Bring permitted records into your software.
Your developer can assess the read interface for signing status, evidence references and obligations. Use the information in a reporting process instead of manually copying updates.
Less copying between systems. Less chasing for status. A clearer next step.
Bring your existing software and approved AI tools into the same working process, with access and review under your control.
Connections are assessed per engagement. The examples below distinguish the documented read interface from planned actions.
The tools your team already works with.
Illustrative connection model. Client compatibility and deployment setup require confirmation.
An employee preparing a review needs to know what is still outstanding. The useful result is a concise answer linked to the records behind it.
The documented interface includes reads for signing envelopes, evidence references and obligations. It does not grant an assistant unrestricted access to your workspace.
“Which signatures are outstanding, and which obligations need attention?”
Return to the envelope status before following up.
Review obligation · Due for attentionOpen the recorded obligation and confirm its owner.
Your developer can assess the read interface for signing status, evidence references and obligations. Use the information in a reporting process instead of manually copying updates.
MCP gives an AI client access to selected tools. The documented catalogue is read-only, with workspace scope bound to the access grant. Discuss a Claude or ChatGPT connection with us; the exact client mode, authentication and supported tools need testing.
The evidence-verification CLI serves a different purpose: inspecting an exported record. It is not a general command-line client for running your business. See the verification guidance or request the supported tooling.
A background connection should have a specific purpose, permitted inputs and a result someone can inspect. We scope that job with your technical and operational teams.
Retrieve permitted status information and prepare a summary with references. Confirm deployment readiness before relying on it.
Task creation, messages and other writes are separate capabilities. They are not included in the documented read-only MCP catalogue. Any future write tool needs its own permissions and approval design.
An assistant may help organise information. Compliance decisions remain with the responsible human reviewer.
Connecting an external AI can send permitted information to that provider. It is a different data flow from the planned self-hosted Exedro assistant.
Explore deployment and data boundaries ↗Agree the workspace, credentials, allowed reads, recipient service and revocation process.
Scope the destination, information exchanged, permitted actions and failure handling. An inbound connection does not automatically authorise an outbound one.
We will map the systems, information and review points, then confirm the supported connection and implementation scope.
EXEDRO AI · PRODUCT PREVIEW
Explore how Exedro can turn permitted records into source-linked findings, prepared follow-ups and reports. Scope external connections alongside your hosting and approval requirements.
Explore Exedro AI ↗Scope a customer chatbot around your approved knowledge base, calendar and enquiry routing. Give the receiving employee a useful summary.
Explore chatbot configuration ↗