Makix Chat

Connection & Credentials

Choose your AI provider, or enter custom credentials manually.

Connection Profiles

Switch providers without losing your API key, models, or limits. The fields below edit the active profile.

Credentials

API Key Base URL

Model Engine Settings

Retrieve and configure your preferred models for each modality.

Language & Reasoning

Conversational LLM & Planning Engine

Chat Model

Powers conversational replies, reasoning logs, agent steps, and goals.

Image Generation

Visual Asset Creation

Image Model

Generates high-fidelity artwork and assets from description prompts.

Text to Speech

Voice Synthesis Settings

TTS Model

TTS Voice

Converts responses into realistic audio. Set target voice (e.g. alloy, echo, fable, onyx, nova, shimmer).

Speech to Text

Transcription Settings

ASR Model

Streaming transcription Show live partial transcripts.

Translates your voice recordings and speech inputs into text messages.

Agent Capabilities

Manage native device capabilities and remote MCP tools accessible by the AI.

MCP Servers

Connect one or more Model Context Protocol servers. MCP is independent of your AI provider and shared across all connection profiles.

AI Engine Orchestration

Configure Zeagent's native agentic and orchestration engine features.

Agentic Engine

Goal Planning Allows the AI to formulate step-by-step plans to resolve complex requests.

Goal Verification Automatically validates outputs against original goals and retries on errors.

Continuation Planning Allows the agent to chain multiple conversation turns autonomously if needed.

Parallel Tool Calls Enable concurrent execution of multiple tools when actions are independent.

MetaRouter Intent Classifier Dynamically detects intent and registers relevant tools and skills per turn.

Goal Injection Position

Where the decomposed goal block is placed. Only applies when Goal Planning is on.

Interaction

Tool Confirm Haptic Vibration Vibrate when a sensitive tool call requires your confirmation.

Response Audio Notification Play an audio notification chime when the assistant completes a response.

Inline Autocomplete Suggest the rest of your message as greyed-out ghost text while you type — tap the suggestion (or press Tab) to accept. Uses a small AI call per typing pause.

Engine Limits & Budgets

Max Iterations Max Context Tokens Max Completion Tokens Max Tool Calls Per Session Max Calls Per Individual Tool

Persona & Appearance

Configure your user identity, styling hints, and application theme.

Your Name Toulouse, France

Profile

Your Name Your City Style Hints (tone only · 100 chars max)

Appearance

Dark mode Switch between light and dark appearance.

Collapse stats Collapse performance metrics badges by default.

Data management

Permanently remove every chat session from this device.

Long-Term Memory

Long-term memory lets the assistant remember durable facts across chats fully on-device.

Enable Long-Term Memory Recall relevant memories into context each turn and expose the remember / recall / forget tools.

Auto-Reflection Extract durable facts at the end of each turn. Adds one LLM call per turn — off by default.

Recall Budget

Recall Token Budget (caps the injected memory block)

Stored Memories 0

Access & Security

Manage application lock credentials, unlock PIN, and endpoint verification.

App Lock

Require PIN on launch Show the PIN unlock screen when the app starts. Turn off to open the app directly without entering a PIN.

Warn on untrusted endpoint Confirm before sending messages to the LLM if the API endpoint is not a secured Makix node.

Change PIN Password

For convenience security, you can change your 6-digit unlock PIN here.

Current PIN

New PIN

Confirm New PIN

Lock Session

Immediately lock the application. You will need to enter your PIN to gain access again.

About Makix Chat

Application version, autonomous system properties, and author details.

Makix Chat

Autonomous Multi-Agent Assistant

v1.2.0 Production

ReAct Loops

Dynamic reasoning-acting loop to select and execute tools autonomously.

Vision Analyzer

Understand context, documents, and scenes through image analysis.

Voice & Audio

Integrated low-latency ASR speech recording and custom TTS playback.

Cordova Bridge

Access native sensors: Contacts, SMS registry, Calendar and Camera securely.

System Diagnostic Details

Active Model gpt-4o

Platform Context Cordova / Android

Context Window Limit 12K Tokens Cache

Engine API Status Connected

How can I help you today?

Welcome back