Enter your details to view and download the full PDF version
Let's discuss your options and get a demo underway
The future of debugging is here
A new debugging experience driven by Artificial Intelligence. Resolve issues with unprecedented speed and accuracy, using AI-powered suggestions on the root cause of errors and how to fix them.
With one click, Raygun sends a detailed prompt to your chosen LLM preloaded with crucial context from your stack trace, environment, and affected code. AI Error Resolution analyzes and diagnoses the error group to suggest fast and accurate solutions.
It’s hard to keep your cool when you’re alerted to an urgent error in production. AI Error Resolution is on-call 24/7 to support you, no matter what. Get instant solutions to resolve issues before they become incidents.
AI tools are already part of the debugging process for most developers. With a secure integration that you control, you can remove the hurdles of context-switching and separation from your codebase. With built-in prompts refined by our team of engineers, you’re asking better questions to get better answers.
Isolation is inefficient. AI Error Resolution keeps everyone informed and involved with multiple live user messaging, conversation history, and comment capture, and queries your stack trace directly to give you more precise and informed diagnostics. Give your people and systems the visibility to get the job done.
AI Error Resolution is an enhancement of Raygun Crash Reporting. To see it in action, sign up for a free 14-day trial of Crash Reporting — no credit card required. Existing Crash Reporting customers can start generating solutions right now!
Connect securely to your organization’s OpenAI account using your API key.
Crash Reporting will surface the errors in your codebase, which you can sort by your chosen criteria.
Select your first error instance and open AI Error Resolution. You’ll immediately get three suggested solutions and can even generate the code to execute.
What languages and platforms does AI Error Resolution support?
Raygun Crash Reporting supports all major languages and platforms. You can see a full list of our supported technologies here.
How will I be charged for use of tokens/responses generated by the AI provider?
Raygun makes one API call to OpenAI for each new request to resolve an error and each subsequent follow-up question. The costs depend on OpenAI’s pricing and the LLM model you select to use.
What data is AI Error Resolution trained on?
Raygun doesn’t train any models. AI Error Resolution is powered by the AI provider of your choice, and draws on the LLM’s general knowledge while using the prompts and parameters built into the feature. Read more about Raygun’s data handling policies here.
What data is sent to the provider and to Raygun?
To retrieve a resolution, Raygun crafts a prompt that includes the specific data of that error group (e.g., the stack trace). Raygun securely posts the prompt for inference only to the selected third-party AI integration using your plan’s own API key. No model training takes place with any customer data.
Who owns the data transmitted between Raygun users, Raygun, and the model?
Your interactions with the AI provider are securely stored with Raygun so that you don’t have to prompt the AI provider again to view previous discussions on an error group. You can always permanently delete any discussion thread.
How long will Raygun retain data?
We’ll hold your data for 90 days in line with our data retention policies.
What safeguards are in place to protect against the “hallucinations” or malfunctions that AI models have been known to exhibit?
Raygun does not offer any guarantees on the quality of the LLM’s answer, which is controlled by the AI provider (OpenAI). We have tested and experimentally validated our prompt, which, to the best of our knowledge, has not resulted in any hallucinations so far.
Do I have to be a Crash Reporting customer to use AI Error Resolution?
Yes, AI Error Resolution is built into Crash Reporting and queries the stack trace data surfaced via error reports.
