Read-only by design
Writes are refused at the database, not left to the model.
Memory
Your team already knows which ideas worked and why others failed. Refix carries that history into its next recommendation, so your experience keeps informing the work.
Used by teams at
What counts as an active customer at your company becomes context for future questions across the team.
New information is checked against existing memory, so a changed definition can replace an outdated one.
Findings and useful methods carry forward. The next investigation starts with more of the context your team has earned.
Keeping it current
If your definition of an active customer changes, future answers should reflect it. Refix checks new facts against existing ones and updates the relevant memory while keeping unrelated context.
Better judgment
A previous experiment may explain why an obvious change is a poor bet. Refix brings relevant history into the investigation and shows the context behind its recommendation.
Completed bookings per search, Jul 6 to Aug 24
See how Refix finds an opportunity and follows it through.
Choose what Refix can access and set the approvals it needs. Review the data and reasoning behind its recommendations.
Writes are refused at the database, not left to the model.
Require your team's approval before Refix takes action in connected tools.
Decide who connects data, approves actions, and manages billing.
Daily credit use, with an alert before you reach your limit.
Model providers keep nothing and never train on your data.
Self-hosting
Run the Refix agent on your own Kubernetes cluster. Your platform team controls access and model connections, with our engineers helping you deploy.
Refix keeps business context such as metric definitions, past decisions, and useful findings. It also remembers personal preferences, such as how you want results presented.
Yes. Ask Refix to remember a fact or forget something it has stored. For example, tell it that your team now reports revenue by week.
Refix checks the new information against existing memory and updates what it supersedes. Facts with an end date can expire, so temporary context doesn't become a permanent assumption.
Personal preferences apply to your own conversations. Shared business facts, such as a metric definition, provide context for the team's work.
Memory is stored in your workspace's database, isolated from other customers. With self-hosting, that database runs on your infrastructure.