Natalia Connolly

About Natalia Connolly

Natalia Connolly is part of the AI and data team at AlixPartners, the management consulting firm. In an AI Engineer talk with Kevin Madura, she lays out the business context of a post-merger procurement review and the contract-data problems that led Claude Code to wrong answers.

Notable Positions

On Messy Contract Data

Real contract portfolios contain long documents that silently time out in preprocessing, orphan amendments not linked to their parent, amendments that restate the total contract value, and rate tables that break across pages without headers. None of it is exotic, and all of it misleads an agent that only sees extracted numbers.

On the Pipeline Before the Agent

She and Madura spent about two weeks on a five-stage pipeline: OCR and document intelligence, classification (MSA, amendment, NDA), extraction of terms, rates and spend, mapping services to spend categories, and only then analysis by Claude Code. Structured tables also made every answer checkable in code.

Key Quotes

• "The agent said December 1st. The actual date was November 1st, right?" (on the missed auto-renewal deadline)
• "40,000-plus contracts were processed overnight once it was done." (on the pipeline payoff)

Related Reading

• Kevin Madura - Co-presenter at AlixPartners
• Context Engineering - Structuring what agents see