Most of what we publish starts as a research question: Does this certificate actually help you get hired? What can an AI agent safely do with access to your inbox? Which security bug should you patch first? We answer those questions with primary sources and hands-on testing, and we publish what we find, including the parts that don't flatter anyone.
Research areas
AI tools, agents, and workflows
Hands-on evaluations of AI assistants, coding agents, connectors, and local models. We focus on what holds up in real work, what it costs, and where it breaks. Where practical, we publish the prompts, configurations, and steps so you can reproduce our results. See our coding agent comparison and local agentic workflows guide.
AI in professional work
How AI changes specific jobs and industries: law, real estate, cloud security, design, and 3D work. We look at the workflows that save real time, the compliance lines professionals can't cross, and the new ways to earn. Examples include Claude artifacts for legal work and Claude and Zillow for real estate.
AI security and the software supply chain
Vulnerabilities in AI tools and their dependencies, safe agent design, prompt injection, and supply-chain compromises. This research is strictly defensive: who's affected, how to check, and what to fix. See our write-up on the LiteLLM supply-chain compromise.
Security news and vulnerabilities
Our Write-ups desk turns vendor advisories and CVEs into plain-English guidance for defenders. We never publish proof-of-concept code, payloads, or targeting details.
Courses, certificates, and career outcomes
Do credentials pay off? We review course content in detail, compare training platforms, and track what employers and recruiters actually ask for. Start with what recruiters actually care about and getting a cybersecurity job without a degree.
The economics of AI for individuals
What AI tools cost, what they replace, and how independent workers and small businesses are turning them into income, with realistic numbers and clear risk warnings. See making money with AI connectors.
Privacy and security tools
Side-by-side testing of the tools people trust with their data, like password managers and VPNs.
Research principles
- Primary sources first. Vendor advisories, official documentation, syllabi, filings, and our own testing come before secondary coverage.
- Facts, estimates, and opinions are labeled. When a number is an example or an estimate, we say so next to it.
- Methods and limits are published. We explain how we tested and what we couldn't check.
- AI-assisted, openly. Much of our research is done by autonomous AI agents under editorial review, and every byline says who did the work.
- Independent judgment. Sponsors, partners, and affiliate programs never decide our conclusions. Relevant commercial relationships are disclosed on the page.
- Corrections are public. When new evidence changes a finding, we update the article and its date.
Read the full standards on our How we work page.
Collaborate on research
We'd like to hear from:
- Researchers and developers with datasets, benchmarks, open-source tools, or reproducible findings to share or test together.
- Companies offering early or evaluation access to AI, security, or learning products. Access is welcome, but we publish what we find, good or bad.
- Security researchers whose findings have been disclosed and patched. We cover vulnerabilities after a fix or advisory is public, not before.
- Educators and employers with insight into which skills and credentials lead to jobs.
- Organizations interested in sponsoring a research project. Sponsorship is disclosed and never guarantees favorable findings. See Advertise for details.
How to reach us: use the subject line "Research inquiry" and include who you are, what you'd like to share or study, and any timeline. Open an issue on the site's repository with the subject line shown above, and we'll reply there or move the conversation somewhere private.