OpenAI Prompts for Telco Support
Budget: $250 – $750 USD
YOU MUST PROVIDE AN EXAMPLE URL IN YOUR BID
I need a set of well-crafted OpenAI prompts that will let our telecommunications support bot handle the bulk of account, billing, and service-management questions without human intervention. Most incoming chats revolve around account status updates—“Is my bill paid?”, “When does my plan renew?”—and simple service-management requests such as adding or removing features. The goal is to generate responses that sound human, stay on-brand, and guide the customer to a resolution or the correct self-service link in just one or two exchanges.
We provide billing services to multiple telcos each of which might allow only subset of actions/queries so your solution MUST support a mechanism to identify which telecommunications company this customer is from then limiting the actions. This can be achieved using prompt directives or seperate vector stores for each telecommunications company HOWEVER this must be in one instance (api key) at OpenAI.
Here’s what I’m expecting:
• A reusable prompt library organised by intent (account status, plan changes, billing issues, etc.) that I can drop straight into our API calls.
• Structured Json Response so we can call the correct REST API for the client.
• Support for more than one telecommunications company with different options (see above)
• Instructions on how to chain or enrich these prompts with our internal data (customer name, balance, renewal date) so answers stay accurate and personalised.
• A short test script or Postman collection showing the prompts in action against OpenAI Chat completion endpoints.
• Acceptance criteria: the bot must correctly identify the inquiry category and return a helpful, brand-compatible answer in under 500 tokens for at least 90 % of the provided test cases.
This is the minimum list of actions we need supported (as noted some telephone companies do not have all of these options)
Account Level
1. Account Balance
2. Extra time to pay
3. Resend my last bill
4. Resend bill for specific month (August, January etc)
5. Pay my bill
6 Change credit card details
Service Level
First if the customer has more than one service must identify which service(telephone) they want to perform an action for the
1. Change Plan
2. Suspend Service
3. Termination Service
4. Query amount billed for Service todate this month
5. Recharge service (for prepaid mobile phone for example)
6. query amount of data available
If you’ve built conversational flows for telcos—or any high-volume support environment—and know how to squeeze the best out of GPT-4, I’d love to see your approach and a quick sample.
I need a set of well-crafted OpenAI prompts that will let our telecommunications support bot handle the bulk of account, billing, and service-management questions without human intervention. Most incoming chats revolve around account status updates—“Is my bill paid?”, “When does my plan renew?”—and simple service-management requests such as adding or removing features. The goal is to generate responses that sound human, stay on-brand, and guide the customer to a resolution or the correct self-service link in just one or two exchanges.
We provide billing services to multiple telcos each of which might allow only subset of actions/queries so your solution MUST support a mechanism to identify which telecommunications company this customer is from then limiting the actions. This can be achieved using prompt directives or seperate vector stores for each telecommunications company HOWEVER this must be in one instance (api key) at OpenAI.
Here’s what I’m expecting:
• A reusable prompt library organised by intent (account status, plan changes, billing issues, etc.) that I can drop straight into our API calls.
• Structured Json Response so we can call the correct REST API for the client.
• Support for more than one telecommunications company with different options (see above)
• Instructions on how to chain or enrich these prompts with our internal data (customer name, balance, renewal date) so answers stay accurate and personalised.
• A short test script or Postman collection showing the prompts in action against OpenAI Chat completion endpoints.
• Acceptance criteria: the bot must correctly identify the inquiry category and return a helpful, brand-compatible answer in under 500 tokens for at least 90 % of the provided test cases.
This is the minimum list of actions we need supported (as noted some telephone companies do not have all of these options)
Account Level
1. Account Balance
2. Extra time to pay
3. Resend my last bill
4. Resend bill for specific month (August, January etc)
5. Pay my bill
6 Change credit card details
Service Level
First if the customer has more than one service must identify which service(telephone) they want to perform an action for the
1. Change Plan
2. Suspend Service
3. Termination Service
4. Query amount billed for Service todate this month
5. Recharge service (for prepaid mobile phone for example)
6. query amount of data available
If you’ve built conversational flows for telcos—or any high-volume support environment—and know how to squeeze the best out of GPT-4, I’d love to see your approach and a quick sample.