Confidential mandate

Conversational Commerce AI Recovery Leader

Urgent / Unplanned

Conversational Commerce AI Recovery Leader mandate in Mexico City, Mexico · Omnichannel Consumer Retail

After a holiday assistant issued unauthorised discounts and contradictory order actions, a retailer needs a ten-month executive to restore Spanish conversational commerce and transfer controlled tool-use governance.

The mandate

The conversational-commerce director was removed after a peak-season assistant combined obsolete promotion content with over-broad ordering tools, issuing discounts beyond delegation and creating contradictory cancellation states. The service is now limited to informational replies, while product, operations and engineering dispute whether grounding, tool permissions, dialogue policy or commerce APIs created the failure.

The replacement must assume the Mexico City seat within fourteen days and lead a ten-month recovery. A permanent search begins after two controlled sales cohorts complete without high-severity conduct exceptions, expected in month five, with five weeks reserved for the successor to shadow one release and chair the final peak-readiness review.

Handover is achieved when Spanish intent and grounding evidence covers the priority journeys, every transactional action has scoped permission and confirmation rules, three releases meet signed conduct and service corridors, two simulated commerce failures are contained, and the permanent appointee accepts the residual tool, content and fulfilment risk register. Sales uplift alone cannot close the assignment.

The interim may disable intents or tools, stop releases, change evaluation and knowledge priorities, revise temporary staffing and reallocate up to MX$180 million inside the approved recovery budget. Customer restitution, pricing policy, permanent leadership hires, core commerce replacement and commitments above MX$60 million require committee or board approval; the role may not trade confirmation safeguards for conversion.

General contact-centre redesign, merchandising strategy, warehouse operations and loyalty-platform replacement are outside scope. The brief covers Spanish and code-switched conversational journeys, their knowledge and tool interfaces, release measurement, incident command and the operating system transferred to lasting leadership.

Why this seat is open

The peak failure was not an isolated hallucination; it exposed unclear authority spanning content, agent policy and irreversible commerce actions. Removing the former director left urgent remediation split among teams whose measures reward different outcomes. A temporary executive must restore bounded customer value and accountability before the company chooses its long-term conversational-commerce leader.

What you will own

  • Reconstruct affected conversations from retrieved content, prompt and policy versions, tool calls, API state, confirmations and eventual customer or fulfilment outcome.
  • Define allowed informational, reversible and irreversible actions with authentication, spending, discount, confirmation, escalation and human-review thresholds.
  • Establish Spanish evaluation coverage across regional phrasing, code switching, ambiguity, vulnerable customers, adversarial requests and incomplete order context.
  • Rebuild knowledge grounding around source authority, offer validity, inventory freshness, conflict resolution, citation and automatic expiry of promotional content.
  • Install release gates combining task success, unsupported claims, conduct exceptions, tool precision, latency, containment and controlled commercial evidence.
  • Command simulations covering stale promotion data, partial order execution, identity uncertainty and downstream timeout before restoring broader agent authority.
  • Transfer the action registry, evaluation suite, incident decisions, content controls, launch calendar and talent assessment through a successor-led release.

Candidate qualifications

  • Held executive or enterprise-director accountability for a production conversational commerce, customer-service agent or tool-using consumer AI platform.
  • Recovered an incident in which generated dialogue triggered an incorrect transaction, entitlement, refund or customer commitment.
  • Built Spanish-language and code-switch evaluation with regional reviewers and customer-conduct cases rather than translated English benchmarks.
  • Designed action permissions, confirmations, idempotency and compensating controls across an AI agent and transactional commerce APIs.
  • Balanced conversion and automation benefits against fraud, vulnerable-customer, fulfilment and complaint evidence under peak traffic.
  • Handed a recovered AI product to permanent leadership with tested release, incident and content-governance mechanisms.

Non-negotiables

  • Can begin in Mexico City within fourteen days, work on site for the first twelve weeks and travel monthly to another operating hub.
  • Will join high-severity customer incident command and accept committee boundaries over pricing, restitution and conduct risk.
  • Brings direct Spanish production-AI ownership; English-only chatbot strategy or contact-centre outsourcing leadership is insufficient.
  • Must disclose current retail, model-provider, commerce-platform and conversational-AI vendor relationships before investigation access.
  1. 49 words maximum. State your earliest Mexico City start date and any obligation preventing exclusive leadership through the first six months.
  2. 49 words maximum. Describe a generated conversation that caused a wrong transaction and the control you imposed before restoring service.
  3. 49 words maximum. Which conduct signal would make you disable a profitable commerce action despite improving conversion?

This mandate is confidential. The client is named only under a mutual NDA, and your own record is never listed, sold or shown to a company under your name until you release it for this specific mandate.