The Future of Software

The world of software is undergoing a shift not seen since the advent of compilers in the 1970s. Compilers were the original vibe coding: they automatically generate complex machine code that human programmers had to manually write before. Over time, compilers became fully trusted, nobody has to look under the hood, most programmers won't understand a thing. Are AI coding agents the new compilers? Will we simply trust whatever code they generate?

Compilers, revisited

[Draft] Set up the analogy in full: what compilers actually automated, why trust in them accrued over decades, and what "looking under the hood" used to mean. Establish the parallel to today's coding agents.

If code is now reasoned into existence rather than written, the unit of trust moves from the source to the proof.

What formal verification changes

[Draft] Explain Reasonable's core bet: grounding coding models in formal verification so their output can be checked, not just believed. Define the frontier of learnable programming for a technical-but-broad reader.

The setup

[Draft] Describe the experimental setup with NVIDIA Nemotron. Models, tasks, verification harness, hardware. Keep it reproducible.

Figure 1 — chart placeholder

Figure 1. Replace with the results figure (e.g. verified-pass-rate vs. model scale).

Results

[Draft] Report the headline numbers here. Until then, the table below is structure only.

Method	Verified pass @1	Notes
Baseline	—	[Draft]
+ Nemotron	—	[Draft]
+ Verification	—	[Draft]

[Draft] A short code sample renders like this:

theorem sorts_correct (xs : List Int) :
  Sorted (verifiedSort xs) ∧ Permutation xs (verifiedSort xs) := by
  -- proof obligation discharged by the verifier
  sorry

What's next

[Draft] Close on the trajectory: what trusted, verifiable code generation unlocks, and how readers can follow or join the work.

Reasonable · Inspect. Verify. Trust. — We're hiring →