July 17th 2024
We are delighted to introduce our latest features and fixes from the recent period, enhancing your experience with improved functionality and performance.
Session Explorer
We are delighted to introduce our Session Explorer. Get instant, live logging of every prompt and response in the Session Explorer table. Track conversations and gain a level of transparency into your AI’s behavior. Learn which messages violated which policy and the exact action taken by Aporia’s Guardrails to prevent these violations.
PII Policy Expansion
We added new categories to protect your company’s and your customers’ information: SSN (Social Security Number), Personal Names, and Currency Amounts.
Policy Catalog
You can now access the Policy Catalog directly, allowing you to manage policies without entering a specific project and to add policies to multiple projects at once.
New Policy: SQL Hallucinations
We have announced a new SQL Hallucinations policy. This policy detects hallucinations in LLM-generated SQL queries.
New Fine-Tuned Models
Aporia’s Labs are happy to introduce our new fine-tuned models for prompt injection and toxicity policies. These new policies are based on fine-tuned models specifically designed for these use cases, significantly enhancing their performance to an entirely new level.
Flexible Policy Addition
You can now add as many policies as you want to your project and activate the number allowed in your chosen plan.
Log Action Update
We ensured the ‘log’ action runs last and doesn’t override other actions configured in the project’s policies.