> initializing C1_LABS

Adopt AI.
Fearlessly.

C1 Labs is the C1.ai research studio for identity security in the age of autonomous agents. Deep research and hands-on labs - open to everyone.

read the research labs - coming soon

What we research

Non-human identity

Service accounts, agents, and workloads now outnumber people. We map how they get access - and lose it.

Agent access perimeter

Once an LLM can act, the trust boundary moves. We pressure-test where prompt injection meets real grants.

Standing access & JIT

Long-lived grants are how breaches escalate. We study just-in-time models that cut standing access to zero.

Detection & anomaly

Catching the anomalous grant before it is abused. Signals, baselines, and the noise floor of real organizations.

Latest signal

view all →

# latest signal$ research show how-is-null-broke-our-query-plan-estimates-in-postgreskind: Database · Aug 19, 2026status: published

DatabaseAug 19, 2026

How IS NULL broke our query plan estimates in Postgres

Our query generator emitted the same deleted_at IS NULL clause more than once, and Postgres multiplied the selectivity of every copy. The row estimate collapsed, the planner picked the wrong plan, and a 293ms query took 50 seconds. Here's the bug, and the patch we sent to pgsql-hackers.

read the write-up →

Aug 11, 2026

Evolution of Our AI Agentic Engine

SELF-PACED · RUNS IN YOUR BROWSER

Learn by doing.
Coming soon.

Hands-on tutorials, guided walkthroughs, and CTF-style challenges - write real least-privilege policies, hunt over-permissioned accounts, and escalate through a misconfigured connector before the timer hits zero.Coming soon

preview →

// OUR PREMISE

Security shouldn't be the reason you slow down on AI.

C1 Labs exists to close the gap between how fast teams adopt autonomous agents and how fast identity security keeps up. We publish what we learn and hand you the labs to try it yourself. No gatekeeping.