What problem does it solve?
Telecom offers are hard to compare. Plans differ by data, speed, contract length, device instalments, bundles and promotions that change often, and eligibility depends on the customer's current contract, credit and address. Customers who cannot compare offers may pick a plan that does not fit, or call to ask, and the answer can depend on how current the advisor's knowledge of the promotions is.
Operators already score who is likely to upgrade, but the moment of decision happens in a conversation: a question in the app, a chat about roaming before a trip, a call about a slow connection. Rule based chatbots could list plans, not reason about which one fits this customer, and advisors lose time during sales calls looking up the current offer and writing up the call.
How does it work?
- Understand the need. The assistant asks about usage, household, devices and budget, or reads the customer's actual usage and contract after authentication.
- Check eligibility and current offers. It retrieves the offers, promotions and upgrade eligibility that apply to this customer from the product catalogue and decisioning engine, never from memory.
- Recommend and compare. It proposes a small number of options with the reasons and the total cost, including what changes on the next bill, and compares devices on request.
- Complete or hand over. For simple purchases (a roaming pass, an extra, a plan change) it completes the order with confirmation; for contracts, devices on credit or complex bundles it prepares the quote and passes it to an advisor or the checkout.
- Assist advisors. In stores and contact centres the same engine suggests the next best offer and answers product questions for the advisor during the conversation.
- Audience
- Customer facing
- Autonomy
- Supervised agent
- Adoption
- Early adopters
- Channels
- Mobile app, Web chat, WhatsApp, Phone and voice, Agent desktop
What is it worth?
Benchmarks are computed from the public deployments below: one data point per organization per KPI, with who made each claim.
| KPI | Median | Reported range | Data points | Claimed by |
|---|---|---|---|---|
| Interactions handled | Not pooled | 70,000 to 45 million | 3 | 2 organization, 1 vendor |
| Conversion uplift | Too few to pool | 75% | 1 | 1 vendor |
| Users served | Not pooled | at least 83,000 | 1 | 1 vendor |
Value drivers: Revenue growth, Customer experience, Employee productivity.
Indicative value
An operator with 3 million postpaid mobile and broadband customers
USD 180,000 to USD 1.4 million
Incremental annual margin from assisted upgrades per year
How this is calculated
Formula: customers * engagedShare * incrementalConversion * marginPerSale. The low scenario uses every low input, the high scenario every high input.
| Input | Low | High | Basis |
|---|---|---|---|
| Postpaid customers customers, customers | 3,000,000 | 3,000,000 | The reference operator. |
| Share of customers who have a sales conversation with the assistant or an assisted advisor each year engagedShare, fraction of customers | 0.2 | 0.4 | Editorial assumption, replace with your own digital and assisted sales reach. |
| Additional upgrades or extras bought per engaged customer incrementalConversion, fraction of engaged customers | 0.01 | 0.02 | Editorial assumption and deliberately low against the benchmark on this page (Pega reports a 75% increase in offer acceptance at Telenet), because that figure is relative, has no stated baseline or control group and comes from one vendor story. |
| Incremental annual margin per upgrade or extra marginPerSale, USD per sale per year | 30 | 60 | Editorial assumption, replace with your own margin per upgrade. |
What it leaves out: Counts only the first year's margin on additional sales. It leaves out lower churn from better fitting plans, the cost of the AI and catalogue integration, cannibalisation of sales that would have happened anyway, and discounts given to close.
Who already uses it?
9 public deployments, strongest evidence first. Grades: A regulator or audit, B the organization itself, C vendor case study, D anonymous or estimate.
Singtel
Singapore · Telecommunications · 2026
In a partnership announced on 4 March 2026, Singtel upgraded the customer care assistant Shirley of Singtel Singapore with Sierra's agentic AI, starting with a pilot that went live in under ten weeks. Shirley verifies customer details, resolves mobile and home troubleshooting, completes roaming sign ups and understands local expressions including Singlish; customers purchased more than 200 roaming add ons independently. In its first six weeks it handled over 70,000 cases. Singtel Singapore also says it will deploy voice AI agents for outbound sales calls within defined compliance and governance standards.
- Containment rate: 73%, mobile and home troubleshooting cases, initial results after launch
"73% of mobile and home troubleshooting cases were resolved without requiring a Customer Care officer."
Claimed by: organization - Automation rate: 76%, roaming sign up requests, initial results after launch
"76% of roaming sign-up requests were completed successfully without requiring a Customer Care officer."
Claimed by: organization - Interactions handled: at least 70,000, first six weeks after launch
"Singtel went live in less than 10 weeks, and in the first six weeks since launch, ”Shirley” handled over 70,000 customer cases involving high volume requests in areas such as mobile issues and roaming services."
Claimed by: organization
Orange France
France · Telecommunications · 2025
Orange France deployed Mon Assistant IA (MAIA) with Verint to 3,000 Orange sales advisors, launching it in early December 2025. During calls it understands the conversation, detects customer needs, retrieves relevant information and summarises the exchange to update the customer file; the advisor validates the proposed responses. Orange also announced Sharlie, a speech to speech voice assistant for its digital brand Sosh built with Microsoft and ILLUIN Technology on the ILLUIN Dialogue and Microsoft Foundry platforms, with capacity for over 3 million conversations a year once deployed.
- Interactions handled: about 1 million, conversations supported per month, as of March 2026
"Launched in early December 2025, Mon Assistant IA already supports nearly 1 million conversations per month."
Claimed by: organization
Verizon
United States · Telecommunications · 2025
In June 2025 Verizon announced a customer experience overhaul: a Customer Champion who owns complex issues end to end, drawing on Google Cloud AI including Gemini models, 24/7 live chat with human agents, and a new My Verizon app, which includes an AI powered Verizon Assistant, in which customers can become a customer, manage upgrades, add lines and ask billing questions. Verizon describes the assistant as voice enabled for mobile customers. Verizon's chief executive framed the programme as a way to build loyalty and improve retention. No outcome figures were published.
No outcome disclosed.
Virgin Media O2
United Kingdom · Telecommunications · 2025
Virgin Media O2 built its own tool, Lumi AI, that analyses a live conversation and prompts the advisor with resolutions that worked for similar customers and with the products and services most likely to interest this customer. In July 2025 it was in pilot with a cohort of advisors in care, telesales and retentions, with a wider rollout planned. Alongside it the operator uses an AI contact centre service from Amazon Web Services that routes callers by their stated reason, software that flags potentially vulnerable customers, and automatic call summaries. The retention effect of Lumi AI has not been published.
No outcome disclosed.
T-Mobile
United States · Telecommunications · 2025
T-Mobile built PromoGenius on Power Apps to give retail and call centre representatives one place for current promotions, discounts and trade in values, used on iPads on the shop floor. An agent built in Copilot Studio reads more than 20 device makers' websites, answers technical questions in natural language during a customer conversation and builds comparison tables that can be shown to the customer. Microsoft reports over 83,000 unique users and 500,000 launches a month for the app, which supports all T-Mobile retail stores and call centres.
- Users served: at least 83,000, unique users, retail and call centre staff
"The app, called PromoGenius, is the second most popular app at T-Mobile, supporting all T-Mobile retail outlets and call centers, with over 83,000 unique users and 500,000 launches a month."
Claimed by: vendor
Vodafone
United Kingdom · Telecommunications · 2024
TOBi is Vodafone's digital assistant on the website, the My Vodafone app, messaging and telephony, first launched in Italy and extended to 15 language versions. It handles billing questions, contract updates and simple technical troubleshooting, hands over to a live agent with a summary, and during the pandemic made the same sales offers as human agents, such as data boosts and upgrades. Microsoft reports that TOBi now fully resolves 70% of inquiries arriving through digital channels; an earlier Microsoft story quotes Vodafone on a 12% year on year fall in contacts to call centres after launch.
- Containment rate: 70%, inquiries arriving through digital channels
"Currently, TOBi handles nearly 45 million customer calls a month, fully resolving 70% of customer inquiries coming through the company’s digital channels."
Claimed by: vendor - Interactions handled: about 45 million, per month
"Currently, TOBi handles nearly 45 million customer calls a month, fully resolving 70% of customer inquiries coming through the company’s digital channels."
Claimed by: vendor - Contact deflection: 12%, frequency of customer contacts to call centres, year over year after the TOBi launch
"Since launching TOBi, we’ve reduced the frequency of customer contacts to call centers by 12 percent year-over-year"
Claimed by: organization
Reliance Jio
India · Telecommunications · 2023
Jio runs a WhatsApp assistant built with Haptik with more than 900 intents. It covers the 5G customer lifecycle end to end, from lead generation and buying a 5G device to porting into Jio, choosing and buying plans and customer care, with separate journeys for prepaid and postpaid, and sends proactive top up and recharge reminders. Haptik reports that the channel acquires 8,000 new Jio Fiber and 5G customers a day.
No outcome disclosed.
Telenet
Belgium · Telecommunications · 2023
Telenet, a provider of connectivity and entertainment services in Belgium, uses Pega's AI based Customer Decision Hub as a single decisioning system that responds to customer signals in real time, anticipates how behaviour may change and proposes the next best action, such as personalised upgrades and solutions, with the stated goals of reducing churn and raising offer acceptance. Pega reports a 20% reduction in churn, a 75% increase in offer acceptance and a 33% increase in cross sell. These figures cover the whole decisioning programme, including upgrades and cross sell, not retention alone.
- Churn reduction: 20%
"20% reduction in churn"
Claimed by: vendor - Conversion uplift: 75%
"75% increase in offer acceptance"
Claimed by: vendor
Mobily
Saudi Arabia · Telecommunications · 2022
Mobily deployed customer facing AI agents on eight channels, including WhatsApp, Twitter and Apple Business Chat, connected to its internal systems. The agents answer billing, balance and data usage questions, change subscriptions, sell add ons, take payments and recharges, and handle feedback and complaints, with a warm handover to a specialist who can take over or hand back. NiCE Cognigy reports that the first response time fell from 20 minutes to about 6 seconds. The deployment was already live in 2022, when the case study described it as conversational AI; the current version presents it as agentic AI.
- Response time reduction: 99.5%, first response time on messaging channels
"An AI agent picks up any inquiry in around 6 seconds, reducing first response times significantly from the previous 20 minutes: a 99,5% improvement."
Claimed by: vendor
How do you implement it?
A model agnostic playbook: what to prepare, the order to build in, and what goes wrong.
Data you need
- A current product, price and promotion catalogue with start and end dates
- Upgrade eligibility and credit rules available by API
- Customer usage and contract data for authenticated recommendations
- Marketing consent and contact preferences per customer
Systems to integrate
- Product catalogue and pricing engine
- Decisioning or next best action engine
- Order management and checkout
- CRM with contract, usage and consent data
- Contact centre and store systems for advisor assistance
Complexity: Medium
Recommending is easy; recommending correctly is not. The work is in a clean, current product and promotion catalogue, real eligibility rules and an order API. Selling on credit or under long contracts adds pre contract information and consent duties.
- 1
Fix the catalogue before the conversation
Put every plan, extra and promotion, with its eligibility and end date, into one source the assistant reads through a tool. Stale promotions are the fastest way to lose trust.
- 2
Start with simple, reversible purchases
Roaming passes, data extras and plan changes without a new contract are low risk. Add devices on credit and new contracts later, with the full pre contract information flow.
- 3
Show the total cost
Make the assistant state the monthly and total cost and the change on the next bill for every option, so customers do not feel sold to and complaints stay low.
- 4
Give advisors the same brain
Use the same recommendation and product answers in stores and contact centres, so a customer hears the same offer on every channel.
- 5
Measure against a control group
Hold out a random share of customers or conversations to measure real incremental conversion, not sales that would have happened anyway.
Guardrails
- Prices, promotions and eligibility come only from catalogue tools, never from the model
- Total cost and contract length stated before any order is confirmed
- Marketing consent checked before any proactive or outbound offer
- No offers to customers flagged as vulnerable or in financial difficulty without human review
- Explicit confirmation before every order, with a record of what the customer saw
KPIs to instrument
- Incremental conversion against a holdout group
- Order cancellations and returns within the cooling off period
- Complaints about sales or pricing that mention the assistant
- Advisor handling time on sales calls with and without assistance
- Revenue and margin per assisted sale
Human in the loop
Advisors complete contract and device on credit sales from the prepared quote. The commercial team approves every new offer the assistant may present, and a quality team reviews a weekly sample of sales conversations for mis selling and unclear pricing.
Common failure modes
- Stale or wrong offers
- The assistant quotes a promotion that ended. Keep all offers in a tool with end dates and test daily.
- Mis selling
- Customers are pushed to bigger plans they do not need. Recommend on actual usage, show the total cost and review samples.
- Ignoring consent
- Proactive offers reach customers who opted out of marketing. Check consent in the tool, not in the prompt.
- Credit decisions by the back door
- The assistant decides who may buy a device on credit. Keep credit checks in the existing, governed process.
What are the risks and rules?
EU AI Act
Depends on design
A sales assistant is limited risk with an Article 50 duty to disclose AI. If it assesses the creditworthiness of individuals for devices on credit, that part is high risk under Annex III point 5(b), so keep credit decisions in the existing governed process. Selling that uses manipulative or deceptive techniques, or exploits a customer's age, disability or economic situation, to materially distort a purchase decision in a way likely to cause significant harm is prohibited under Article 5(1)(a) and (b).
Rules that apply
Guidance
- Article 50, transparency obligations for providers and deployers of certain AI systems (European Union, Europe). People must be informed that they are interacting with an AI system unless this is obvious from the context.
- Directive on privacy and electronic communications (Directive 2002/58/EC) (European Union, Europe). Sets consent rules for unsolicited electronic marketing, which apply to proactive offers sent by messaging, email or automated calls.
- Customers to get clearer broadband information (Ofcom, Europe). Ofcom guidance says UK broadband providers must describe the underlying network technology in clear terms before a customer agrees to buy, whether online, by phone or in person, so a sales assistant has to give the same information.
Controls to put in place
- AI disclosure at the start of every sales conversation
- Offer approval workflow with owners, start and end dates
- Consent and vulnerability checks enforced in tools
- Record of the offer, price and terms shown before each order
- Monthly review of cancellations, returns and sales complaints
Frequently asked questions
- Does AI actually increase telecom sales?
- There is early evidence, mostly vendor reported. Pega reports that Telenet saw a 75% increase in offer acceptance and a 33% increase in cross sell with AI decisioning, and Singtel reports that customers bought more than 200 roaming add ons independently through its assistant Shirley, among the initial results after launch. Neither states a control group, so measure against a holdout group before you trust the uplift.
- Should the assistant sell to customers or help advisors sell?
- Many operators do both. Simple extras and plan changes suit self service, while devices on credit and new contracts often go through advisors, who benefit from the same offer and product information: T-Mobile gives store and call centre staff its PromoGenius app, and Orange France gives 3,000 sales advisors an AI assistant during customer calls.
- What rules apply to AI generated offers?
- The usual telecom and consumer rules: clear pre contract information, total cost, cooling off rights and marketing consent for proactive offers. An AI assistant has to follow them in every conversation, which is easier to prove when prices and terms come from tools.
How to cite this page
Blits.ai AI Use Case Library, "AI assistant for telecom plan upgrades, add ons and sales", last verified 26 September 2026, https://www.blits.ai/ai-use-cases/plan-upgrade-and-sales-assistant. Licensed under CC BY 4.0. Method: how we verify use cases.
Changelog
- 27 September 2026: First published