Platform

Getting started

Qorpera onboarding is a deliberate four-phase sequence. The goal is not to connect tools and start firing — it is to arrive at a confirmed organisational model that the reasoning engine can trust as its baseline.

1. Connect

An admin accepts the invitation, signs in, and authorises the company connectors. Google Workspace and Microsoft 365 each carry their sub-tools (Gmail/Drive/Calendar/Sheets, Outlook/OneDrive/Teams/Calendar) under a single OAuth consent. HubSpot, Slack, and Stripe are authorised individually.

Employees then authorise their own personal connectors (Gmail, Outlook, Teams) — the AI uses each person’s tokens to act on their behalf rather than via a shared service account. Total admin time: 3–5 minutes. Employee time: under a minute per person.

2. Research

The discovery pipeline analyses your connected data in parallel:

People discovery. Every person the company interacts with is catalogued — employees, clients, partners — with email deduplication and internal/external classification.
Evidence extraction. Content is read and structured claims are extracted using per-source-type prompts.
Synthesis. An Opus 4.7 pass produces the organisational model — departments, roles, situation type recommendations, and an uncertainty log.
Frame-driven page synthesis. Person profiles, external contacts, domain hubs, processes, and relationships are drafted using tuned synthesis frames.

3. Confirm

The admin reviews the AI-generated model. Departments, team assignments, detected situation types, and unresolved questions are presented for human verification. You correct errors, answer questions the AI couldn’t resolve from data alone, and confirm the model. This is where your human judgment enters the loop structurally — not as a one-off correction but as the signal the reasoning engine will trust going forward.

4. Operate

With the model confirmed, continuous operation begins. Connectors sync on scheduled intervals. The situation detection engine evaluates new content and entity state. Detected situations flow through the reasoning engine and surface in My Work for approval.

Most teams see their first detected situations within a few hours of confirming the model. The reasoning engine gets meaningfully better within the first two weeks of active use, as approval patterns inform per-type heuristics.

Previous← How it works NextThe wiki →

1. Connect

2. Research

The discovery pipeline analyses your connected data in parallel:

People discovery. Every person the company interacts with is catalogued — employees, clients, partners — with email deduplication and internal/external classification.

Evidence extraction. Content is read and structured claims are extracted using per-source-type prompts.

Synthesis. An Opus 4.7 pass produces the organisational model — departments, roles, situation type recommendations, and an uncertainty log.

Frame-driven page synthesis. Person profiles, external contacts, domain hubs, processes, and relationships are drafted using tuned synthesis frames.

3. Confirm

4. Operate

Most teams see their first detected situations within a few hours of confirming the model. The reasoning engine gets meaningfully better within the first two weeks of active use, as approval patterns inform per-type heuristics.