screen for you to check.
The conversation starts with the data you already have.
Bring the spreadsheet that gives you the most trouble or the system nobody ever queries, and we will look together at what they become in here.
Data
What starts to exist in your space
- Ready-made connectors and your own API, with credentials encrypted and the source read-only
- A catalog with the columns, the rows and the provenance of every table
- Questions in plain language with the SQL on screen, and the number always coming from the table
- Certified metrics, so the same question always returns the same number
- Transformations without SQL that prepare a table and materialize it in the catalog
- Notebooks that hold text, queries and metrics in one living document
It does not start with a configuration screen: the first conversation is about your source, not about ours.
Why you can take this number into the meeting.
- A spreadsheet never becomes textCSV, XLS and XLSX go down the columnar track and never into similarity search. That is what keeps sums, exact filters and joins working, instead of turning a spreadsheet row into a loose paragraph.
- The number comes from the tableBluey writes the query, but the database is what calculates. The query runs read-only on the source, with limits and an audit record, and the SQL it used stays on screen for you to check.
- You only see what you are allowed to seeOn both tracks the permission comes before the result, and search returns only what your account can reach.
- Credentials stay encryptedThey are stored encrypted and used only to sync your source. A personal table you save in the Data Lab is invisible even to the workspace administrator.
On TVBluey, the company behind Synapse, on TECH SC by Globo NSC
Questions before you start
Does my spreadsheet become text for the AI?
No, and that is a rule rather than a preference. A tabular file goes down the columnar track, where a column stays a column and can be summed, filtered and joined. Vectorizing a spreadsheet row would destroy exactly that, so the product does not do it.
Where does the number in the answer come from?
From the table. You ask in plain language, Bluey writes the SQL, the database calculates and the screen shows the query that ran so you can check the column it summed and the filter it applied.
Do I have to replace my systems?
No. Adoption starts from the systems your company already runs, the connector available for each one and the sync frequency that makes sense for that source.
What if my system has no ready-made connector?
It comes in through the API. The connection is always one of the two, a ready-made connector or your own API, with credentials encrypted and the source read-only.
Who sees my tables?
It depends where you save them. A personal table is visible only to you, not even to the workspace administrator, and a corporate one is published to a schema your team shares.
Can I prepare the data without writing SQL?
You can. Transformations filter, compute fields, categorize, join datasets and aggregate step by step, and each one materializes the derived table in the catalog, ready to feed metrics, notebooks and agents.
What happens in the demo?
We start from a question your company asks every week and follow it all the way through: where the data comes from, which track it takes and how the number arrives with the query in view.
How do we assess the investment?
The proposal considers the scope of the platform, the AI usage and the support for adoption, with the requirements assessed before anything is signed.
The data is already there. It just has to answer.
Point at the first source, let the catalog build itself and ask the question that costs you a whole afternoon of spreadsheet work today.
Talk to us about your data