Featured system

DavosBot

A personal AI operating system living inside iMessage.

DavosBot started as a simple assistant and evolved into a text-native AI operating system: reminders, scheduled messages, memory search, sports intelligence, command routing, personas, and automations — all reachable from the place I already communicate.

Statusshipping
InterfaceiMessage
CorePython
MemorySQLite
RuntimeMac Mini / PM2
ModelsLocal + cloud

Personal Assistant

Reminders, scheduled texts, daily briefings, todos, and accountability.

AI Agent Layer

Natural language commands, reasoning, summaries, and workflow execution.

Memory

Persistent context for people, preferences, projects, and recurring tasks.

Sports Intelligence

Scores, team tracking, fight data, betting context, and sports workflows.

Group Chat Mode

Shared reminders, bot mentions, polls, and group automation concepts.

Admin + Control

Permission-aware commands, protected actions, logging, and reliability checks.

Personas

Custom assistant personalities for different use cases and tones.

Life OS

Habits, workouts, routines, changelogs, and personal operating cadence.

Architecture

Small production system, text-native interface.

DavosBot is built like a small production system: persistent memory, command parsing, model routing, scheduled jobs, process monitoring, and a text interface that makes the whole thing feel invisible.

iMessagetext interface
AppleScript / Message Listenerinbound capture
Message Routerintent + permissions
DavosBot CorePython orchestration
Memory + Commands + Automations + AI Modelsworkflow layer
Response back to iMessagetext-native output
Command palette

Example command simulator

Click a command to see the kind of response DavosBot could return. These are example responses, not live calls.

DAVOSBOT RESPONSE

Reminder created for Friday.

command parsed permissions checked workflow routed
Control plane

Built for personal control.

DavosBot is designed around personal infrastructure, local-first workflows where practical, and explicit control over sensitive actions.

Personal infrastructure

Runs around tools and infrastructure I control.

Local-first where practical

Uses local workflows when they make sense for speed and privacy.

Permission-aware commands

Sensitive actions should be explicit, logged, and protected.

Protected admin features

Admin-style commands should be separated from normal chat behavior.

No unnecessary app switching

The interface is text, because that is where the workflow already happens.

Real daily use

Designed around actual reminders, messages, sports, planning, and routines.

Projects

Portfolio

Data · Sports

UFC Fight Database

A searchable fight database with thousands of UFC fights, filters, and structured sports data exploration.

Python JavaScript Sports Data Netlify
Travel · AI

TravelLens

AI-powered travel planning and discovery for destinations, itineraries, and exploration.

React Vite AI Travel