Saleswolf

Lead Conversion AI: Analyze startup prospects—fetch LinkedIn insights, business data, web trends, AWS solution designs, and delegate follow-ups—to turn leads into cloud-native customers.

Comment

Inspiration

We wanted to bring every sales use-case under one roof. Working and interning at different startups, we’ve seen firsthand how time- and resource-starved sales teams struggle to:

  • Scrape LinkedIn for decision-makers
  • Analyze business models for the right pitch
  • Craft technical AWS solutions
  • Coordinate follow-ups without dropping the ball

What it does

Saleswolf is your end-to-end Lead Conversion AI Agent. It ingests a startup lead’s core info (company name, industry, contact), then:

  1. 🔍 Scrapes LinkedIn for org structure and decision-makers
  2. 📊 Analyzes business model, funding stage, and cloud needs
  3. 🌐 Searches the public web for recent news or launches
  4. 🏗️ Generates a tailored AWS architecture plan (top 5 services, 3-step diagram, cost tips)
  5. 🤝 Delegates follow-up tasks to the right team members with relevant collateral
    All orchestrated in real time via n8n, so sales teams get a turnkey, personalized conversion playbook—no manual legwork required.

built with

  • Orchestration: n8n workflows glue together each sub-agent
  • Agents & APIs:
    • Apify for LinkedIn scraping
    • OpenAI / AWS Bedrock for LLM-driven analysis & architecture design
    • Operant AI for privacy redaction and compliance guardrails
    • Minimax Audio for optional voice-based inputs
  • Architecture:
    1. Parent agent receives { company_name, company_info, lead_name, lead_email }
    2. Executes sub-workflows via Execute Workflow nodes
    3. Collates outputs into a single dashboard and Slack/email notifications
  • Tech Stack: Node.js micro-services, Python scripts, Chart.js for analytics, AWS Lambda for hosting

Challenges we ran into

  • API orchestration: Chaining 6+ disparate services in a single flow
  • Data normalization: Merging LinkedIn scrape results, business insights, and web data into a uniform schema
  • Privacy & compliance: Auto-redacting PII before any LLM call
  • Error handling & retries: Ensuring idempotent workflows when a node or external API fails
  • Time constraints: Building a robust, demo-ready platform within a 5-6 hour hackathon sprint

Accomplishments that we’re proud of

  • 🚀 Fully automated, end-to-end demo in under 6 hours
  • 🔄 Modular agent architecture: easily add, remove, or swap sub-agents
  • 🔒 Built-in privacy guardrails via Operant AI, no manual redaction needed
  • 📈 Dynamic analytics dashboard showing vibe classification, service breakdown, and conversion strategy in real time
  • 🤖 Seamless handoff: auto-delegate follow-up tasks to your internal team with contextual resources

What we learned

  • Clear data contracts between agents are critical for smooth orchestration
  • Few-shot prompting can replace custom model training for classification tasks
  • Low-code tools like n8n drastically accelerate API integration
  • Early investment in error handling and retry logic pays off in stability
  • Tenant-specific endpoints (Operant’s Gatekeeper) require careful discovery and configuration

What’s next for Saleswolf

  • 🌐 Multi-channel outreach: Email, SMS, and in-app messaging agents
  • 📊 Advanced reporting: SLA dashboards and ROI forecasting
  • 📚 Human-in-the-loop feedback: Let reps tweak and fine-tune AI suggestions
  • 🔗 RAG integration: Pull in internal docs (playbooks, case studies) for richer context
  • ☁️ Multi-cloud support: Extend beyond AWS to Azure and GCP for broader applicability

Built With

Share this project:

Updates