Blog

Insights on the AI workforce

Engineering deep-dives, research notes, product updates and customer stories — from the team building governed, autonomous agents.

Hiring our first AI Safety engineer
Featured

Company

Hiring our first AI Safety engineer

Why a 30-person company is dedicating a full role to safety from day one — and what we want this person to own in the first 90 days.

Maya Chen Maya Chen · Apr 14, 2026 · 6 min read
Why we built a sandboxed runtime
Engineering

Why we built a sandboxed runtime for AI agents (instead of trusting prompts)

Prompts are not a security boundary. Here's the architecture we landed on after our first six months in production.

Sofia Alvarez · Apr 8, 2026 · 8 min
Anatomy of an Email Responder
Product

Anatomy of an Email Responder: 100M tokens later

What we learned shipping the most-deployed agent on the platform — and why "drafts only" beat "full automation".

Daniel Becker · Apr 2, 2026 · 7 min
Human-in-the-loop is a contract
Research

Human-in-the-loop is not a feature — it's a contract

A framework for thinking about approval gates as enforceable contracts between operators and the agents they deploy.

Priya Raman · Mar 27, 2026 · 10 min
Customer story: Northwind cuts ops headcount
Customer stories

How Northwind cut sales-ops workload by 40% with one CRM agent

A B2B SaaS team rolled out the CRM Specialist in two weeks. Here's the rollout plan, the metrics, and the surprises.

Jamal Okonkwo · Mar 19, 2026 · 6 min
Deterministic guardrails
Engineering

Engineering deterministic guardrails on probabilistic systems

A look at the policy engine that sits between the model and the world — written in Rust, powered by CEL.

Marcus Hale · Mar 11, 2026 · 9 min
Multi-tenant isolation
Engineering

Why we chose multi-tenant isolation over single-tenant for security

Single-tenant feels safer. We argue, with the threat model in hand, that the opposite is true for agent workloads.

Sofia Alvarez · Mar 4, 2026 · 11 min
The case against autonomous-by-default
Research

The case against autonomous-by-default

Most "autonomous" agent demos are autonomy theatre. We unpack what the data says about user trust and adoption.

Priya Raman · Feb 25, 2026 · 8 min
Q1 product roundup
Product

Q1 2026 product roundup

17 new integrations, the agent simulator, granular blast-radius controls — and what we're shipping next.

Daniel Becker · Feb 18, 2026 · 5 min
Evals are infrastructure
Engineering

Treating evals as production infrastructure

How we run 12,000 eval cases on every PR — and what we do with the cases that flake.

Sofia Alvarez · Feb 9, 2026 · 7 min

Newsletter

Subscribe for the long-form essays

One email a month. No fluff, no growth-hacking — just the deepest piece we wrote that month.