Aller au contenu
Partagez un workflow. Dring AI vous appelle en environ deux minutes et qualifie le besoin. Demander un rappel par l’IA
Cette page est disponible en anglais pour le moment. Voir la page en anglais
Admin and partner panel

Manage production AI agents from one role-scoped admin panel

Prompts, voices, telephony routes, campaigns, tools, post-call workflows, users and keys stay in one operational view. Teams and partners can manage the workflows they own, while controlled releases, CRM outcomes and human handover keep the customer experience accountable.

Hear Dring AI walk through a practical workflow, then submit the one your team wants to improve. Dring AI calls in about two minutes, identifies the need and helps our team shape the right project. Hear a Dring AI demo, see the callback flow or review the operations benchmark.

What you control

Ten areas, one company view.

Pick a company and the panel shows its agents and everything they depend on. Partners see the companies they run; Dring sees all of them.

Agents

Start from a tested template for the use case, or build one from a blank agent: opening line, system prompt, end-call behaviour and the tools each agent may use. Speech recognition, language model and voice chosen per agent from the approved providers.

Phone numbers

Inbound routing to an agent, outbound caller IDs, number provider and background sound. New numbers are requested from the panel and go live after telco setup.

Campaigns

Outbound lists with scheduling, calling days and windows, retry delays, do-not-call marking and SIP rules. Start, pause and monitor from the same screen.

Contacts

Contact lists from a CSV upload or a CRM sync, grouped into segments, assigned to one or more campaigns, with do-not-call flags and a call history per contact.

Tools

What the agent can do during the call: knowledge search, price check, location lookup, policy retrieval, email, custom API requests with fallbacks. Bound to agents explicitly.

Post-call workflows

What happens after the call: analysis fields, then delivery by webhook or email, each step with its own outcome condition, green, yellow or red.

Voices

A voice library per language and brand. Bring your own voice from a supported provider with a voice ID, assign it per agent.

Vendors

Your own API keys for speech, language and voice providers when your contract requires it, or Dring's managed keys by default.

Users and roles

Owner, operator and viewer roles per company. Recording download permission is a separate switch.

API keys and secrets

Keys for your integrations and a key-value store for the secrets tools need, never pasted into prompts.

Controlled release

Change the agent, not the customer experience.

Prompt and behaviour changes do not go live on save. They enter a release pipeline: the test suite runs, judges score, and the change is deployed in stages. Actions and workflow steps that carry no conversational risk apply immediately.

  • Prompt edits: tested and reviewed before they reach a live line
  • Status visible in the panel: live, or awaiting approval
  • Tools, workflows and number assignments: applied instantly, logged
  • Number conflicts prevented: one inbound number, one agent
Agent · support linePrompt · awaiting approval
Opening lineUpdated · shorter greeting
Test suiteWorkflow checks · passed
ReviewRelease criteria · passed
RegressionNo drift detected
RolloutStaged → approved
Live versionCurrent approved release
Operations views

Usage and capacity, without a ticket.

Two screens the operations team lives in: what was used, and what is running right now.

Usage report

Minutes and conversations per company, per agent and per channel for any period. The same numbers your invoice is built on, so there are no surprises.

Live slot monitor

Concurrent calls in progress, by company and by line, with capacity headroom. Peak-hour inbound and a running outbound campaign side by side.

For partners

The same panel, your brand on it.

Partners work in the same Dring panel with account-level roles, reports and tested release controls. They can create customer workspaces from ready agent templates, manage lines and campaigns, and take each deployment live with Dring running telephony, orchestration and analysis behind it. A branded customer-facing layer can be scoped separately when the model requires it.

Partner viewAssigned companies
CompaniesScoped workspaces · assigned agents
TemplatesOrder status · COD confirmation · Abandoned cart · Survey
ReportsPartner logo · monthly · per company
UsersOwner · operator · viewer per company
Agent templates

Start from a template built for the job, not a blank page.

Every agent in the catalogue started as a tested template for its use case: order status, appointment booking, collections, qualification and more, each adapted to a brand's tone, policies and tools. A new agent follows the same path from the panel: pick a template, or start from a blank agent, and adapt the opening line, prompt, voice and tool access.

  • Templates by use case, or a blank agent to build from scratch
  • Opening line, system prompt, voice and tool access, editable per agent
  • Every change tested against the simulation harness before release
  • Version history: what changed, and what is live right now

Browse the agent catalogue

New agentFrom template
TemplateAppointment booking
VoiceAssigned from library
ToolsCalendar, CRM lookup
Test suiteWorkflow simulations
JudgesTwo independent AI judges
StatusAwaiting approval
Campaign scheduling

Calls happen inside the hours you set.

Every outbound campaign carries its own calling window, calling days and retry rules. Nothing dials outside them.

Set the calling window and the days a campaign is allowed to dial, per campaign. A retry rule per outcome decides what happens next: no answer, busy and voicemail each get their own delay before the next attempt, up to the attempt limit that campaign is given. A daily and a weekly cap on attempts keeps a single number from being dialed too often in a short time.

When a customer names a preferred time on the call, the agent sets that as the next call time, still inside the campaign's calling window. A do-not-call request is honoured immediately: the contact is removed from every campaign, not just the one that reached them.

  • Calling window and call days, set per campaign
  • Retry delay per outcome: no answer, busy, voicemail, unreachable
  • Daily and weekly attempt caps per contact
  • Do-not-call marking removes a contact from every campaign
Campaign · WinbackActive
WindowConfigured local time
Call daysWeekdays
No answerRetry in the next permitted window
VoicemailRetry next window, no message left
Callback requestedAgent-set time · inside window
Do-not-callRemoved from all campaigns
Contacts

One list, kept honest across every campaign.

Upload a contact list as a CSV, or sync it from your CRM through Dring Connect. HubSpot is the documented reference path; Salesforce, Zendesk, Freshdesk, Bitrix24, Zoho, Google Sheets and custom CRM environments are mapped to their real objects and permissions. Contacts are grouped into segments and assigned to one or more campaigns.

  • Upload by CSV or sync from your CRM through Dring Connect
  • Segment contacts and assign them to one or more campaigns
  • Do-not-call and consent flags, honoured across every campaign
  • Per-contact history: every call, its outcome and the next action

See Dring Connect

Contact · phone maskedCall history
SegmentTrial ended, no reply
CampaignsWinback
Last outcomeGreen · interested
Next actionRep to send proposal
Do-not-callNo

Twenty minutes in the panel says more than a deck.

Submit context and consent for a callback. Dring AI calls in about two minutes, identifies itself and walks through the panel workflow with your team.