What We're Building

Four products. Two people. AI-native.

We're building in public. Here's what we're working on, where each project stands, and what's coming next.

Running Now

Herald

AI-powered X/Twitter engagement agent

Herald monitors X for relevant conversations, drafts contextual replies, and routes them through Slack for human approval before posting. Two-tier LLM architecture: Haiku filters for relevance, Sonnet/Opus generates responses.

How it works

  • Monitors 50+ accounts in your industry
  • AI drafts replies matching your voice
  • Slack approval workflow—nothing posts without you

Results so far

  • Running our own @wildpinesai account
  • Human-in-the-loop, always
  • Consistent daily engagement

Posts assisted by AI, human-approved. We eat our own cooking.

Early but Running

AutoPundit

Document-to-video pipeline

Turn any document into a narrated video with AI avatar and Ken Burns-style B-roll. Claude writes the script, Fish Audio generates the voice, SadTalker animates the avatar. End-to-end automation.

The pipeline

  • Document in → Claude writes script
  • Fish Audio TTS with custom voice
  • SadTalker lip-sync avatar
  • Ken Burns B-roll compositing

By the numbers

  • ~$4-6 per video
  • 5 videos on our YouTube channel
  • Fully automated pipeline
Watch our videos on YouTube
Research

Pinecone

Our OpenClaw clone — autonomous agents on local hardware

Our own OpenClaw-style setup — autonomous agent teams running on local GPU hardware with open-source models. No API dependency, no per-token costs. Early research, building on our own cluster.

The approach

  • OpenClaw architecture on our own GPU cluster
  • Open-source models, fully self-hosted
  • Autonomous multi-agent coordination

Why local

  • Full control over the stack
  • No per-token costs at scale
  • Data never leaves the building

Early research. Learning what autonomous agent teams can do when you own the hardware.

Shipping Soon

Wild Companion

AI companion app

A privacy-first AI companion built with our Liquid Glass design system. Runs locally with Ollama, persistent memory, relationship tracking. Cross-platform: iOS, Android, Windows, Mac. Built end-to-end through our AI-native workflow.

What's built

  • Local-first with Ollama
  • Persistent memory across sessions
  • Liquid Glass design system

Why it matters

  • Privacy-first alternative to Character.AI
  • Your conversations stay on your device
  • Heading to the App Store

Built and nearly deployed. App Store submission coming soon.

Interested in what we're building?

Want to follow along or talk shop? Reach out.