Platform
Marella reads your documents, answers questions from them with the source passage attached, and runs agents only with the sources and tools you allow.
Upload files, sync Confluence, or connect another source once we have checked it
Supported files are read into passages, and you can see where each one came from
The people, companies and topics it finds, in a structure your team can correct
Drawn only from the documents in scope, with the source open beside them
Agents set up for one job, with the tools they are allowed to use written down
Architecture
A bad answer can start with a misread scan, a missed passage or a permissions gap, so each layer has its own page and its own test.
Reads your files into passages it can cite, and shows which files failed to parse.
See document processingAnswers from the documents you approve, with each claim linked to the passage behind it.
See cited searchKeeps the people, companies and topics found in your documents in a structure your team can correct.
See the knowledge networkRuns repeatable work with the sources and tools you allow, and stops to ask where you say it must.
See agents with limitsSets who can see what, where the system runs and which model provider it uses.
See deployment optionsThe workbench and the control room
Ask across the documents you have access to, open the passage behind each answer, and hand repeatable tasks to an agent.
Administrators choose which sources are in scope, which model provider runs and which tools an agent may call. Decision traces show what an agent did and what it relied on.
The platform test
A demo on our sample documents proves little. Run real questions over documents like yours, agree the expected answers first, and name who reviews the result.
For example: answer a policy question, compare a clause or assemble a supported brief.
Include the stale, conflicting and awkward material, not just the clean examples.
Write down where documents are stored, which model processes them and who is responsible for what.
Agree what the system may propose and what an authorised person must review or approve.
Platform questions
Deciding where it should run? Read what private AI means and compare the deployment options.
Marella is a private AI platform for an organisation's own documents. It reads them, answers questions with the source passage attached, holds the knowledge your team curates, and runs agents over the same material within limits you set.
No. Search is one workflow. The same documents and controls also support curated knowledge, configured agents and output templates where those are enabled.
No. Your source systems stay authoritative. Marella works over approved copies, and before a connector goes live we test what it does with permissions and what happens when a page is updated or deleted.
Yes: managed by us, in a cloud environment you control after an architecture review, or self-hosted on infrastructure you choose, including on-premise. We write down where your data is stored, which models process it and who supports it for the configuration you pick.
Tell us about the workflow, not the confidential details. We agree how documents are handled before any are shared.
Plan an evaluation