Docusign Slashed AI Costs With Specialized Models

The company reduced contract processing costs by 50-fold after switching from large general-purpose AI systems.

Updated on Sept. 21, 2026 in Artificial Intelligence

Isometric editorial illustration showing a cluster of hexagonal metallic processors with detached modular data segments on a neutral background. — Docusign has slashed AI operating costs by 90 percent after transitioning from general-purpose models to smaller, specialized AI systems for contract processing. AI Illustration. Upload story photo >

Live Poll

Do you believe companies are successfully turning their current investments in AI into profitable results?

Docusign has successfully transitioned to smaller, specialized AI models for contract processing, cutting its AI operating costs by 90 percent. By filtering specific contract segments for analysis, the company increased its throughput eightfold.

Why it matters

The shift addresses the unsustainable costs associated with using massive general-purpose models for high-volume production. By lowering expenses, Docusign aims to make its AI agent product line commercially viable at scale.

Docusign processed more than 1 million agreements per day while maintaining accuracy within 2 percentage points of larger models. The company retired a previously necessary cleanup step for output correction to achieve a 50-fold reduction in per-document costs.

The players

Docusign

This company provides e-signature technology and digital transaction management services to millions of global users.

Microsoft

This global technology firm provides cloud computing infrastructure and AI training platforms like Microsoft Foundry.

The details

Docusign utilized Microsoft Foundry to train specialized models that extract over 50 facts from every agreement. A new filtering mechanism identifies pertinent contract segments, ensuring only essential data is processed by the AI engine.

Per-token AI prices have trended downward since 2022.
Docusign unveiled its Iris assistant and Agent Studio in May 2026.
PYMNTS released a report on enterprise AI billing in June 2026.
The Enterprise AI Benchmark report was published in August 2026.
Microsoft released a customer case study on Docusign on September 15, 2026.

The Tech Race

Docusign's shift reflects a broader enterprise movement to counteract the 320% rise in AI bills since 2022 by moving away from expensive, general-purpose models. This pivot highlights the growing pressure on firms to transition from experimental AI deployments to cost-efficient, specialized infrastructure.

Business users can expect more affordable AI-powered document tools as Docusign lowers its overhead through specialized model efficiency. These technical improvements facilitate faster turnaround times for contract reviews and fact extraction in high-volume digital workflows.

The takeaway

Companies are increasingly finding that smaller, task-specific AI models often outperform larger systems in terms of both cost and speed. Leaders looking to optimize AI spend should evaluate whether their current workloads truly require large general-purpose engines.