Fieldguide launches long-horizon agent for audit testing

Audit and advisory solutions provider Fieldguide announced the launch of Field Orchestrator, touted as a long-horizon agent capable of completing substantive testing workflows. 

While the specific definition can vary, in general a long-horizon agent can be described as an AI agent that is used not for singular tasks but managing complex goals over days, weeks or months, executing thousands of individual steps while maintaining a persistent state and memory the whole time to keep it focused on its function. 

Purpose-built for audit and advisory to plan and execute engagements from start to finish, Field Orchestrator is said to sustain reasoning and action across the full arc of an engagement, coordinating multi-agent workflows without continuous practitioner oversight. It treats testing not as a discrete task to be done in response to specific prompts but as a continuous workflow across population analysis, sample selection, evidence collection, extraction, testing and workpaper generation. Auditors interact with the Field Orchestrator through a conversational interface inside the audit program. It reads the engagement context, generates a multistep plan they can review and approve, and delegates execution across Fieldguide's AI agents, making sure to pause for human reviews.

Supporting the new agent are the newly released Field Board, described as a Kanban-style project management view, and the Agent Review Experience, a workspace for reviewing agent-tested outputs. Field Board organizes every control and deliverable into a visual Kanban lifecycle that gives managers and partners engagement-level visibility and shows preparers and staff a clear "what's next" view. The Agent Review Experience consolidates every agent-tested control into a single workspace for reviewers to validate evidence and signoff. 

"Testing isn't a series of independent tasks. It's a continuous reasoning chain, and managing it is what most AI in audit gets wrong," said Fieldguide CEO and co-founder Jin Chang in a statement. "Field Orchestrator coordinates the work through one conversation, moving from AI that runs at discrete task level to AI that runs an entire engagement, and Field Board makes every action from both humans and agents visible."

Fieldguide is premiering Field Orchestrator, Field Board and Agent Review at AICPA Engage 2026 in Las Vegas this week. While the long-horizon agent is currently being used for audit testing, Fieldguide described that as only its first use case.

The launch comes only a few days after Fieldguide announced a partnership with Mercia, a U.K.-based audit methodology and compliance provider, to embed Mercia's audit methodology throughout the Fieldguide platform. This adds capacities for dynamic risk assessments and audit programs that adapt to engagement context and autonomously complete full workflows, automated diagnostics and validation checks that proactively identify inconsistencies or missing evidence, and continuous methodology updates as Mercia publishes new guidance and standards.