Skip to content

Migrate Assistants

OpenAI retired the Assistants API on 2026-08-26. Open Chat Studio's Assistants pages were removed on 2026-09-02 — assistants can no longer be created, viewed, edited, or archived in OCS. The assistant pipeline node was removed next, so a pipeline that still contains one no longer builds. All stored assistant records have since been deleted permanently. See the OpenAI Assistants (Removed) page for background.

Affected chatbots are already down

A chatbot whose pipeline still holds an assistant node cannot answer participants. Follow the steps below to replace that node with an LLM node and bring it back.

Assistant records no longer exist in OCS — they have been permanently deleted. If a chatbot or pipeline still holds an assistant node, replace it with the native OCS features shown in the table below:

Assistant Feature Replacement Feature
Threads Open Chat Studio sessions
Code Interpreter OpenAI Code Interpreter tool in LLM nodes
File Search Indexed Collections

Migrating Code Interpreter

Info

This guide assumes that you have enabled the chatbots feature

To use OpenAI's code interpreter tool without using Assistants:

  • Create a Chatbot with an LLM node.
  • Select an OpenAI LLM Provider
  • Check the "Code Execution" checkbox under the "Builtin Tools" section of the configuration.

Make sure you're familiar with the different types of Collections before continuing.

General Steps

  1. Create an indexed collection using the same files your assistant used under its "file search" tool.
  2. Set up or update your chatbot to reference this collection.

Step 1: Create the Collection

Create the collection manually — there's no automated import from an assistant, since assistant records no longer exist in OCS.

  • Click on the "Collections" tab in the sidebar and click "Add new".
  • Choose Indexed Collection and give it a name.
  • Choose between a Remote Index and a Local Index. A remote index is closest to how an assistant's "file search" tool worked — the files are indexed by the LLM provider you select, so pick the same provider your assistant used.
  • Upload the same files your assistant used for its "file search" tool. If you no longer have local copies, download them from your file storage at OpenAI.

Step 2: Update your chatbot

Once your collection is created:

  • Open your Chatbot's pipeline editor.
  • If the pipeline still contains an assistant node, it renders as a Removed Node and the pipeline will not build. Delete it.
  • Add an LLM node to the pipeline in its place.
  • Within the node, select your newly created indexed collection.

You're done!