A practical, production-ready guide to configuring BSNL SIP PRI (IMS SIP Trunk) on modern Asterisk (PJSIP) with OpenVPN. Covers authentication, routing, user_eq_phone, scoring logic, debugging, and real-world carrier behavior. BSNL now delivers SIP PRI over FTTH fiber using IMS-based SIP signaling routed through VPN (OpenVPN or SoftEther). Key realities: SIP Proxy is reachable only via VPN SIP authentication may use: IMS expects proper user=phone format RTP may traverse different subnets If your PBX doesn’t support OpenVPN natively, you must deploy a router or gateway device that does. 🛠 Step 1 – Configure OpenVPN BSNL provides: VPN Server Primary IP VPN Server Secondary IP Virtual IP (Client) Gateway Mask After connecting: ip addr show tun0 ip route ping 10.x.x.x # SIP Proxy Ensure: VPN interface is UP Route to SIP proxy goes through VPN SIP proxy is reachable 🧩 Step 2 – Configure BSNL SIP Trunk (PJSIP – NOT chan_sip) Modern Asterisk uses PJSIP , not sip.conf. Below is a ...
🧠 Human-Like Conversations at Scale: How We Used Android Prompts + n8n to Revolutionise SMS Lead Qualification for Debt Help Services 🇬🇧📱
In today's world, no one wants to talk to a bot—especially when they’re vulnerable, in debt, or under emotional stress. But what if your AI assistant didn’t feel like a bot at all? What if it sounded like a real human, used UK idioms, responded with empathy, and followed sales psychology techniques like SPIN Selling?
We recently implemented a fully customisable Android-style Prompt system in n8n for a debt help service that was drowning in inbound leads but struggling with qualification and conversion.
Let me walk you through what we built, how it works—and the emotional impact it's had, through a few stories that stuck with us. 💬
✅ The Challenge: Too Many Leads, Not Enough Conversations
The client was generating hundreds of inbound leads weekly through web forms. But many were falling through the cracks. They needed to:
- Qualify interest (without sounding robotic)
- Schedule calls only with warm leads
- Respect UK opening hours
- Handle sensitive objections (like shame around debt)
- Never alienate a user or make them feel like a statistic
💡 The Solution: Human-First Android Prompts Powered by n8n
We built a modular Android Prompt system in n8n using:
- Dynamic field inputs (e.g., date, first message, bump messages)
- SPIN selling principles baked into response logic
- Real-time SMS & WhatsApp automation via Twilio
- Intelligent booking logic based on business hours
- Custom rules for anger detection, follow-ups, and re-engagement
Unlike generic AI responses, these prompts felt human, empathic, and localised for UK audiences.
Let’s get real. People can sniff out a bot instantly. So we made sure every message sounded like “Sarah from admin”—not GPT-4.
🧵 Real People. Real Debt. Real Relief.
Here are three real stories from the front lines, showing how this automation connected humans, not just contacts.
🧍♂️ “I’m embarrassed to talk to anyone about this…”
Tom, a 42-year-old tradesman from Manchester, replied to our first message with:
"I don’t want to talk to a robot, I just want help."
But instead of auto-responses, he received this:
"Thank goodness, my calendar just pinged me to call, but I didn’t want to disturb you—are you still looking for help?"
He paused. Then shared his biggest debt—£18,000 on 5 credit cards. Our AI continued:
“Totally understand, how’ve you tried to reduce the debt so far?”
He opened up: sold tools, skipped holidays, but was still stuck. Tom booked a call that afternoon. After the chat, he said:
“Didn’t feel like a bot at all. Felt like someone finally heard me.”
👩👧 “I’m a single mum and scared to pick up the phone”
Jasmine, a lead who came in after midnight, didn’t respond for 2 days. Then she saw our empathetic bump message:
“Hiya, I didn’t want to bug you—just wanted to check if you still need help? We can book a time that suits you.”
She replied:
“I’m a single mum, I work nights. I can’t talk now but I’m drowning.”
Instead of pushing, the AI asked:
“When would be a good time to reach back out?”
She picked Tuesday at 10 AM. We scheduled it. She later told the advisor:
“I didn’t feel judged. It didn’t rush me. Just gave me space.”
That’s trust at scale.
👨💼 “I thought it was spam, until it didn’t give up”
Mark, a business consultant, originally ignored the message—then replied “not interested”. Our prompt respectfully replied:
“No worries. Just to confirm—are you looking for help with your debt?”
He paused, then said:
“Actually, yeah. Just didn’t think you’d actually read that.”
He had over £25K in tax debt and felt ashamed. The conversation gently peeled back his objections, one question at a time. He booked in a call the next day. That’s powerful re-engagement—without pressure, fluff, or scripts.
🔁 Key Features That Made It Work:
- ✅ Dynamic SPIN Selling logic
- 🕰️ Real-time hour-based logic (respects open/closed times)
- 📅 Smart Booking Protocols (no weekend calls!)
- 😡 Angry message detector (triggers “goodbye” rule)
- 🧠 Memory-aware bump replies
- 🇬🇧 UK tone, slang, and rhythm throughout (“hiya”, “no worries”, “Brilliant, I’ll reach back out then.”)
🔧 Under the Hood: How We Built It in n8n
We used:
- HTTP Request nodes for Twilio SMS
- Switch nodes to handle positive/negative responses
- Webhook triggers for CRM updates
- Function nodes to inject date/time logic
- Custom JSON logic for conversational memory
The whole prompt system is configurable, scalable, and deployable in under a week.
🎯 Final Thought: Automation Shouldn’t Feel Automated
If you're working in high-trust spaces like debt help, mental health, or financial services, your automation needs to feel like a friend, not a funnel.
This Android Prompt + n8n system proves that you can qualify leads, follow up persistently, and still treat people with dignity.
🔗 If you’re curious about how this could work for your niche—whether it’s legal support, health clinics, recruitment, or wellness services—drop me a DM. Let’s make your AI feel human.
📩 Or comment "SMS AI" below and I’ll send over a demo of the exact prompt framework.