Back to all solutions
Telecom

Branches, call centers, and B2B sales - one AI agent. From plan recommendations to contract, without seams.

Branch concierge, call-center agent takeover, and B2B sales assistant. Instant retrieval across plans, devices, and T&Cs via Graph N-hop Agentic RAG; real service-change execution via Function Calling.

Overview

Telecom has hundreds of plans, device lineups, T&Cs, and promotions updated weekly, while branches, call centers, and B2B sales each run separate systems. Existing AI chatbots stop at simple FAQs - they cannot actually execute activation, number portability, or plan-change APIs, and offline branches, call centers, and sales teams cannot share the same up-to-date knowledge.

AICLUDE unifies branches, call centers, and B2B sales on a single agent platform. Plans, devices, T&Cs, and promotions are retrieved instantly via Graph N-hop Agentic RAG; human intervention flows naturally through agent takeover mode; and real service changes execute via Function Calling. Native kiosk and signage lift the offline branch experience to digital-store quality.

Key Capabilities

Real-time Search across Plans, Devices, T&Cs

Instantly search hundreds of plans, devices, options, T&Cs, and promotions via Graph N-hop Agentic RAG - auto-retrained as weekly changes land.

Call-Center Agent Takeover & Real-time Coaching

Agent takeover mode hands off seamlessly from agent to human. Real-time coaching and response approval suggest scripts and discount options to the agent.

Offline Branch Kiosk & Concierge

Native kiosk and signage + STTS voice. Waiting customers self-serve plan comparison and device consultations, and context is handed off to the human agent.

B2B Sales RFP & Proposal Automation

Analyze enterprise requirements → draft tailored pricing/circuit plans → build competitor comparison. Sales reps only review and revise.

Function Calling Real Service Changes

Activation, number portability, plan changes, and add-on enrollment are registered as external API tools - the agent executes the real API after customer consent.

Customer Persona-Based Recommendations

A persona & customer context engine auto-recommends the optimal plan based on usage patterns, family composition, and history.

Case Stories

Self-contained Application Scenarios

Every case is shown in full: Pain, AICLUDE Apply, Scenario, Impact, and Tech: without collapse.

Case 01

Offline Branch AI Concierge

Customer Pain

  • Branch waits are long, and agents struggle to memorize every new plan and promo.
  • After 30+ minutes of consultation, customers walk away saying "I'll just re-check on the website."
  • Consultation quality is inconsistent across branches.

AICLUDE Apply

  1. 1Native kiosk and signage - unified kiosk & signage operations let waiting customers self-explore.
  2. 2STTS voice (real-time voice recognition, response generation, and voice reply) lets seniors use the kiosk too.
  3. 3Plans, devices, and T&Cs trained via Graph N-hop Agentic RAG - auto-retrained weekly.
  4. 4Function Calling actually executes device-stock checks, reservations, and ticket creation.
  5. 5On agent takeover mode, the kiosk conversation context is auto-handed to the agent screen.

Scenario

Show Me
Input · 60-year-old (at kiosk)
"We're a family of 4 - can we lower the bill?"
  1. 01Step
    STT recognition
  2. 02Retrieve
    Persona RAG:search family-bundle and senior discount policies
  3. 03Retrieve
    Plan RAG:compare with current plan
  4. 04Step
    Present 3 alternative plans + visualize expected savings
  5. 05Step
    Customer:"I'll take the second one"
  6. 06Generate
    Function Calling:create agent ticket + pass context
  7. 07Step
    Agent called - paperwork only

Impact

  • Shorter branch wait times.
  • More cases handled per agent.
  • Less variance in consultation quality across branches.

Tech

  • Native Kiosk & Signage
  • STTS
  • Graph N-hop Agentic RAG
  • Function Calling
  • Agent Takeover Mode
Case 02

Call-Center Real-time Assistant

Customer Pain

  • Average Handle Time (AHT) is long; agents spend time searching manuals.
  • Training on new plans and promos is heavy, and new-agent ramp time is long.
  • Quality management depends on post-call QA - hard to improve in real time.

AICLUDE Apply

  1. 1Agent coaching mode - real-time coaching & response approval guide the agent live.
  2. 2Customer utterance → auto intent analysis → recommended plan/script shown on agent screen.
  3. 3Complex inquiries flow to agent takeover mode; simple ones are handled by the agent alone (hybrid).
  4. 4Audit Log records the full conversation - for compliance and QA automation.
  5. 5PII encryption (AES-256-GCM) + searchable-encryption index - protect customer ID and account info.

Scenario

Show Me
Input · Customer
"My bill is too high - can we lower it?"
  1. 01Retrieve
    Auto-lookup customer ID and current plan (Function Calling)
  2. 02Deliver
    Analyze usage patterns (data, voice/SMS)
  3. 03Step
    Show 3 best alternatives + projected savings on agent screen
  4. 04Step
    Agent explains option 2 to the customer immediately
  5. 05Step
    Customer consents
  6. 06Execute
    Function Calling executes the plan change
  7. 07Audit
    Audit Log auto-recorded

Impact

  • Shorter Average Handle Time (AHT).
  • Faster new-agent onboarding.
  • Less quality variance · QA automated.

Tech

  • Agent Takeover Mode & Real-time Coaching
  • Function Calling
  • Graph N-hop Agentic RAG
  • AES-256-GCM & Searchable-encryption Index
  • Audit Log
Case 03

B2B Sales · RFP Proposal Automation

Customer Pain

  • A single enterprise RFP takes sales, engineering, and legal teams several days.
  • Combinations of circuits, dedicated networks, IoT, and cloud products are complex - drafting takes time.
  • Competitor price-competitiveness analysis is done by hand.

AICLUDE Apply

  1. 1Upload RFP document → auto-extract requirements (file connector + Graph N-hop Agentic RAG).
  2. 2Train on internal product catalog, pricing, and past wins to generate a tailored proposal draft.
  3. 3Web-scrape published competitor prices/terms → auto-build a comparison table.
  4. 4DAG autonomous execution engine runs quoting → tech review → legal review in parallel and integrates them.
  5. 5Sales rep only reviews and revises the AI draft → moves to approval workflow.

Scenario

Show Me
Input · Sales rep
"Prep RFP for Customer A - HQ + 50 branches, 5G + dedicated line"
  1. 01Step
    Upload RFP
  2. 02Step
    extract requirements (circuit count, regions, SLA)
  3. 03Retrieve
    Catalog RAG search
  4. 04Step
    derive optimal combination
  5. 05Step
    3-year TCO calc + volume discount applied
  6. 06Deliver
    Web-scrape published rates from 2 competitors
  7. 07Step
    comparison table
  8. 08Generate
    Auto-draft proposal in Google Docs
  9. 09Step
    Sales rep reviews and revises
  10. 10Step
    internal approval workflow

Impact

  • Shorter RFP response lead time.
  • Consistent proposal quality.
  • Higher sales-org productivity.

Tech

  • Graph N-hop Agentic RAG
  • Web Scraping (Isolated Extension)
  • DAG Autonomous Execution Engine
  • Function Calling
  • Workflow Automation
Case 04

Omnichannel Customer Journey - Web, App, Branch, Call Center

Customer Pain

  • Web, app, branch, and call center are separate systems - customers re-explain when changing channels.
  • Customer shows plan interest on web → branch agent has no visibility into that history.
  • Membership, perks, and points are split per channel - inconsistent experience.

AICLUDE Apply

  1. 1Web widget, app, branch kiosk, call center, Slack, and social all share the same agent and RAG context.
  2. 2A persona engine applies the same identity across channels - branch and call center see web interest history instantly.
  3. 3Physical multi-tenancy separates data by region, franchise, and corporate.
  4. 4SSE real-time notifications sync state across channels - app push and branch kiosk are consistent.
  5. 5Audit Log unifies cross-channel consultation history.

Scenario

Show Me
Input · Web
customer consults a 5G premium plan
  1. 01Step
    saved
  2. 02Step
    (persona context sync) Branch visit (membership recognized)
  3. 03Step
    kiosk:"About the 5G plan you asked about online - want to see the estimated bill with the current promo?"
  4. 04Execute
    Function Calling computes the price in real time
  5. 05Audit
    Simulation shown on signage
  6. 06Step
    Customer:"Great, let's do it"
  7. 07Step
    Agent takeover mode handles paperwork only

Impact

  • Higher web-to-offline conversion.
  • No re-explaining when changing channels.
  • Higher membership return rate.

Tech

  • Omnichannel
  • Persona RAG
  • SSE Real-time Notifications
  • Physical Multi-tenancy
  • Audit Log

Apply AICLUDE to this industry

We shape each PoC around your data, security requirements, and operating flow.