Small tools built because something kept wasting time.

Prospecting was split across scrapers, spreadsheets, enrichment steps, GHL handoffs, and manual follow-up.

A prospecting system designed to identify leads, enrich records, support outbound AI phone-call experiments, and hand useful prospects back into GoHighLevel.

Feeds agency outbound experiments by moving scraped and enriched prospects toward GHL, AI phone-call tests, and follow-up workflows.

Cold email worked only when list building, inboxes, send pacing, replies, suppressions, and offer testing were treated as infrastructure.

A code-managed outbound layer for lists, verification, warmed inbox rotation, reply triage, deliverability notes, and campaign learning loops.

Tests lists, offers, inbox strategy, and follow-up logic for Heaviside agencies while turning the lessons into DAGA training material.

Manual local-business list building was too slow for niche market testing.

National-scale Google Maps scraping and deduping for paving, garage door, and other home-service lists.

Creates the raw market and prospect lists that feed outbound, competitive research, niche launch decisions, and CRM enrichment.

Scraped local-business lists were useful only if they could be cleaned, deduped, mapped to CRM records, and handed to a follow-up system.

A lead-gen platform and GoHighLevel sync companion built around Apify/Google Maps data, dedupe logic, and CRM handoff patterns.

Preserved the path from raw local-market data to usable CRM motion, even after the Supabase-backed version was paused for cost.

Ad strategy was too dependent on staring at Meta's ad library by hand and trying to remember which hooks kept showing up.

A competitor ad review loop that scans ads, extracts offers and landing pages, groups formats, and feeds new creative tests.

Turns public competitor ads into concrete creative refresh decisions for PVM, GDM, HG PPC, and future vertical launch research.

Google and Meta dashboards show recent performance. They do not explain whether an account is deteriorating against years of vertical data.

A vertical-labeled warehouse and scoring engine across 165+ ad accounts with daily feature views, risk scores, and threshold alerts.

Gives HG PPC and MarketingOS a benchmark layer so account reviews can compare current performance against real vertical history.

Lead forms and CRM events were not reliable enough as ad-platform conversion truth.

A GHL to GTM Server to Facebook CAPI / GA4 relay so booked calls and conversion events can be tracked with less guesswork.

Repairs the conversion-truth path for niche ad accounts so Meta and Google optimize against cleaner downstream events.

SEO and paid-media tools needed a safer, shared way to call DataForSEO without every app owning its own brittle integration.

A Go gateway with HMAC auth, rate limiting, Redis caching, and circuit-breaker protection around DataForSEO calls.

Provides a reusable data-access layer for market research and SEO tooling without scattering API logic.

Aged and expired domains needed a safer way to be sourced, scored, filtered, and grouped by niche before any SEO infrastructure decision.

A domain sourcing and vetting pipeline for evaluating domains by niche relevance, risk, history, and usefulness for private local SEO infrastructure.

Supports local SEO infrastructure decisions by vetting domains before they enter niche clusters or become part of any owned-media buildout.

Citation work gets messy fast when dozens of businesses need consistent NAP data across 100+ directories.

A local business citation tracker with directory monitoring, scraping, and consistency checks.

Keeps citation cleanup measurable across vertical brands instead of letting directory work live in spreadsheets and vendor promises.

Each vertical needed owned media that could rank for trade plus city searches and feed leads back to the agencies.

A Go + PostGIS directory engine powering pavinglist.com, garagedoorlist.com, and electricianslist.com.

Creates owned local-search properties that can rank, structure market data, and feed leads or research back into the niche agencies.

Google Business Profile work gets hard to govern when multiple locations, posts, updates, and service areas are spread across accounts.

A GBP management layer for multi-location work across Heaviside-managed businesses.

Keeps Google Business Profile work connected to the broader SEO workflow.

Paving websites needed trade-specific features like project galleries, service calculators, and conversion tools without rebuilding them every time.

A paving-specific WordPress plugin with project galleries, service calculators, and AI-chat concepts.

Captured vertical feature ideas that later informed PavingOS, HG Websites, and niche website templates.

Website rebuilds needed a faster way to capture existing public-site structure, styling, and design tokens before moving into Next.js.

A CLI concept for converting a public website into a Next.js app using computed-style capture and design-token extraction.

Informed the website-platform and clone-generation thinking behind HG Websites and HG Web Commander.

Video creative got expensive at the same time the ad accounts needed more testable variations.

A Next.js and Remotion video-generation pipeline with HeyGen, voice, captions, still generation, local GPU options, and programmatic rendering.

Turns scripts, offer angles, and campaign inputs into reusable video assets for personal proof, PVM/GDM creative, and future paid-social testing.

Local AI experiments needed a usable front door instead of scattered scripts on a homelab GPU box.

A browser-based hub for local LLM and video generation on an RTX 3090.

Gives local models and GPU video experiments a front door that can be reused by content, creative, and infrastructure workflows.

Client-funnel QA and competitor checks needed isolated sessions, controlled geography, and repeatable browser automation.

Tailnet-only Chromium with isolated fingerprint, residential proxy support, FastAPI, and Playwright.

Lets agency systems inspect funnels, local search results, and competitor experiences from controlled sessions instead of personal browsers.

Coolify, queues, databases, VPS servers, home-network gear, and agent jobs were spread across too many dashboards.

A self-hosted dashboard for deployments, BullMQ queues, PostgreSQL health, Hermes jobs, and home-network status.

Acts as the status layer for the Heaviside infra ecosystem, making deployments, queues, databases, and agent jobs visible in one place.

Task extraction and follow-up ideas were scattered across notes, inboxes, meetings, and chat logs.

A PWA task manager that pulls action items out of notes and chat, with an Android-capable Capacitor shell.

Became part of the broader pattern: capture work where it appears, hand it to the right place, and keep the owner out of manual triage.

Every serious local or VPS project needed secrets loaded safely without copying raw credentials into notes, code, or shell history.

A Bitwarden Secrets Manager initializer and repeatable helper pattern for project-scoped secret injection.

Supports deployments, API integrations, MCP servers, and agent workflows without leaking tokens into notes, code, or shell history.

Agents needed access to where the work actually happens: email, calendar, files, tasks, and Microsoft Graph workflows.

One of the first 100 published OpenClaw skills, built when OpenClaw was still called ClawdBot, connecting Microsoft 365 workflows into an early agent toolchain.

Shows early hands-on work in agent tooling by connecting everyday Microsoft 365 work to an emerging open agent ecosystem.

AI agents are only useful when they can reach the systems where work already happens.

MCP servers and wrappers for MainWP, Meta Ads, GoHighLevel, x-accel, DAGA, and related business APIs.

Lets agents operate inside real business tools instead of sitting outside the workflow as a chat interface.

Agent orchestration needs a place where jobs, workers, queues, and status are inspectable.

An open-source fork of an agent orchestration dashboard with a broad panel set for monitoring and control.

Provides UI patterns for the agent fleet, Infra Dashboard, and Open Agents work.

A real-time freight-rail monitoring idea needed a way to visualize movement and public rail signals.

A monitoring and visualization prototype using real-time streams, WebSockets, and a React interface.

Preserved as a data-product experiment and a reusable pattern for map-like dashboards.

Daily Catholic family prayer and liturgical content needed a simpler home-screen rhythm.

A personal app concept for daily Catholic family devotions and liturgical material.

Personal prototype that sharpened recurring-content, daily-state, and mobile-app patterns outside the agency context.

Earlier agency operations needed WordPress management, SEO task handling, and multi-tenant control before the current HeavisideOS stack existed.

Earlier platform and SEO tooling that later gave way to HG Web Commander, HG SEO Commander, and the broader MarketingOS stack.

Acts as the fossil record of the current system: useful ideas were absorbed, stale tools were retired, and the operating core got cleaner.