Question 1

What is the Portfolio AI System?

Accepted Answer

The Portfolio AI System is three interconnected AI systems that run together as a business development pipeline: Phase 1 is a RAG-powered chatbot that answers questions and builds context, Phase 2 is a multi-agent proposal generator that produces a BRIDGE-framework business case in 90 seconds and captures the lead, and Phase 3 is a LangGraph autonomous lead follow-up pipeline that researches and scores each lead, drafts personalised follow-up emails, and places VAPI voice calls to leads scoring 60 or above. All three systems run on Wandile Lokwe's own business development pipeline at wandilelokwe.co.za.

Question 2

What does Phase 1 (RAG chatbot) include?

Accepted Answer

Phase 1 includes a LangChain RAG pipeline with ChromaDB vector store, Anthropic Claude API streaming responses, document upload for brief and specification injection, conversation analytics tracking messages/tokens/response times, LangSmith tracing on all retrieval calls, and a Gradio or Next.js UI. Delivered at R45,000–R85,000 depending on scope.

Question 3

Can I buy just one phase or do I need the full system?

Accepted Answer

Each phase is a standalone deliverable. You can commission Phase 1 only, Phase 1 and 2, or the full three-phase system. Each phase is independently functional and adds value. The full system creates a complete pipeline from visitor arrival to autonomous voice follow-up, but the phases are designed to be deployed incrementally.

Question 4

How much does each phase cost?

Accepted Answer

Phase 1 (RAG chatbot): R45,000–R85,000. Phase 2 (multi-agent proposal generator): R65,000–R120,000. Phase 3 (LangGraph lead pipeline): R85,000–R160,000. Full system (all three): R180,000–R320,000. Monthly retainer for maintenance and monitoring: R15,000–R25,000/month. Payment terms are 40/40/20 on all engagements.

Question 5

What is LangGraph and why does it matter?

Accepted Answer

LangGraph is a state machine framework for building multi-step AI pipelines where each step depends on the results of previous steps. It is used in Phase 3 to orchestrate the lead follow-up pipeline: the outer state machine manages the overall flow, an intelligence subgraph researches each prospect and assigns a lead score from 0 to 100, a communications subgraph drafts personalised follow-up content, and the routing logic decides whether to send an email, place a VAPI voice call (for leads scoring 60+), or schedule a follow-up check. LangSmith traces the entire pipeline.

Question 6

What are the payment terms?

Accepted Answer

40/40/20. 40% on scope sign-off. 40% on delivery. 20% on go-live. You never pay in full upfront. A clear scope document is prepared before any build commitment. The scope call is free.

Question 7

Is the system demonstrated on a live site?

Accepted Answer

Yes. Phases 1 and 2 are live at wandilelokwe.co.za — the RAG alter ego chatbot and the 4-agent proposal generator both run in production on Wandile Lokwe's own business development pipeline. Technical evaluators can verify every architecture claim via the LangSmith traces included with each proposal run.

A live AI portfolio that proves capability and captures leads autonomously.

Visitor arrives. Business case delivered. Lead followed up autonomously.

Each phase is a standalone deliverable. All three form a complete pipeline.

RAG-Powered Chatbot

Multi-Agent Proposal Generator

LangGraph Lead Follow-Up Pipeline

All Three Phases

A portfolio that generates leads and proves capability simultaneously.

Common questions

Talk to Wandile directly