whereismyai
On the Record
On the Record is an open-source software checkpoint that makes AI agents prove their work and verifies search citations before delivery.
On the Record is an open-source software project built to replicate the Tow Center study on AI search citations. It acts as a checkpoint between AI models and users, enforcing rules that require AI agents to prove their work before claims ship. When tested against four AI search engines with 800 citations, the system blocked fabricated and dead links by refusing unverified claims and providing audit records with receipts. The software works for any claim a script can check, including numbers, code, and legal references, without requiring model retraining. The dataset, grader, and audit records are public and available in the repository.
On the Record at a glance
- Best for: Software engineer, Researcher
- Runs on: open source
- Pricing: Free
- Released: October 2026
- Category: AI research tool
- Website: marcus-ai.dev
Key features
- Verifies AI search citations
- Blocks fabricated and dead links
- Generates audit records with receipts
- Exports full audit bundles
- Operates as a checkpoint between AI and user
- Checks claims against sources
Use it for
- Verifying AI search engine citations
- Auditing AI agent claims
- Checking numbers against sources
- Validating legal and medical references
Alternatives to On the Record
- past.dev — Provides agents with accurate, timestamped context instead of static retrieval.
- HyperResearch — Generates verified reports with citations checked against full source texts and a persistent vault.
- OpenResearch — Run parallel research workflows locally or on remote compute.
- Scite — It uses direct publisher agreements to verify scientific claims with exact source links.
- Jonex — Provides traceable, agent-ready knowledge from diverse enterprise data sources.
- Keryx — Compensates verified creators directly in USDC for cited sources
- Benchpress — Evaluates document artificial intelligence models against specific user pages for accuracy, cost, and speed.
- AutoResearch — It provides a stateful, reproducible workflow for automated AI research.
- Paperchecker — Runs locally to prevent data theft and protect privacy.
- Lune — Accesses trusted knowledge from top conference papers for AI agents.
- PageIndex — It provides 98.7% accuracy on FinanceBench without vector databases.
- Leaked System Prompt AI — Explores how major AI systems are instructed and configured.
All On the Record alternatives