T
TEKREIGN
KAMPALO — Agentic AI Marketing Platform (Kai)
AI Marketing Technology

KAMPALO — Agentic AI Marketing Platform (Kai)

AI marketing analytics with Kai—a supervisor plus specialist mini-agents stack for Google Ads, Meta, search, and cross-channel insights

Overview

KAMPALO is TekReign’s AI-based marketing platform. At its core is Kai—an agentic supervisor plus specialist mini-agents stack that answers marketing questions over synced campaign data. Teams get unified analytics for Google Ads and Meta, plus conversational AI that runs tools, applies deterministic ROAS/CTR/CPC decision logic, and synthesizes clear answers.

Project Details

Client

TekReign Product

Duration

Ongoing product

Category

AI Marketing Technology

Live ProjectVisit Live Project

Project Gallery

Key Results

Agentic Kai Chat
Hybrid ROAS Decisions
Google & Meta Suites

The Challenge

Marketers juggle multiple ad platforms and reporting tools. Generic chatbots invent metrics; raw dashboards don’t explain winners. Teams needed an AI that could query synced stats safely, specialize by domain (ads, analytics, search), and combine rule-based decisions with LLM narrative—not live ad-API calls inside every chat turn.

Our Solution

We shipped KAMPALO with Kai’s production agent stack: • Supervisor + specialist mini-agents (LangGraph): decode → parallel suite agents → synthesize • Suites: search_engine, analytics, google_ads, meta_ads, meta_insights • Agents use DB-only tools on synced stats (no live Google/Meta calls in chat) • Hybrid pipeline: tools → structured findings → deterministic decision engine (ROAS/CTR/CPC winners); LLM synthesizes mixed-domain answers • Curriculum maps known Kai prompts before the supervisor so smaller local models stay in-scope • LLM providers via AILLMService → LangChain ChatOpenAI: Ollama (default, e.g. qwen3.6) or optional Groq • MCP (FastMCP) exposes the same namespaced suite tools in-process and over HTTP for external clients • Product surfaces: agent query/chat (HTTP + SSE streaming), WebSocket chat, conversation history in Postgres, plus rule-based trends/compare/insights analytics

The Results

KAMPALO gives marketing teams an agentic operating layer on top of synced data: • Ask Kai in chat and get suite-specialized analysis across Google and Meta • Faster “what’s winning?” answers grounded in tools + deterministic ROAS/CTR/CPC logic • Safer chats that read synced DB stats instead of hammering live ad APIs • Extensible tools via MCP for API and external agent clients • Live product at kampalo.com with streaming and conversation history

Technologies Used

LangGraphLangChainKai (supervisor + mini-agents)Ollama / Groq LLMsFastMCP / MCPPostgreSQLSSE + WebSocket chatGoogle Ads & Meta (synced)Decision engine (ROAS/CTR/CPC)
Kai’s supervisor and suite agents turn synced ad data into clear winners—tools and ROAS logic first, then a clean AI summary when we need the full picture.

Marketing Operations Lead

KAMPALO user