AI-Native Product Studio
We find worthwhile problems and build software that solves them.
Sometimes for clients. Sometimes with clients. Sometimes for ourselves.
What This Isn't
Shape & Signal is a repeatable product-creation studio built to turn worthwhile problems into working software — not to sell hours.
Ways of Working
Three ways in, one standard of work.
Build Ours
Shape & Signal originates and owns the product. We identify the problem, develop the thesis, own the roadmap, test the opportunity, and decide whether the experiment deserves further investment.
Learn more →Build For
A client hires Shape & Signal to solve a meaningful business problem or create a useful product. We start with what's actually happening, not a list of features to build.
Learn more →Build With
A deeper, more selective partnership with a founder or operator. Shared problem, shared thesis, shared build, potentially a shared stake in the outcome.
How We Work
Signal → Shape → Ship → Learn → Reshape
A decision loop, not a checklist. Every product moves through the same five stages — and can move backward when the evidence says so.
- 01SignalIs the problem real?
- 02ShapeWhat's the smallest useful solution?
- 03ShipInto real use.
- 04LearnWhat did reality say?
- 05ReshapeWhat should change?
Projects can move backward — Reshape often leads back into Shape, not forward into a new Signal.
Selected Work
Evidence, not a portfolio gallery.
Municipal AI Pilot Lab
PausedAn independent evaluation of whether AI can assist municipal correspondence triage while preserving employee oversight — testing a governance and evaluation methodology as much as a model.
- Problem
- Can AI assist municipal correspondence classification while preserving employee oversight?
- What we built
- A classifier evaluated over 200 synthetic resident messages, 3 independent trials each (600 classifications total), schema-validated, with a deterministic human-review router and a 10-area governance assessment.
- What we learned
- Accuracy and safety-gate performance alone aren't sufficient evidence for a deployment decision — cybersecurity was assessed as a gap before production, and records management wasn't assessed at all.
- Current state
- Recommendation: proceed to a bounded, controlled next-stage evaluation — not to production. No further work is currently active.
Framli
Active ExperimentInvestigating what happens after consultative B2B sales conversations, and whether sellers need better ways of helping internal champions carry a buying decision forward.
- Hypothesis
- Framli is investigating what happens after consultative B2B sales conversations and whether sellers need better ways of helping internal champions carry a buying decision forward.
LossPoint
LiveAn AI-powered business diagnostics platform that helps service businesses find where they're losing time, money, and operational efficiency.