Practical systems that make work feel lighter.

A podcaster wanted to stop scrambling before guest interviews. I helped him connect the tools already in his process so each new booking creates the prep, folders, tasks, tags, messages, and follow-up steps he needs.

• A Calendly booking starts the workflow automatically.
• Zapier routes each guest through the right path, creates files and folders, and keeps Trello, Kit, Gmail, Dropbox, and SMS in sync.
• The result is simple: he shows up prepared every single time without rebuilding the process by hand.

A podcast network wanted a better way to showcase its shows without asking every host to learn new tools. I built BPN Megafeed to gather episodes from many podcast feeds into one network feed, then built Clipfeed to turn new episodes into short, branded clips ready for podcast and social publishing.

• BPN Megafeed creates one public RSS feed from many existing podcast feeds while leaving every audio file on the original host, so downloads still count in the right place.
• Clipfeed watches the network feed, finds new episodes, creates transcript-backed clip options, adds short intro and outro bumpers, and publishes the selected audio clip to Transistor.
• The system also renders vertical videos with artwork, captions, and BE Podcast Network branding, then prepares them for social publishing.
• Behind the scenes it uses Supabase for workflow status, ElevenLabs for transcripts and voice bumpers, ffmpeg for clean audio cuts, Remotion for vertical video, Transistor for podcast publishing, and Postiz for Shorts publishing.
• Hosts and admins see a simple app where they can review episodes, suggest timestamps, approve clips, redo a run, or hold publishing. The AI stays in the background where it belongs.

Be Podcast Network needed its public site to stop depending on Elementor for core layout and daily updates. I rebuilt the site around a lightweight native WordPress theme and a custom BPN Network Manager plugin so design, show data, sponsors, intake, and homepage controls each had a stable place to live.

• The BPN Native theme now handles the public templates for the homepage, sponsor page, join page, contact page, application pages, posts, search, and 404 page without relying on Elementor-generated layouts.
• BPN Network Manager syncs shows from Transistor with pagination support, fixes unstable artwork URLs, and gives admins simple screens for homepage show ordering and publishing controls.
• Sponsors can move between current, past, and not-shown states, while self-serve sponsorship intake collects campaign details, uploads logos, creates pending sponsor records, sends notifications, and redirects to Stripe.
• Cloudflare Turnstile, validated Stripe redirects, removed hardcoded secrets, and safer upload rules made the sponsorship workflow less fragile and easier to operate.
• A BPN Megafeed landing page and safe homepage fallbacks turned backend cleanup into visible public-site improvements the team can keep maintaining.

Success Signal Analysis helps Kit creators turn recent broadcast data into evidence-backed marketing experiments. A visitor enters a Kit API key, names the outcome they care about, and gets a report that compares winning and losing campaigns instead of leaving them to stare at open and click rates alone.

• The app pulls recent Kit broadcasts, broadcast content, and stats server-side, then scores campaigns against the visitor's chosen goal.
• Two OpenAI passes separate the reasoning: one extracts structured campaign features, and the second writes an evidence-first report with signals to repeat, reduce, and test next.
• The workflow stays intentionally stateless. The visitor's Kit key is used for one analysis, never stored, and discarded after the report is generated.
• The finished report renders on screen, can be copied or downloaded as Markdown, and is emailed to the visitor for follow-up.
• The public GitHub repo includes the build notes, prompts, tutorial, and deployment shape so other builders can study how a focused AI tool moves from idea to shipped workflow.

A customer had 8 cameras tied to a closed cloud account, subscription-gated features, and a vendor-controlled app. I migrated the Wyze Cam v2 fleet to thingino open-source firmware, then connected the local streams into Apple Home so the customer could simply open the Home app on their iPhone and see their cameras.

• Every camera's factory firmware was backed up and SHA-256 verified before the migration, keeping a safe path back to stock firmware.
• The off-the-shelf installer had an incomplete image for this camera sensor variant, so I rebuilt the installer with the correct official image and verified it before migrating the fleet.
• The camera bootloader could not read large SDXC cards, so I isolated the card compatibility issue and standardized the install path on reliable smaller cards.
• Two half-installed, non-booting units were recovered with a self-flashing recovery card instead of replacing hardware, soldering, or using a chip programmer.
• The open RTSP/ONVIF streams became the integration point for Apple Home, preserving the simple iPhone experience while removing the cloud account and recurring software fee underneath it.
• The result: all 8 cameras run locally with no monthly software fee, no cloud dependency, no vendor lock-in, Apple Home access, and streams ready for an NVR, Home Assistant, Frigate, or AI vision layer.

Life Lab had important automations spread across scripts, webhooks, schedules, and credentials. I turned that scattered backend into one clear dashboard where the team can see what is running, what failed, and exactly what to do next.

• Scheduled workflows run automatically, and incoming webhooks land in one secure place.
• Run history, logs, API credentials, and failure alerts are managed inside the dashboard instead of across disconnected tools.
• When something breaks, the team gets an email with a link to the failed run and a one-click Copy for Claude button that packages the log and source code for fast diagnosis.
• It runs on Cloudflare Workers, D1, R2, KV, Durable Objects, and Queues, keeping the system lightweight and inexpensive while still giving Life Lab a real operations command center.

Darrell was carrying the stress of a driving school's back office while trying to focus on teaching teenagers how to drive. I made the AI disappear into the workflow so he could run the school with fewer clicks, fewer reminders, and less computer anxiety.

• One click creates a new class cohort, pricing page, Stripe product, and CRM smart list.
• Automated emails keep students informed about upcoming classes and required information.
• The system handles the repetitive office work in the background so Darrell can spend his attention on students, not software.