Delegating General Administrative Work to Lyric

Overview

During a 30-day pilot, John Collens, a system administrator at Bark, evaluated how Lyric performed general-purpose administrative work: itinerary preparation, supplier email processing, document and briefing preparation, coordination, routine data extraction, and scheduling. The pilot measured time saved, accuracy, verification of critical details, independent task completion, and compliance with data security and access requirements.

The challenge

Administrative work such as travel itineraries, supplier correspondence, briefing documents, and coordination competes for time with system administration. The pilot assessed whether Lyric could take on this work reliably, with every critical detail verified and access to data kept within policy.

Scope

Lyric performed general-purpose administrative work for one system administrator over 30 days, including:

• Itinerary preparation
• Supplier email processing
• Document and briefing preparation
• Coordination
• Routine data extraction
• Identification of scheduling conflicts

Accuracy and verification

Lyric verified every critical detail against its source, and routine data extraction reached at least 98.9% accuracy. On defined test cases, Lyric identified at least 95% of scheduling conflicts.

Learning and independent work

• At least 92.2% retention and application of established instructions on defined test cases.
• A measurable reduction in repeated errors after correction over the pilot.
• At least 70% of eligible tasks completed without substantive human intervention.
• At least 90% of eligible workflows completed.
• A decreasing trend in human review and correction time over the 30 days.

Business value and compliance

Net labor hours saved were measurable and documented, and the pilot showed positive net value after software and implementation costs. The pilot recorded 100% data security and access compliance.

Assessment

Over 30 days, Lyric took on a substantial share of general administrative work with high accuracy and verification of every critical detail. It retained established instructions, repeated fewer errors after correction, and required less review and correction time as the pilot progressed.

These results reflect one 30-day pilot with a single administrator. Figures described as test case results were measured on defined test cases, and completion figures apply to eligible tasks and workflows. Longer evaluations across additional roles would establish how consistently these results hold.