Inbound call
A caller dials the live Aetherlex line.
LIVE01 / LIVE SYSTEM
One call. Five production handoffs.
I connect the AI agent, phone system, server events, and messaging into one workflow. The result: caller needs become a structured owner summary and a consent-based confirmation for the caller.
Twilio
The number routes voice traffic into the SIP path.
LIVERetell AI
Elastic SIP trunk to the agent: Claude 5.5 Sonnet with the “Susan” voice.
LIVEWebhook
Call events reach the production server running on Render.
LIVEPost-call SMS
Follow-up is sent over an approved A2P 10DLC campaign.
VERIFIEDOwner summaryCaller, business, need, urgency, and next step.
Caller confirmationSent after SMS consent is captured on the call.
02 / HOW I LEARN & BUILD
Build. Test. Learn. Refine.
17
Agent versions shipped
I learn by building working systems, testing their behavior, and refining the details. Across 17 agent versions, I tuned response timing for natural turn-taking and refined the greeting to include recording disclosure.
HUMAN TEST / PASSED / OCT 1, 2026
A no-heat HVAC emergency, handled end to end.
I validate the complete workflow: how the agent handles a need, what the owner receives, and whether the caller gets confirmation. A human test using a no-heat HVAC emergency verified all three.
Emergency scenario handled correctlyPASS
Owner summary delivered accuratelyPASS
Caller confirmation deliveredPASS
03 / OPERATING RANGE
Building the system around the AI.
I work across AI behavior, integrations, and business operations. That means shaping the conversation, connecting the tools, and checking that each handoff produces the right result.
Voice AI & applied LLMs
I shape agent behavior through prompt iteration and live testing, then connect it to telephony and application logic. My work includes natural call behavior, SIP routing, webhooks, appointment-request capture, and post-call communication.
- Retell AI agent
- Twilio telephony
- Render production server
Automation systems
I use self-hosted n8n and Zapier to connect lead intake, scheduling, and follow-up processes. Four production workflows cover lead intake, booking sync, error handling, and suppression + reply intake.
- Website lead intake
- Booking sync
- Error handler
- Suppression + reply intake
From discovery to delivery
I run Aetherlex AI from technical discovery and live demos through intake, onboarding, and production operation. That experience connects technical implementation with how a business receives leads, brings clients on board, and follows up.
- Client intake via Formspree + email notifications
- Welcome email + onboarding checklist
- Technical discovery + live demos
Production validation
I check agent responses and delivered messages together, with consent and error handling built into the surrounding workflow. LIVE and VERIFIED mark the implementation and test status shown on this page.
- Inbound answering
- Owner summaries
- Caller confirmation SMS
- Appointment-request capture
04 / COMPLIANCE
Compliance is part of the architecture.
Not a footnote added after the system works.
A2P 10DLC brand + campaign
Approved sole-proprietor brand and messaging campaign for the SMS path.
SMS consent
Consent is captured on the call and recorded as “yes” before follow-up.
Recording disclosure
The disclosure is built into the greeting callers hear on the live agent.
05 / AI PORTFOLIO REEL
Creative work. An AI workflow behind it.
My mockup reel brings together AI-produced spec films, motion, and editing. It also shows how I use Claude Code to analyze my editing style and turn it into a repeatable creative workflow.
AI production + Claude Code-assisted workflows
Hiring someone to build useful AI?
I bring hands-on experience building AI agents, connecting automations, and testing business workflows. I’m looking for AI implementation, automation engineering, and applied AI roles where I can keep learning and ship useful systems.