TypeSafe founder: tool-calling LLMs pay a huge model to pretend to be human
thursdai_pod · x · 2026-09-18
On the ThursdAI podcast, Allie Laabs of TypeSafe offered a sharp analogy for tool-calling LLMs: it's machine code dressed up as a person, with a huge model paid to pretend, then translated back — "the dishwasher already exists. Do not build the robot."
The episode also covers Jev, TypeSafe's non-LLM model: $42 per billion input tokens with free output. It doesn't write answers — you define the question, it returns a probability, with choice, score, and null modes.
More from Models
- Jev claims first model optimized for low/medium-entropy tasks; frontier models only for high-entropy — multiply_matrix · 2026-09-18
- Image generation used Transformers for years; autoregression plus language-only pretraining is the novelty — benno_krojer · 2026-09-18
- How TypeSafe Jev turns LLM token probabilities into snap judgments — and why frontier labs could clone it — JnBrymn · 2026-09-18
- Cryptographer mocks OpenAI: sandboxes can't even run a recent Linux kernel — matthew_d_green · 2026-09-18
- GPT-6 weekly limits spawn 'usage maxing': users race the reset timer — rschu · 2026-09-18
- Researcher: real-time driving-game models were doable in 2016 — industry has LLM tunnel vision — gandamu_ml · 2026-09-18