Automate the reasoning. Stop before the send.

01 / Measurement

A tracker that prices its own next move before making it

Brands don't know whether Google's AI Overview cites them when someone asks a category question — or which competitor gets cited instead. Nothing on the market prices out that answer before running it.

How it works

1. Makes one test call to DataForSEO's AI Overview endpoint and reads the real cost back from the response — not an estimate.
2. Projects the full run's total from that single data point, before spending another cent.
3. Refuses to run the remaining prompts if the projection would cross the $3.00 cap — cuts prompts and says so, rather than overspending silently.
4. Runs the set — unbranded discovery, branded comparison, and problem-led prompts — and scores presence rate, citation rate, and a competitor/domain leaderboard.
5. Surfaces the citation gap list: the specific prompts where a competitor gets cited and the brand doesn't. That list is the actual product.

The live demo runs four brands measured independently: Cocokind, Hospitable, All In Solutions Counseling Center, and Thryv Marketing Solutions. Every number traces to a stored DataForSEO response. This was an independent measurement exercise built as an application artifact — not client work, and not an engagement with any brand shown.

PythonDataForSEO APIGoogle AI Overview dataSelf-contained HTML report

02 / Outreach

An outreach engine that drafts the message and stops at the send

Most outbound tools automate the sending. This one automates the judgment before it — finding a real reason to reach out — and requires a yes from a human before anything leaves the building.

How it works

1. Local discovery: searches Maps/GBP by radius and category for businesses in a given market.
2. Enrichment: pulls site, reviews, phone, and social presence for each business found.
3. Hook detection: flags a real reason to reach out — dead website, no reviews, a ranking gap, or no phone listed.
4. Drafts an opener from the hook it found.
5. Holds at a human yes/no approval gate — nothing sends or gets tracked until a person clears it.

PythonGoogle Maps / GBP dataOpen source, MIT licensed

03 / Client reporting

A report generator that drafts, but never signs its own name

Raw security event exports are unreadable to the non-technical client paying for the service. Someone still has to turn CSV rows into a report a business owner will actually read and trust.

How it works

1. Ingests raw security event exports, CSV or JSON, from whatever the client's tools produce.
2. Computes a 0–100 security posture score against a documented, transparent formula.
3. Builds metric tiles, a threat breakdown chart, and a top-targeted-users list.
4. Writes a plain-English executive summary aimed at a non-technical reader.
5. Stops at a DRAFT with a CSM signature line — no automatic send to the client.

PythonCSV / JSON parsingRuns offline, no external calls