Vaibhav Bansal

Vaibhav Bansal

VP Everest Group
Vaibhav Bansal

“Vaibhav is a Vice-President at Everest Group and has been associated with the firm for ~3.5 years. At Everest Group, he leads research and advisory across Automation and AI, including areas such as IDP, Agentic AI, Gen AI, Enterprise AI platforms, RPA, Process intelligence, and Process orchestration. Vaibhav comes with ~16 years of overall professional experience in areas such as Automation, AI, Digital transformation, Strategy, and Business Process Management. He has worked across consulting, technology, and research-oriented roles in his entire career. Prior to joining Everest Group, Vaibhav has worked for esteemed organizations such as IBM, Cognizant, EXL, and ZS Associates.”

Day One

9:00 AM AI Outcomes Start Upstream: Why Cleaning Document Pipelines Beats Tuning Prompts

Most AI misses the mark not because models are weak, but because inputs are. Unstructured, inconsistent, and low-quality documents force RAG and LLMs to guess. This session maps the upstream work that reliably lifts precision/recall, reduces hallucinations, and shortens time-to-value, beyond what IDP alone can do.

We’ll unpack:

  • Why AI stalls: messy scans, mixed layouts, legacy files, weak metadata/lineage.
  • IDP’s ceiling: field-level OCR/extraction  normalization, validation, version control, or compliance at scale.
  • Fix the input layer: automate document cleanup, standardize formats, preserve structure, enrich/validate, route exceptions, and log audit trails.