Forever Healthy Releases AI4L 1.0 for Practical Longevity
- An Open-Source AI System for Evidence-Based Reviews of Longevity Interventions
We are super excited to share with you that today we released AI4L – “AI for Practical Longevity”, an open-source system that enables anyone to produce rigorous, evidence-based reviews of health and longevity interventions using frontier AI models.
Our novel “Audit-Driven Prompting” method enables iterative self-auditing and eliminates hallucinations with live citation verification and zero-tolerance quality gates.
The 1.0 release is available under the MIT license at github.com/forever-healthy/AI4L
The first generation of longevity therapies is available today, but the evidence is scattered
Senolytics, NAD+ restoration, mTOR modulation, geroprotectors, peptides. Unfortunately, the underlying evidence is incoherent and distributed across journals, clinical trials, expert commentary, and specialist communities.
A conventional approach does not scale
We previously produced intervention reviews with a dedicated research team, each requiring more than two months of work for a team of two. That approach did not scale to the full universe of interventions, let alone keep existing reviews up to date.
Conventional AI summaries are not really helpful either
They often sound equally confident whether they’re right or wrong. Due to the heuristic nature of AI, models often hallucinate studies and URLs, misrepresent evidence, or miss critical nuances.
AI4L takes a novel approach
Instead of instructing an AI to “write a review,” the project’s prompt describes a 390+ item quality assurance audit — the same kind of specification one would hand to a human auditor. The AI is then asked to generate a review capable of passing that audit and, afterward, to perform the audit itself.
Independent, history-isolated agents handle creation and auditing to avoid context bias and self-confirming hallucinations.
Auditors are required to actively fetch URLs, retrieve metadata, and verify citations against live sources.
Reviews cycle through creation, audit, and correction until they achieve a 100% pass across all QA criteria.
We call the approach Audit-Driven Prompting.
AI4L is designed to be model-agnostic. It runs as a single prompt in any major chat interface (including Claude Desktop) or in CLI environments for repeatable, automated workflows. The project ships with example reviews, audit transcripts, and documented limitations.
Resources
- Repository: github.com/forever-healthy/AI4L
- Example reviews & audits: github.com/forever-healthy/AI4L/tree/main/examples
- Discussion: github.com/forever-healthy/AI4L/discussions
About Forever Healthy
Forever Healthy is a private, humanitarian initiative with the mission of enabling people to vastly extend their healthy lifespan. More at forever-healthy.org
Media contact
About the author
Press release
