Skip to content
Sweet Gum.
Tools

Internal tools we build to do the work better.

We automate our own workflows before we automate anyone else's. These are the tools we run in-house for research, scraping, video and publishing. They are not products and are not available to third parties.

Keyword research tool

A browser extension that shows Google Keyword Planner metrics (monthly search volume, competition, and cost per click) directly alongside Google search results while we research a topic. It connects to the Google Ads API on behalf of our own Google Ads account and is used to plan content and advertising for Sweet Gum Solutions and the recommendations we make to clients.

The tool is read-only, is used only by our own team, and does not access any client's Google Ads account. See our privacy policy for how it handles data from Google APIs.

Research

Know the market before we build

Every site starts with data on what people search for, who already ranks and whether the market is worth it.

Market validator

Type a service and a city and it pulls live Google results and map pack data, layers in domain authority, and returns a score, a grade and a plain-English verdict on whether that market can pay. It can sweep hundreds of city and service combinations in one run.

Market scoring engine

Scores every city and service pair in a state from structural signals, then audits the survivors against live search and map data. Markets that pass become a site spec. It is how we picked where our tree and concrete sites live.

Lead list builder

Finds businesses by category and area, enriches each one from its website and public records, scores it, and holds it for a human review before delivery as a CSV or a CRM push. A shared catalog means the same business is never scraped twice.

Scrapers

Turn the web into research we can use

Polite, rate-limited collectors that respect robots.txt and only touch content we have legitimate access to.

Ranking site harvester

Crawls sites that already rank in a niche and reverse-engineers their page structure, section lengths and topic vocabulary into a reusable spec. Collection and analysis are separate programs, and no copied text ever makes it into a spec.

Skool community intake

Pulls every lesson from a Skool community we belong to, including videos, attached files and linked posts, transcribes the video locally and captures a screenshot at every on-screen change. The output pairs what was said with what was on screen, ready for an AI to read.

Circle community intake

The same intake for Circle communities, which each run on their own domain. It reads Circle's own data feed instead of scraping pages, keeps lessons in the creator's order and links each line of transcript to the screenshot on screen at that moment.

YouTube research intake

Pulls transcripts and details from YouTube videos so we can research a topic, a competitor or a format in minutes instead of watching hours of footage.

Video

Video without a film crew

The same pipelines behind our own channels, used for client explainers, ads and social clips.

Video production pipeline

Script, a real human voiceover, a timestamped transcript, one image per line, optional motion and sound effects, then automatic assembly into a finished video. The long version is cut into vertical shorts with burned-in captions.

Faceless video studio

A browser app that walks from idea to finished video on one screen: define the channel's style, record the voiceover, review a storyboard, then render. Every step has an approval gate and a live cost meter, and a short clip costs about ten cents to produce.

Build and publish

Ship sites and content on a schedule

Generators and publishing pipelines that keep sites fast, consistent and fresh.

Static site generator

Turns a site spec into a fast static local service site: home page, one page per service, one per town served, with schema and internal links built in. It also builds preview sites that show a business owner what their site could look like before we talk.

Automated publishing

Researches, writes, cross-links, adds structured data to and publishes a new article on a schedule, then deploys the site. A person reviews the output. It runs twice a week on our own sites.

AI operations

Keep the AI honest and the costs low

Small tools that make the AI parts of our work more reliable.

Second opinion council

Sends a hard question to several competing AI models and brings back their answers to compare, so one model's blind spot doesn't become our mistake. Free models go first and a paid model backs them up, so most checks cost nothing.

TokenTrim

A small, fast local program that compacts large data exports and logs before they reach an AI, cutting the size by roughly 40 to 58 percent with no loss of information.

Want tools like these working for your business?

We build the same kind of research, automation and video systems for small businesses.