Lernf is an AI risk literacy platform for engineering teams that trains developers to critically evaluate AI model behavior rather than passively accept its outputs.
What is Lernf?
Lernf is a web-based platform that trains engineering teams to recognize and counter AI sycophancy—the tendency of models to agree with users up to 58% of the time. It takes real-world engineering tasks as input (e.g., code reviews, design proposals) and produces structured exercises, dependency diagnostics, and adversarial critiques. The platform is built by the Lernf team, whose name stands for "learning on the fly" rather than being a sitting duck.
Key Features
- Workplace Scenarios — Practice handling real engineering situations before AI handles them for you, building the muscle memory that erodes when you outsource every decision.
- Intent Gradient Analysis — Visualize agreement patterns, information filtering, and dependency signals across models like Claude, GPT, and Gemini, revealing what AI won't tell you about itself.
- Devil's Advocates — Structured AI-generated dissent that attacks your proposals from distinct analytical lenses, preventing calcified thinking.
- PR Review — Catch what AI-generated code misses before production, simulating the review skills that matter when AI writes 80% of the code.
- GRIP Compass — Map your team's AI relationship against historical patterns of institutional dependency, showing where you're heading.
- Multi-Model Comparison — See how different models handle the same input differently, using your own API keys to avoid vendor lock-in.
- Simulated Engineering Conversations — Practice real dialogues with AI-powered NPCs that role-play teammates, stakeholders, or critics.
Who is it for?
- Junior developers who need to build critical thinking skills alongside AI—without the mentorship that's declining as entry-level job postings drop 67% since 2020.
- Senior engineers tasked with reviewing AI-generated code and catching blind spots before they reach production.
- Team leads managing AI adoption who want measurable guardrails against the 17% decline in unassisted performance documented in the Bastani PNAS study.
- Compliance officers preparing for EU AI Act Article 4 compliance (August 2026), who need structured governance diagnostics.
What can you do with Lernf?
- Stress-test engineering proposals — Run a design decision through multiple Devil's Advocates and compare how each model attacks it from different analytical angles.
- Simulate a code review session — Feed AI-generated PRs into the PR Review tool and practice catching logical errors that the original model missed.
- Audit your team's AI dependency — Use the GRIP Compass and Intent Gradient Analysis to produce a diagnostic report showing where your team is on the spectrum from active adaptation to passive dependence.
How does it work?
Sign up for a free account, bring your own API keys for OpenAI, Anthropic, or Google models, and access the full suite of exercises. Each tool presents a real engineering scenario (e.g., a code diff, a design proposal) and guides you through analyzing model behavior—then compares your evaluation against the system's metrics.
Pricing
Freemium. Free tier with bring-your-own API keys. Paid plans from $20/month for individuals and $15/seat/month for teams. See per-person and team pricing. By signing up, you agree to the Terms of Use and Privacy Policy.
FAQ
Is Lernf free?
Yes—the free tier lets you bring your own API keys and use all tools. Paid plans unlock additional collaboration features and usage limits.
