Menu

SCCAI — SCcomms AI | How it works

Follow one enquiry through SCCAI.

SCCAI — SCcomms AI — is the AI customer service platform from SCcomms. This example follows a routine question, a request for an exception, a human takeover and a review of the service work.

View the service loop

The SCCAI operating loop

Illustrative example. The policy, messages and records are fictional; they are not a customer transcript.

SCCAI / THE OPERATING LOOP
PreparePeople review the guidance.

01 · Prepare

Prepare the guidance before the question arrives.

The team supplies its business information, separates public material from internal reference guidance and instructions, and reviews what the service will use.

Fictional policy for this example; not a statement of legal entitlement.

Your business knowledge
Public information
Customer-facing business information.
Internal reference
Background guidance with a distinct role.
Response instructions
How the service should respond.
People reviewSources, summaries and instructions.
Illustrative knowledge relationship. Source quality and answer review remain with your team.
Fictional example guidance

Returns can be requested within 14 days of purchase.

Not a statement of legal entitlement.

02 · Meet

Meet the customer on a supported channel.

The customer asks through WhatsApp or web chat. SCCAI answers the suitable routine question from the example guidance.

Fictional web-chat exchange
“What is your return policy?”
Customer-facing agent

You can request a return within 14 days of purchase.

Illustrative example. Fictional 14-day policy, not a legal entitlement or a customer transcript.

03 · Understand

Keep available customer context in view.

For a known customer, SCCAI can use profile information, configured fields and recent ticket context. Teammates can review earlier messages and the customer record as they handle the situation.

Context follows the available customer record and the team's access.

Available context for a known customer
Customer profile
Known record and configured fields.
Recent tickets
Available follow-up context.
Earlier messages
The teammate can review the conversation.
Illustrative relationship. Context depends on known records and granted access. Team notes are not automatically all supplied to AI.

04 · Resolve

Bring in a person when the request needs judgement.

An authorised teammate takes over, reads the earlier messages and reviews the exception. The person chooses when to hand the conversation back to AI.

Where ticketing is enabled, follow-up work can receive an owner and status. The conversation keeps the discussion; the ticket keeps track of the remaining work.

One conversation · ownership changes
Customer

What is your return policy?

Customer-facing agent

You can request a return within 14 days of purchase.

Customer

It has been 20 days. Can your team review my request?

Ownership: AI-held

An exception needs a person.

An authorised teammate can initiate takeover and read the earlier messages.

Ownership: authorised teammate

The person reviews the situation.

“I can see your earlier messages. Let me review the situation.” The exception is a person’s decision.

Ownership: handed back deliberately

The teammate chooses when to hand back.

Earlier messages remain available. Follow-up can have an owner and status where ticketing is enabled.

Illustrative example. Messages and the 14-day policy are fictional, not a customer transcript or a statement of legal entitlement. These controls change the explanation.

05 · Supervise

See where the team needs support.

A manager can inspect waiting customers, ownership and workload within the permitted operating scope. Important activity records help the team review what changed and who acted.

Your service day, in view
Waiting customers
Who still needs attention?
Conversation ownership
Who is responsible now?
Current workload
Where could the team help?
Permitted branch contextThe manager decides where attention goes.
Illustrative service overview. Visibility follows the user’s permitted branch context.

06 · Improve

Improve the guidance through review.

The team reviews real conversations and service performance. Where questions are missing or guidance needs correction, people update the sources or instructions and check later replies. Surveys can add feedback where configured.

Review → update guidance → check later replies.

  1. Review conversations and performance.
  2. People update guidance.
  3. Check later replies.
Human review returns the service to Prepare. Surveys add feedback where configured.

The staff copilot helps with a different kind of work.

Inside the portal, a teammate can ask the copilot to perform a supported customer-record update. The copilot uses that teammate’s permissions, pauses for confirmation, and lets the teammate inspect the saved result.

A staff-copilot example, separate from the customer-facing conversation.

Explore Trust & Control
Staff copilot · separate internal task
A teammate’s access · supported customer-record update
  1. 01Proposed change

    The teammate requests a supported customer-contact update. The copilot presents the proposed change within that person’s permissions.

    Existing contact detailProposed contact detail
  2. 02Explicit confirmation

    This customer update pauses for the teammate to confirm the proposed change. Selected sensitive tasks have their own confirmation rules.

    A person confirms this selected change.
  3. 03Record to inspect

    After confirmation, the teammate can inspect the saved customer record and relevant activity.

    Inspect the result and recorded activity.
Illustrative staff-copilot sequence. This is a separate internal task, not an action triggered automatically by the customer conversation.

The next useful step

See the workflow in a demo.

Bring one repeated enquiry and an exception your team should own. Explore the supported work, takeover and review together.