Skip to content

// THE WORK

Real work. Real numbers.

We don't show you polished case studies. We show you what we actually built, what broke, and what the numbers look like.

LegalDormant

Running daily, spending money, writing zero rows.

A legal intelligence platform ran daily, spent money and wrote zero database rows. Wrong data source, wrong endpoints, wrong cost model. We rebuilt all three.

5

Root-cause bugs found

3

Counties covered after rebuild

0

Rows before the rebuild

Read the full case study →

DistributionProduction

101,862 properties. We found our own waste and reported it.

Broken dedup queries and backwards name parsing. We documented every mistake and reported our own waste to the client.

101,862

Properties in the database

~40%

Share enriched

In production

Status

Read the full case study →

EnergyProduction

77% of ZIP codes over a federal legal limit.

Across roughly 300 Connecticut ZIP codes, 77% had a contaminant above its federal legal limit. The version that worked stopped leading with the technology.

~300

Connecticut ZIP codes analyzed

77%

ZIPs over a federal legal limit

233 of 301

Recount, independently

Read the full case study →

Enterprise SecurityOngoing

Country-level risk, tied to enterprise exposure.

HPE needed to know whether their people and facilities were safe, anywhere in the world, at any time. We built the intelligence layer that makes that possible.

Worldwide

Coverage

47

Active data sources

Government and open feeds

Source basis

Read the full case study →

InsuranceOngoing

Five platforms. One operational picture.

Data scattered across five systems. Zero unified customer view. We built the operational picture, and then stayed to build everything else.

5

Data sources unified

None

Unified customer view before

Ongoing

Status

Read the full case study →

Safety TrainingNot deployed

Years of training delivered. Zero digital presence to show for it.

A behavioral safety firm was invisible in search and AI answers. We built the edge SEO fix. It was never deployed, and this page says why.

Edge worker + schema

Delivered

Multiple

Category rivals cited by AI

DNS migration

What stranded it

Read the full case study →

// WHAT WE SHIPPED

Things we built. Go check them.

Case studies are easy to write. These are products you can open and check right now, including the one where we caught ourselves publishing data that wasn't real.

ProjectKids

Consumer marketplace, built and operated by us

An activity and camp platform that scans program sites daily and turns them into a searchable, bookable directory across several metros.

It is also where we graded our own data: a confidence spine that scores every listing by how it was obtained, scraped, inherited, or model-derived, and serves it by trust tier instead of quietly hiding the weaker rows. That work established how many listings were genuinely bookable, meaning a real future session and a real price: 906 of 1,290.

projectkids.io

Edwin

Local-first AI memory, in pre-release

A memory layer for the AI tools you already use. It stores each fact with the date it was true and the reasoning behind decisions, checks new facts against what it already holds, and keeps the store on your own machine. Not yet generally available.

Its benchmark is pre-registered and self-run, with the harness, judge prompts and per-call cost ledger published. It reports the result that does not flatter it: on accuracy, Edwin tied simply giving the model the full history. Its case is cost, and the site says so.

heyedwin.io