Case studies

Working demos of what we build for clients, and systems we have built.

What we build for clients

Working demos for trades shops and local businesses. Every business, price and phone number is fictional.

Aldermoor Water
Water filtrationAnswer every call packageBuilt example, fictional business

A website and an AI receptionist, working together.

The site writes down every installed price and gives an instant estimate. Calls the shop can't take are answered by the AI receptionist, which texts the owner the lead.

Systems we have built

QUANTMECHANIX

An options research desk that shows its reasons.

Ideas run on paper first and are scored once they settle.

See how QuantMechanix works

Built for GreenRoom Technologies Inc.V1 liveV2 in private beta release

Professional network for the music industry.

An AI assistant that finds collaborators from member credits and connections, and shows why they fit.

Read the GreenRoom case study

SGNL

An AI operator that finds the few things only you can do.

It verifies work across your tools, then returns 3 to 5 moves for the next 18 hours.

See what SGNL is building

The experience behind the firm

Moving high-volume financial data and cutting query time for the research built on it.

  • 6,500+companies' ESG data in the Kafka migration
  • 45%lower query latency
How it was built

ESG company data

  1. FromESG data on 6,500+ companies
  2. BuiltA Kafka migration
  3. ResultSame data contracts, 45% lower query latency

Bank filings

  1. FromBank regulatory filings
  2. BuiltNormalized ingestion pipelines
  3. Used byCredit analysis in the Terminal

Turning data onboarding and model calibration from bottlenecks into repeatable systems.

  • 85%faster source onboarding
  • 100M+parameter search space
  • 2 hrscalibration, down from 10 to 16 days
How it was built

Vendor data

  1. FromBloomberg, Moody's, ICE, MMD
  2. BuiltA Snowflake warehouse on Data Vault 2.0
  3. ResultFirm-wide credit datasets, with new sources onboarded 85% faster

Model calibration

  1. From100M+ parameter combinations
  2. BuiltA distributed grid search on AWS ECS
  3. ResultAbout two hours, down from 10 to 16 days

Start with fifteen minutes.

A free call. You leave knowing which engagement fits, or that none does.

Book a call