Automorphic provides Conduit, a tool for integrating knowledge into language models. It addresses the limitations of prompt stuffing by enabling language model fine-tuning, behavior/knowledge adapter training, and dynamic combination of these elements.
This tool speeds up model iteration through human-in-the-loop feedback and streamlines model deployment. Conduit enhances language models by continuously updating them based on user input and manual labeling.
It ensures high performance and deployment efficiency by enabling rapid loading and stacking of fine-tuned adapters. Its compatibility with the OpenAI API allows seamless integration of Conduit into existing codebases. The Automorphic Hub offers publicly shared models trained and enhanced using Automorphic, available for inference.
Users can access these models on the hub for their own applications. Automorphic also offers TREX, which transforms unstructured data into a structured format like JSON, XML, YAML, or any format specified by a regular expression or context-free grammar.
TREX provides a highly customizable alternative to OpenAI's functions. Aegis, a firewall tool by Automorphic, protects language models and users from adversarial attacks.
It defends against prompt injections, prompt and personal identifiable information (PII) leakage, and harmful language. Aegis constantly learns from usage to improve its detection abilities.
Fine-tune language models.
Combine models in a dynamic way.
Rapid model iteration is possible.
Requires manual feedback.
Adapts slowly to initial feedback.
TREX has a complex setup.

Released 3 months ago
Free + from $25/month

Released 3 years ago
From $49/month

Released 3 years ago
Contact for pricing

Released 2 years ago
Free + from $50/month

Released 1 year ago
Contact for pricing

Released 2 years ago
Contact for pricing

Released 2 years ago
Contact for pricing

Released 1 year ago
Free + from $0.50/unit