Skip to Main Content
Shape the future of IBM watsonx Orchestrate

Start by searching and reviewing ideas others have posted, and add a comment (private if needed), vote, or subscribe to updates on them if they matter to you.

If you can't find what you are looking for, create a new idea:

  1. stick to one feature enhancement per idea

  2. add as much detail as possible, including use-case, examples & screenshots (put anything confidential in Hidden details field or a private comment)

  3. Explain business impact and timeline of project being affected

[For IBMers] Add customer/project name, details & timeline in Hidden details field or a private comment (only visible to you and the IBM product team).

This all helps to scope and prioritize your idea among many other good ones. Thank you for your feedback!

Specific links you will want to bookmark for future use
Learn more about IBM watsonx Orchestrate - Use this site to find out additional information and details about the product.
Welcome to the IBM Ideas Portal (https://www.ibm.com/ideas) - Use this site to find out additional information and details about the IBM Ideas process and statuses.
IBM Unified Ideas Portal (https://ideas.ibm.com) - Use this site to view all of your ideas, create new ideas for any IBM product, or search for ideas across all of IBM.
ideasibm@us.ibm.com - Use this email to suggest enhancements to the Ideas process or request help from IBM for submitting your Ideas.

Status Submitted
Created by Guest
Created on Sep 19, 2026

Tight adaptive abstention sets for human tie-breaking

Problem: When agents abstain or hedge, they produce over-wide candidate lists (useless) or overconfident single answers (dangerous). Honest uncertainty should be tight and adaptive, not vague.

Idea: Add an Orchestrate abstention module fusing evidential deep learning with conformal prediction, producing small, adaptive candidate sets that keep coverage guarantees — honest uncertainty with less hedging. Grounded in Karimi et al. (COPA 2024), whose evidential uncertainty sets were smaller and more adaptive than three state-of-the-art conformal methods.

Value: Human tie-breakers get short, trustworthy shortlists instead of noise or false confidence — faster resolution, fewer errors.

Idea priority Low