Everything your company knows,ready for anyone to ask.
The contracts, policies, manuals and spreadsheets your company already has become readable by the AI without ever stopping being yours, because every answer shows the file and the page it came from and the folder permission still holds inside the conversation.
From files to an answer with its source, in one short video.
Files and systems flowing into one base, approval before indexing, the folder that decides who sees what, and the answer with the page that proves it.
Watch · 55 s · 29 sIllustrative view of the product, with example data. The video has no sound: the captions are on screen.Read what the video shows
Knowledge Base: files and systems, ready to answer. Upload files, connect your systems and ask. Synapse answers with what the company knows, and shows where it came from.
Your files. Your systems. All in one base. Everything the company knows, ready to answer.
Drag in files or whole folders. PDF, Word, spreadsheets, e-mails: each in its topic's folder.
You decide what Synapse uses. Approved, indexed, and Synapse even writes a summary.
Drive, SharePoint and your systems. The Base updates itself with every sync.
Every folder picks who sees it. Whoever can't open the folder gets none of it in the answer.
The answer, with the page as proof. R$480 a night, and where it came from, one click away.
Sheets become tables: sum, filter, compare. Synapse reads documents and queries tables.
Less searching. More answers. Meet Synapse.
The same material, two ways.
Drag the line and compare today, which is hunting, with afterwards, which is asking.
Can we end the haulier contract without a penalty?
You can, with sixty days' written notice. Once the notice is served there is no termination penalty, and loads already in transit run through to delivery.
The road your company's material travels before it becomes an answer.
Every step happens on a real screen, under the name it carries in the Synapse menu, and the demo follows along as you read.
Knowledge Base
Start from the collection you already have
Drag files one by one or a whole folder from your computer and keep the folder tree in the shape the company already uses. Whoever connects their own Google Drive or SharePoint account gets a folder in their name and keeps deciding who sees what is inside it.
PDF, Word, PowerPoint, text, markdown, HTML, RTF, ODT, e-mail, JSON and XML
Meeting recordings, audio and video become searchable text
You connect it with a ready-made connector or through the API, and whoever owns the content stays the owner
Review queue
Nothing is quoted before someone approves it
Every file arrives as Awaiting approval and only counts once someone with permission asks to approve and index it, when it moves through Indexing and ends up Indexed. Who approves is set by permission, separate from who uses the chat, and the review queue shows in the menu how much is still waiting.
Five states on screen: awaiting, indexing, indexed, rejected and failed
Bulk approval, ticking several documents at once
A document left out of approval exists in the list and does not exist for the answers
Knowledge Base
Every folder picks who gets in
The restriction is by user, role, group or tag, and it does not stop at the screen: whoever cannot open the folder does not receive its content through the chat either. With no restriction at all, the screen says in plain words that the folder is open to anyone in the company.
Restriction by user, role, group or tag
An assistant tied to one folder quotes nothing outside it
Locking a document cuts its live shares on the spot
Chat
The answer arrives with the proof beside it
Whoever asks writes in plain words and gets, under the answer, the Sources block: how many files support it, the name of each one and the page. One click opens the document right there, already on the page it cited, and when the subject is not in the collection the answer says so instead of filling the gap.
The source opens the document itself, on the page it cited
When the subject is not in the collection, the answer says so and asks where it lives
No AI is infallible, which is exactly why the proof comes with it
Chat
A spreadsheet is added up, not summarized
CSV, XLSX and XLS do not come in as text: they become a real queryable table, and the quarter's total is summed over the rows of the file rather than estimated from an excerpt. That is the difference between finding a passage and working out a result, and the platform treats the two as different things.
Plain questions over CSV, XLSX and XLS
The whole file loads, or the upload fails and says why
The spreadsheet opens as a spreadsheet, tabs along the bottom
From the first file to the first answer you can check.
1
Bring what the company already has
Drag the folders from your computer, or let whoever owns the content connect their own account.
2
Decide who sees what
Folder by folder, by user, role, group or tag. The same rule holds later, inside the answers.
3
Approve, and let the team ask
Whatever clears approval can be quoted. The team asks in plain words and checks the source before trusting it.
A layer of the platform, not a product on the side.
The knowledge base comes with the rest of Synapse, so the same approved material answers in the chat, feeds the agents and backs the document that goes out.
Platform layer
Your knowledge base
For the company that runs on documents
Answers carrying the document and the page behind them
A review queue: nothing is quoted before someone approves
Folder permissions that hold inside the answers too
Spreadsheets queried as tables, summed row by row
Documents ready in PDF, Word, Excel, PowerPoint and HTML
Platform, manual and help in Portuguese, English and Spanish
The conversation starts from your own material: a real folder, a real question, and the source on screen.
The material stays your company's.
The source comes with itDocument, page and excerpt under every answer, and when the subject is not in the collection it says so instead of inventing.
Everyone sees what they mayThe folder restriction holds in the answer. Whoever cannot open the folder does not receive its content through the chat.
Nobody publishes aloneAn uploaded file is quoted by no agent until someone with permission approves it.
It stays in your company's spaceEvery search carries your space's scope before any filter of yours, and nothing you upload is used to train AI models.
On TVBluey, the company behind Synapse, on Globo NSC's TECH SC
Questions before we talk
What happens in a demonstration?
We start from one of your processes and show your own material going into the base, through approval, and coming back as an answer with the source beside it. What comes out of it is the map of the sources and integrations the case needs.
Do I have to approve every document?
Yes. Every file arrives as Awaiting approval and is only quoted after approve and index. You can approve several at once by ticking the list, and the review queue shows what is still waiting. This is the number one reason the platform cannot find a file that is already there.
Can I hide material from part of the team?
You can. The restriction is per folder and per document, set by user, role, group or tag, and the same rule holds in the answers. With no restriction, the screen warns that the folder is open to anyone in the company.
Do spreadsheets really work, or is it text reading?
They work as tables: the total is summed over the rows of the file, not estimated from an excerpt. The spreadsheet loads whole, or the upload fails and says why, instead of importing half of it. The same content pasted inside a PDF, though, is text, and text does not add up.
Can I correct a document in place?
Only text, markdown and HTML are editable on screen, and the platform relearns the new content by itself. PDF, Word, PowerPoint and spreadsheets you download, edit in your own program and upload again. And worth knowing up front: there is no bin and no version history, so deleting a document is final.
Do I need to replace my systems?
No. The material stays where it is and comes in through a ready-made connector or the API, and the conversation about adoption starts from the systems the company already uses and the synchronisation each source needs.
How is the investment evaluated?
It starts from the scope of the platform, the AI usage and the support for adoption, with the requirements reviewed before anything is signed. The How to purchase page explains the model.
Will my documents train another company's AI?
No. The material stays isolated in your company's space and is not used to train AI models.
What your company knows is waiting for a question.
Bring a real folder and a real question, and we will watch your own material answer with the source on screen.
