One console. Your team leader runs it.
Eight module groups covering the whole operation. No engineering skill required, no command line, no model configuration. If your supervisor can run a CRM, they can run this on day one.
The state of the operation, in one screen
Live counts across accounts, agents, tasks and escalations. The automation funnel shows what arrived, what the fleet handled, what completed and what is waiting on a person — the single most useful number on the screen being the last one.
- Client management — the accounts allocated to your workspace
- Seller accounts — catalogue size, channel set, service level per account
- Marketplace health — per-channel status and sync drift
| Marketplace | Channel type | Sync cadence |
|---|---|---|
| Amazon | Marketplace | Continuous |
| Flipkart | Marketplace | Continuous |
| Myntra | Marketplace | Scheduled |
| Meesho | Marketplace | Scheduled |
| Own store | Direct | Continuous |
Illustrative rendering. No client data shown.
| Agent | Status | Autonomy | Last run |
|---|---|---|---|
| Inventory AI | Running | Supervised | — |
| Listing AI | Running | Draft only | — |
| Pricing AI | Running | Supervised | — |
| Support AI | Running | Draft only | — |
| Refund AI | Running | Draft only | — |
Illustrative rendering. No client data shown.
The fleet, its balance and its limits
Every agent with its status, autonomy level, tasks completed, accuracy and last run — each with a pause and run control. Two cards sit above the fleet table: the AI operating account balance with its burn rate and per-charge ledger, and the workspace server with processor, memory and storage load.
- Stop the fleet and hand everything to your team, in one control
- Top up the prepaid AI balance without raising a ticket
- The published never-list, shown in the panel where it is enforced
- AI task centre — all, running, queued and failed, with retry and full log
Where your reviewer spends their day
Everything the draft-only agents produce lands here with its confidence, its source context and the action it proposes. Your reviewer approves, edits or overrules — and every overrule is logged with a reason and fed back into the rule set.
Queue ageing is the metric that decides whether this project works for you. It is on the dashboard for that reason.
| Review type | Raised by | Decision |
|---|---|---|
| Listing claim flagged | Compliance AI | Approve / edit |
| Refund above ceiling | Refund AI | Named approval |
| Reply mentions legal | Support AI | Escalate |
| Return dispute | Returns AI | Decide |
| Suspected abuse case | Fraud AI | Open case |
Illustrative rendering. No client data shown.
| Unit | Rate basis | Counted from |
|---|---|---|
| Managed seller account | Per account, per month | Allocation record |
| Order processed | Per order | Panel timestamp |
| Catalogue action | Per action | Listing pipeline |
| Support contact | Per contact | Channel log |
| Return processed | Per return | Returns workflow |
Illustrative rendering. No client data shown.
What you have earned, computed from the same telemetry
Your payout is not a figure someone types in. It is computed from the units the panel recorded, against the published rate card, with the quality score applied. The statement and the scorecard are issued together each month so you can check one against the other.
- Month picker and export for your own reconciliation
- Account-wise breakdown across your allocated accounts
- Running-cost ledger alongside the payout, so net is visible
All panel imagery on this site is a rendering built for illustration. No client, workspace or partner data is shown anywhere. The live console is shown in the demo.
The panel is the product decision.
Fifteen minutes on a working workspace answers more than any page here. Bring your operations head and your worst questions.