TL;DR

Technical deep dive into how AI Bradaa designed the AB Model Family—AB Lite, AB 1.0, AB Pro, and ABO-84—with intelligent routing and the 84-Mentor Council.

The Problem with Generic AI

Most AI systems are built to be everything to everyone. A single model tries to handle quick casual conversations, deep research, coding assistance, and creative writing — often at the cost of depth, speed, or cost efficiency.

For Malaysian users, there is an additional layer: generic AI does not understand RM pricing context, EPF implications, makan place recommendations, or why saying "alamak" and "terrible" in the same sentence is perfectly natural in Manglish. The cultural layer is missing.

AI Bradaa designed the AB Model Family to solve this — four models, each built for a specific tier of capability.

The Four Models

AB Lite 1.0

Free

The daily companion

Key Features

  • 32K context
  • Optimized for speed
  • Free tier

Best For

Quick conversations, Q&A, casual interactions

AB 1.0

RM15/mo

Full reasoning model

Key Features

  • 128K context
  • Full memory
  • Mentor access

Best For

Serious work, daily use

AB Pro 1.0

RM39.99/mo

Deep research & multimodal

Key Features

  • 256K context
  • Image understanding
  • Document analysis

Best For

Research, voice I/O, image generation

ABO-84 1.0

RM39.99/mo

Flagship coding agent

Key Features

  • Full 84-Mentor Council
  • 50+ coding capabilities
  • Deep codebase understanding

Best For

Complex coding, demanding tasks

Intelligent Model Routing

The AI Pod sits at the center of every interaction. When you send a message to AI Bradaa, the Pod evaluates the query complexity, domain, context requirements, and your current Bond level to select the optimal model:

Simple Queries

→ AB Lite

Deep Research

→ AB Pro

Coding Tasks

→ ABO-84

This is not just cost optimization — it is about depth. A casual conversation should feel fast and light. A strategic planning session deserves full mentor council deliberation. The model selection is part of the experience, not just an infrastructure detail.

The 84-Mentor Council

Behind every response, a council of 84 specialized AI mentors contributes expertise:

11

Executive Mentors

Cross-domain reasoning — strategic planning, emotional intelligence, cultural context

73

Domain Experts

Finance, health, career, tech, relationships, legal, and more

When you ask about PTPTN repayment strategies, mentors with financial and Malaysian context weigh in. When you discuss career pivoting, career development and Malaysian workplace culture mentors contribute.

Malaysian Context as Default

The AB Model Family is trained with Malaysian context as a default layer:

  • Financial context — RM pricing, EPF, SST, PTPTN automatically understood
  • Linguistic respect — Manglish is valid, expressive, with its own grammar
  • Cultural depth — Religious considerations, regional variations (KL vs Penang vs Sabah vs Sarawak)

What's Next

The AB Model Family is actively in development:

ABO-84 enhanced coding capabilities

Expanded Malaysian domain expertise in mentor council

Refined intelligent routing based on real user interactions

The goal is not to build the biggest model. It is to build the right model for each situation — fast for casual chat, deep for research, precise for coding, thoughtful for life decisions.

This is a technical overview of the AB Model Family as it is currently in development. Capabilities, pricing, and availability reflect the current engineering state and may change as development progresses.

Was this helpful?