Meet Mercury

An agentic operating system is a software layer that connects business knowledge, decision-making, and execution across existing systems. It gives AI agents a shared understanding of the organization’s objectives, relationships, and constraints, allowing them to monitor conditions and carry out coordinated work within defined permissions. That context persists over time, so agents can account for previous decisions and future commitments as circumstances change.

For retailers, its value lies in managing the dependencies between commercial decisions. A change in demand may require adjustments to purchasing, allocation, and promotional plans, each owned by a different team. Mercury brings these activities into a shared operating environment, helping teams assess the implications, delegate execution, and retain oversight of decisions that require their judgment.

Mercury uses analogue-based forecasting to establish an initial demand estimate from comparable products. Product attributes such as category, material, intended audience, and SKU type help identify relevant comparisons, while available market signals provide additional context. These inputs inform a demand range and production recommendation before the new product generates its own sales data. As results become available, teams can reassess the initial assumptions against observed performance.

A temporal knowledge graph represents business information, its relationships, and the periods during which that information applies. In retail, this matters because prices, assortments, inventory positions, and supplier commitments change over time. Mercury uses this structure to interpret current conditions alongside the circumstances behind earlier decisions, allowing its agents to account for what has changed and which assumptions remain relevant.

Spreadsheets and business intelligence tools support analysis, reporting, and planning. Mercury adds a shared operating environment in which agents maintain business context and carry work through from analysis to coordinated execution. When conditions change, agents can reassess the implications for related decisions and act within defined permissions. Teams retain oversight through role-specific views and approval requirements, with the context behind decisions available for review.

Mercury supports merchandising, planning, buying, marketing, and finance teams through views organized around their responsibilities. Each function can examine the information, decisions, and trade-offs relevant to its work while retaining visibility into dependencies across the business. Teams bring their domain expertise and commercial judgment; Mercury provides a shared context for applying that expertise and coordinating the resulting actions.

As sales results become available, they provide evidence for reassessing the assumptions behind an initial forecast. Mercury uses that evidence to refine demand estimates and inform subsequent planning decisions. The extent of improvement depends on the quality and relevance of the data, the comparability of products, and changes in market conditions. Forecast performance should therefore be evaluated against actual outcomes across relevant products and planning horizons.

Building internally is an option for organizations with the engineering capacity and domain expertise to develop and maintain the system. The work extends beyond connecting a language model to company data. It includes defining business concepts and relationships, preserving context over time, coordinating agents, managing permissions, and evaluating performance as systems and processes change. Mercury brings these capabilities together around retail operations, allowing teams to focus their implementation effort on their own commercial priorities and operating requirements.

Mercury’s agents operate within defined permissions and approval requirements. Their scope can include monitoring conditions, preparing recommendations, and carrying out authorized tasks, with decisions requiring human judgment routed for review. This allows teams to establish where autonomous execution is appropriate and where commercial accountability requires direct involvement.

Mercury serves as a decision and coordination layer across connected business systems. Its role is to bring their information into a shared commercial context and support actions that depend on several functions or sources of data. The systems connected, information available, and actions agents can perform depend on the integrations and permissions established for your deployment.