Project Chat: ask your project with Gauss.
You ask your question in plain language. Gauss answers from your project's measured data – not from the internet. Every number in an answer comes from a query.
Project Chat
brandzaps.ai · Project 1
Between a general AI chat and a dashboard.
Why has a metric dropped? Which topics are missing? You can't read answers to questions like these from a single dashboard card – and a general AI chat doesn't reliably know your project.
Produces plausible numbers that don't come from your data.
no project contextOnly answers the questions someone has built a view for.
limited questionsMakes the project context queryable and traces every answer back to the measured data.
evidenced project statementGauss works with tools, not guesswork.
Gauss has read access to the data of its own project. A built-in handbook tells the model which paths and tools are available. This lets Gauss look things up instead of deriving a number from a general assumption.
Every answer states which data it relies on. Evidence that doesn't exist in the project is discarded.
Read access to its own project
Action cards instead of silent changes.
Project Chat never writes to the data itself. If you want to change something, it suggests the action as a card – and the card states in plain words what is going to happen.
Even an explicit instruction doesn't lead directly to a change. Only your click carries out the action.
Answers, as they are written.
The answer appears in real time. Instead of waiting for an opaque finished output, you see how Gauss answers the question. The history sits on the right, grouped by time.
The ball shows that Gauss is working
In the centre is a dark area with the animated BrandZaps particle ball, above it a greeting with the time of day and first name.
History grouped by time
“Today” or “5 days ago”: the history shows the first sentence of each conversation so you can quickly find earlier questions.
From the graph into a conversation
“Ask AI” takes you from the Project Brain graph straight into a new conversation – with the finding as the starting point.
View Project Brain →Four safeguards protect the project context.
These safeguards are part of the product logic. They ensure that the chat queries your own project instead of taking new instructions from third-party text.
Competitor pages, AI answer texts, full page texts and the wording of Material do not enter the context.
The project cannot be chosen freely within the tools. The database only returns rows from your own project.
Every change needs confirmation on an action card.
Data only reaches the model as a tool result and is never mixed with instructions.
An assistant for your own project.
Project Chat doesn't search the web and doesn't read uploaded files verbatim. It answers questions about your own project – not about other projects in the same account.
Answers from measured project dataEvidence with every answerTraceable action cardsDirect connection to the Project BrainYou ask. Gauss looks it up. You decide.
Project Chat makes the project context queryable – with evidence, with action cards and without third-party text in the context.