I build AI workforces
that never sleep.
Not demos. Not slideware. A live fleet of autonomous agents — trading desks, front-desk receptionists, research scouts, self-healing watchdogs — running in production right now, on hardware I run, in systems I wrote.
The fleet, by the numbers
Every figure below is measured from the running system — not rounded up from a pitch deck.
Three ways to put agents to work
The harsh, repetitive load in a business — answering, screening, watching, reacting — is exactly what an agent does best. I take that load off people.
AI Front Desk
A receptionist that never misses a call. Missed-call text-back, live web chat that knows your business, service tiers, lead capture straight to your phone.
- Answers instantly, 24/7 — nights, weekends, holidays
- Knows your services, prices, and hours
- Every lead logged and texted to you
Try the live public demo →
Custom AI Agents
Bespoke autonomous systems built for your operation: research scouts, monitoring watchdogs, report writers, Discord/Slack bots, data pipelines that police their own freshness.
- Designed around your workflow, not a template
- Self-supervising — restarts, alerts, kill-switches built in
- Delivered running, with a dashboard to watch it work
AI Resume & LinkedIn Forge
An AI-assisted rewrite factory for job hunters: ATS-tuned resumes, sharpened LinkedIn profiles, and cover letters that sound like you on your best day.
- Keyword-matched to the job you actually want
- Human-reviewed, fast turnaround
- Sold online — order from anywhere
The fleet — running right now
This is the part nobody else shows you, because nobody else has it running. A multi-desk autonomous trading research lab — Project ARIES — plus the command systems that keep it honest. Every green dot is a real process.
♈ ARIES Command
tailnetThe flagship landing deck: fleet-wide sword scoreboard, data-trust health, every strategy's live standing at a glance. One brain, many methods — hold, scalp, mean-revert, long/short — routed by market regime.
Open command deck →🏢 Fleet Office
tailnetA pixel-art "tiny office" where every agent is a character at a desk. Watch the fleet work like a living game world — because a dashboard you enjoy checking is a dashboard you actually check.
Walk into the office →🪙 Crypto Desk
tailnetThe crypto paper-trading R&D floor: multi-strategy swords, catalyst engine reading the news before the chart moves, cinematic HUD with ~1-second live push.
Open crypto desk →🎯 Sports Desk
tailnetOdds scanning, closing-line-value tracking, and a news-led catalyst brain that forms its own opinion on injuries, lineups and weather — strict and loose desks racing side by side.
Open sports desk →📱 Aries Mobile Command
tailnetA phone-first PWA to monitor and control the fleet from anywhere: live positions, health, and a token-guarded kill-switch that halts new entries while letting exits finish safely.
Open mobile command →📞 AI Front Desk Demo
publicThe customer-facing product, live on the open internet: type any business name and watch the receptionist adopt it on the spot, with real service tiers.
Try it — it's public →Note: cards tagged tailnet open only for devices on my private Tailscale network — by design. The Front Desk demo is fully public. Want a guided tour of the rest? Ask me.
The Method is the product
Anyone can wire an LLM to an API. The hard part is an agent you can trust unattended. Trading is my dojo — the harshest environment there is — and every discipline it forced gets baked into everything I ship.
Freshness is first-class
The #1 silent killer of autonomous systems is acting on stale data. Every component in my fleet declares a freshness SLA and passes through an is_fresh() gate — stale input fails to neutral, never to a bad decision.
A data-trust watchdog polices the fleet
A dedicated service continuously measures the freshness, liveness and sanity of every feed. It has caught zombie processes and crash loops that looked alive from the outside — before they could do damage.
Kill-switches and fail-safes, always
Feed-staleness halts, heartbeat fail-safes, regime circuit-breakers, and a manual kill-switch reachable from my phone. Agents get authority only inside hard guardrails.
Learn from every action
Every trade lands in a durable, append-only journal. A trap detector buckets outcomes by condition and flags recurring failure patterns — so the system stops re-falling into the same holes, permanently.
Edge is proven before it's trusted
No strategy graduates on a good week. It needs a clean sample, a bootstrap confidence interval above zero, and it has to beat the do-nothing benchmark. Honest scorecards — including the ugly ones — are the whole point.
Self-healing operations
Everything runs as supervised services with auto-restart, a maintenance agent that quarantines junk reversibly, Discord alerting to dedicated channels, and remote command from anywhere via a private mesh network.
The arsenal
The stack behind the fleet — every piece battle-tested in production, not just listed on a resume.
“The goal was never the bot. The goal is to lift the harshest load off people — with autonomous systems honest enough to be trusted with it.”