Work  ·  Patterns & shapes

The shapes of systems we ship. No client names.

We don’t publish a logos wall or a case-study deck on this site. Most of the work is internal, sensitive, or under NDA — and the parts that aren’t belong to our clients to talk about, not us. Instead, here are the patterns we keep building. If one resembles a problem on your desk, that’s the conversation to start.

68
Weeks, typical build engagement
94%
On budget, on date
100%
Engagements include literacy & handoff
30d
Post-launch support, every project
Filter
structured-extraction · illustrative Queue Review Audit
intake_2841 · standard · 12 pages extracted 87s · 38/38 fields
SubmitterField 010.99
Amount$2,840.000.98
Term12 months0.99
Effective2026-08-010.99
Variationindexed +1.5%0.94
Edge clauseSee section 14(c)human
Add-ons1 unit · included0.96
Hold-back$2,840 deferredhuman
SHAPE 01   Structured extraction  ·  4–6 weeks

Pull structured data out of long, messy documents.

The workhorse pattern. Replace a multi-hour manual review with a 90-second extraction agent that scores its own confidence and routes the edge cases to a human. Evaluated against your real historical documents before it touches production.

93–96%
Typical field-level accuracy
~90s
Avg time per document
5–12%
Routed to a human reviewer
1.2k
Typical evaluation set size
Claude Sonnet 4.5 pgvector Eval harness Human-in-the-loop

SHAPE 02

Classification & routing for high-volume intake.

Classifies inbound calls, forms, and emails into a fixed taxonomy with a confidence floor and a clean human-in-the-loop on edge cases. The kind of work a junior used to do all day.
Build  ·  2–4 wks  ·  in production
SHAPE 03

Internal Q&A grounded in your real documents.

A chatbot that answers from your policies, handbooks, and historical tickets — with citations, a confidence floor, and a clean handoff when the question is outside the corpus.
Build  ·  3–6 wks  ·  in production
SHAPE 04

Workflow automation across the tools you already pay for.

Email, docs, CRM, Slack, and the spreadsheet that nobody admits is load-bearing — stitched together with n8n, Make, or custom code, with LLM steps only where they earn their keep.
Automate  ·  2–4 wks  ·  n8n / Make / code
SHAPE 05

AI-aware product surfaces and websites.

Marketing sites, internal tools, and product surfaces designed and built to work with the AI underneath — not bolted on after launch. Real components, real states, real handoff.
Web  ·  3–6 wks  ·  design + dev
SHAPE 06

AI literacy programs tuned to your team’s real work.

Half-day to multi-week workshops built around your tools, your docs, and the actual workflow we shipped. Plain English. Role-specific quick-start guides. Practice on real artifacts, not toy prompts.
Enable  ·  half-day to 6 wks  ·  on-site or remote
Why no logos wall

A note on discretion.

Most of the systems we’ve shipped sit inside operations, presales, legal, and customer-support stacks — the places a company doesn’t advertise because their competitive edge lives there. We’d rather have the long, repeated, referral-driven relationship than the screenshot.

On an intro call we’ll happily walk you through specific engagements, with named clients and real numbers, under a mutual NDA. What lives on this site is the part we’re free to publish: the shape of the work and the rules we run it by.

Book an intro — we’ll go deeper

Yours next

If one of these resembles your problem, let’s talk.

A thirty-minute call. We read your brief before we meet. If it’s not a good fit, we’ll point you toward someone better — and probably name them.

Start a project All services