Clerk
FAQ
Answers about the S-Rank deep research engine.
What is Clerk?+
Clerk is Ngentech’s flagship deep research agent. It is a distributed, durable execution engine that decomposes a topic into 100+ parallel sub-niches, researches each independently, then statistically analyzes and synthesizes an S-Rank research report.
How is Clerk different from a chatbot?+
Chatbots guess. Clerk computes. It runs 100+ parallel research workflows, scores sources 0.0–1.0 for reliability, tags sentiment, deduplicates claims, and calculates predictive “odds” from the mass of collected data — moving from AI guessing to data science.
How long does research take?+
A full research cycle completes within about a week, powered by Temporal.io durable execution that survives server restarts without losing progress.
What does Clerk produce?+
An S-Rank enterprise research paper (PDF/Markdown) with an executive summary, citations and references, trend visualizations, and predictive odds analysis — plus the full raw data for audit.
Is the research auditable?+
Yes. Every report and its data is pushed to a Git repository with Git LFS, so the full commit history shows how thoughts and odds evolved over the research period.
How does Clerk avoid rate limits and bans?+
Each worker runs as an isolated Temporal workflow with automatic retry and sleep on rate limits, a residential proxy mesh (Bright Data / Oxylabs), and IP auto-swapping — so one failure never affects the others.