---
title: "Manifesto: Composable AI — Build Prod, Not God"
type: entity
tags: [manifesto, mission, composable-ai, positioning]
created: 2026-09-17
updated: 2026-09-17
confidence: high
sources:
  - raw/site/typesafe-ai-manifesto.txt
jev_version: "jev-1.13.0"
summary: "TypeSafe's manifesto argues the bottleneck is composability, not intelligence, and lays out a three-step plan to make AI a dependable software primitive."
---

# Manifesto: Composable AI — Build Prod, Not God

> **TL;DR** TypeSafe's manifesto (https://typesafe.ai/manifesto) argues that raw intelligence is no longer the bottleneck — composability is. Today's AI is a "horseless carriage": chat-shaped because a human was assumed on the other end, which forces humans into the loop. TypeSafe's stated mission is to make intelligence invokable like a database query so software can layer on it. Slogan: "We're building prod, not God."

## Facts

| Field | Value |
|---|---|
| Title | "Composable AI: Build Prod, Not God" |
| URL | https://typesafe.ai/manifesto |
| Author | Not attributed on the page (company voice, "we") |
| Date | Not stated on the page |
| Captured | 2026-09-17 |

## The mission

Verbatim: "Our mission is to pave the shortest path to an AI-based economic revolution by making intelligence composable to catalyze a Cambrian explosion of intelligent software."

Two of the page's footnotes make that concrete: the "economic revolution" footnote says "Specifically, the lack of automation" is the problem, and defines the favored success metric as "global TFP growth reaching 3% within five years and holding at that level for ten (unprecedented in economic history, but achievable if AI's gains diffuse across the real economy)."

## The argument, section by section

**"We already have general intelligence."** TypeSafe says it is "not racing towards the ever-moving goalposts of 'AGI,' because today's models have long since crossed the threshold of intelligence needed for creating massive economic value." The evidence it offers is negative: "after years and trillions of dollars invested in the industry, most software still isn't meaningfully intelligent, and life is mostly the same for the average person. That should tell us something: the bottleneck isn't raw intelligence. It's that today's intelligence is hard to build on."

**"Beyond horseless carriages."** The central analogy (illustrated with a 1903 Ford Model T): "Early cars were designed as horseless carriages: rather than reimagining transportation from scratch, inventors took the familiar carriage and replaced the horse with a motor, while keeping the high seats, the buggy springs, and even the whip socket in some models. It's a clear example of how new technologies are often forced into the shape and assumptions of the thing they replace before finding their own native form."

Applied to AI: "Current AI is trained to be a helpful, articulate, pleasant assistant. Reasonable goals if you assume a human is on the other side of the model. Yet the foreseeable consequence is AI that requires humans in the loop instead of running in the background." The footnote on that sentence is the training-objective critique: "RLHF, the algorithm almost all of current AI is trained with, directly optimizes for human preference. Modern 'reasoning' models are trained with RLVR as well, which optimizes for programmatically evaluatable tasks (i.e., benchmarks)."

The alternative it proposes: "We want AI to work alongside existing software as a primitive that any programmer can invoke for semantic judgement and decisions, while still using code for what it's best at: exact computation." The line the page builds to: "Computers can do so much by just branching on bits, imagine if they could also branch on common sense, understanding, and intent." A footnote calls this "the dream of neuro-symbolic AI: neural networks for perception paired with symbolic logic for reasoning. Sometimes cheekily summarized as 'smart if-statements.'"

**"Safe emergence."** The historical claim: "The people who built databases didn't imagine Google. The people who built internet protocols didn't envision Stripe. They made lower-level capabilities so dependable that they could not only run in the background, but also be layered on top of. A Cambrian explosion of software emerged that nobody could have designed top-down."

The analogy for today: "Intelligence today is like databases before SQL: powerful, but every use is bespoke. Once a smart decision becomes as dependable and invokable as a database query, builders will stack them the same way."

And the safety argument, which is framed as an engineering precondition rather than a policy position: "Safety is a precondition for layering. You let a component run unattended if it's reliable; you only build on top of it if it's trustworthy. It takes trust to bury a dependency five layers deep in a system. People will only do so if they can inspect it, test it, and constrain it piece by piece, the way we've always engineered dependable software."

**"A new shape of intelligence."** Three stated steps:

| Step | Text |
|---|---|
| Step one | "Ship the shape of machine-native composable AI with the highest possible intelligence-per-dollar." |
| Step two | "Make our AI reliable enough to transform the economy via real automation." |
| Step three | "Empower the world with higher-level intelligence abstractions that are stable enough to compose and layer upon, for a collaborative, emergent future." |

Closing line: "We're building prod, not God." (Also the headline of the team page, as "Build Prod, Not God".)

Two footnotes attach to step one and to the appendix: "With a new type of frontier model meant for integrating AI within systems," and "On top of rock solid infrastructure built by people who care A LOT. 🥹"

## How this maps to the product

The manifesto never names Jev. The mapping to what shipped is the launch post's job: typed decisions instead of strings, calibrated probabilities so code can decide when to act, and composition in code rather than in prompts. See [[entities/blog-introducing-system-one]] and [[concepts/how-to-build]]. The "smart if-statements" framing reappears in the launch post's use-case table.

## Related

- [[entities/typesafe-ai]] — the company
- [[entities/blog-introducing-system-one]] — the manifesto's claims turned into a product announcement
- [[entities/blog-bitterest-lesson]] — the research-direction argument behind it
- [[concepts/system-one]] — what the "new shape of intelligence" is in practice
- [[entities/team]] — "Build Prod, Not God" as the team-page headline

## Sources

- raw/site/typesafe-ai-manifesto.txt (https://typesafe.ai/manifesto)