01
Discovery before generation
Generation is cheap. Context is scarce. The system should first understand what already exists before producing more material.
How we think
Software should remember why the work matters.
We do not build AI for its own sake. We build systems that help people think more clearly—starting from knowledge that is usually already present, but scattered.
Principles
Seven commitments that shape every product.
Every 158 West product starts from the same observation: useful knowledge is usually already present across documents, messages, memories, analytics, conversations, and unfinished work. Our job is to help people find it, understand it, and act on it without losing the source, context, or human judgment that made it valuable.
02
Sources survive the summary
A conclusion becomes more useful when the user can inspect where it came from, what supports it, and where uncertainty remains.
03
Memory requires boundaries
Persistent context should be approved, scoped, reversible, and separated by purpose. Business memory, creative memory, evidence memory, and conversation memory should not collapse into one unrestricted pool.
04
Human judgment remains central
AI may surface options, contradictions, and patterns. People still decide what is true, appropriate, persuasive, private, or worth pursuing.
05
Products should cooperate without becoming one giant app
Different work deserves different interfaces. A creative director, evidence analyst, business intelligence system, and conversational personality should not be forced into the same workspace.
06
The blank prompt is a design failure
Users should not have to repeatedly explain their brand, audience, source material, constraints, and prior decisions. Good software carries approved context forward.
07
Real work is the test
The products are dogfooded against real campaigns, documents, business histories, and conversations before being treated as complete.
A shared operating loop
The 158 West operating model
The products serve different domains, but they follow the same underlying pattern: from scattered information to useful work.
Collect
Bring together the approved sources, materials, and context relevant to the work.
Connect
Identify relationships across information that would otherwise remain isolated.
Discover
Surface facts, patterns, tensions, opportunities, and missing pieces.
Decide
Keep the person in control of meaning, approval, and direction.
Create
Turn the approved understanding into useful outputs and retain what was learned.
Remember
Carry approved knowledge and preferences forward so the next cycle does not begin from amnesia.
See the principles in the product family.
ForgIOS, Myoshi, Andosho, and MoodyBot each apply this model to a different kind of intelligence work.