Most companies think of AI customer service and sales development as two separate systems. Customer service is reactive — answering questions, handling complaints, processing returns. Sales development is proactive — qualifying leads, nurturing prospects, booking demos. But what if one agent could do both?
That's the enterprise play with OpenClaw (Clawdbot): a unified AI agent that handles inbound customer service and outbound sales development from a single deployment on Tencent Cloud Lighthouse. Let's look at how real teams are making this work.
The Problem: A direct-to-consumer fashion brand was handling 300+ customer messages daily across WhatsApp and Telegram. Their CS team spent 80% of their time on repetitive queries (sizing, shipping, returns) and had zero bandwidth for proactive sales.
The Solution: They deployed OpenClaw on Lighthouse with a dual-purpose knowledge base:
Customer: "What size should I get for the winter jacket? I'm usually a Medium."
Agent: "For the winter jacket, Medium fits chest 38-40 inches. Based on
customer feedback, it runs slightly large — if you're between sizes,
go with Medium. By the way, customers who bought this jacket often
pair it with our thermal base layer — it's 20% off this week when
bundled. Want me to add it to your cart?"
The Results:
The Problem: A B2B SaaS company received a mix of support tickets from existing customers and inquiry messages from potential buyers — all through the same channels. Human agents had to context-switch between support and sales constantly, and leads often went cold because nobody followed up fast enough.
The Solution: OpenClaw was configured with intent detection in the system prompt:
Classify each incoming message into one of two categories:
1. SUPPORT: Existing customer with a product issue or question
2. PROSPECT: Potential buyer asking about features, pricing, or demos
For SUPPORT: Answer using the product knowledge base. Escalate
technical issues to the engineering team.
For PROSPECT: Qualify the lead by asking about company size,
use case, and timeline. If qualified, offer to book a demo
and collect their email.
The Setup:
ssh ubuntu@<lighthouse-ip>
# Configure with multi-purpose knowledge base
clawdbot onboard
# QuickStart → Use existing values → Configure model → Add channels
# Enable daemon for 24/7 operation
loginctl enable-linger $(whoami) && export XDG_RUNTIME_DIR=/run/user/$(id -u)
clawdbot daemon install
clawdbot daemon start
Security note: The lead qualification flow collects personal information (emails, company names). Ensure your API keys and any CRM integration credentials are stored securely — never hardcoded in scripts. Use the Tencent Cloud console's visual configuration panel.
The Results:
The Problem: An online marketplace with 50+ vendors needed to provide customer service for all vendors through a unified interface. Each vendor had different shipping policies, return windows, and product catalogs. Hiring CS agents for each vendor was prohibitively expensive.
The Solution: A single OpenClaw instance with a structured, vendor-segmented knowledge base:
[VENDOR_A - Electronics]
- Shipping: Free over $50, 3-5 days standard
- Returns: 15-day window, must be unopened
- Warranty: 1 year manufacturer warranty
[VENDOR_B - Fashion]
- Shipping: Flat $5.99, 5-7 days
- Returns: 30-day window, tags must be attached
- Exchanges: Free for size/color changes
[VENDOR_C - Home Goods]
- Shipping: Calculated by weight, 7-10 days
- Returns: 30-day window, buyer pays return shipping
- Damage claims: Photo required within 48 hours
The agent identifies the vendor from the product or order context and applies the correct policies automatically.
Cross-sell integration: After resolving a query about Vendor A's product, the agent suggests complementary items from Vendor B or C:
"Your new headphones from TechStore will arrive by Friday! If you're looking for a carrying case, HomeGoods has a great one that fits perfectly — $14.99 with free shipping on orders over $50."
The Results:
All three case studies share the same deployment foundation: OpenClaw on Tencent Cloud Lighthouse.
Visit the Tencent Cloud Lighthouse Special Offer page:
For enterprise workloads (500+ daily conversations), the 4-core, 8 GB RAM tier provides comfortable headroom. All three teams used overseas regions for global messaging platform connectivity.
Pattern 1: Dual-purpose system prompts — Every deployment combined CS and sales logic in a single prompt, with clear intent classification.
Pattern 2: Proactive follow-up — The agent doesn't just answer and wait. It actively suggests next steps (upsell, demo booking, cross-sell).
Pattern 3: Structured knowledge bases — Information organized by category/vendor/product, not dumped as unstructured text.
Pattern 4: Multi-channel presence — All teams connected at least 2 messaging platforms:
# Typical multi-channel setup
clawdbot onboard # → WhatsApp
clawdbot onboard # → Telegram
clawdbot onboard # → Discord
Channel guides: WhatsApp | Telegram | Discord
Pattern 5: Continuous optimization — Weekly review of conversation data, knowledge base updates, and system prompt refinements.
Whether you're a D2C brand, a SaaS company, or a marketplace, the playbook is the same: deploy OpenClaw, build a dual-purpose knowledge base, connect your channels, and let the agent handle both service and sales.
Head to the Tencent Cloud Lighthouse Special Offer page, select OpenClaw (Clawdbot) under AI Agent, and click "Buy Now". Start with customer service, add sales development, and watch a single agent do the work of an entire team.