AI Agents · Catalog · Live

AI Agents.

6 agents shipping in v1, executing across Plan, Buy and Sell. Each maps to one or more Reliance platforms. Each carries data, design patterns, and evals. More agents to follow.

02 · The agents · by value-chain track
Plan · 3Demand sensing, planning, range and assortment intelligence.
BuildingL4
Cortex Planning Agent
Store-realistic Plan of Record + continuous, exception-led replanning.
L4 Agentic · Workflow level
BuildingL4
Retail Vista
All-India spatial intelligence — site discovery, scoring, and a conversational workspace for the expansion team.
H3 grid · Spatial fabric
BuildingL4
Category Intelligence Agent
Category × season × geography → 40-50pp evidence-tagged trend report. Built for buying teams, not for press.
7 steps · Pipeline
Buy · 1Design-to-buy pipeline.
LiveL4
Trend-to-Design Agent
Trend brief → manufacturable design package — image, BOM, costing, line sheets — without humans in the middle of the cycle.
718 designs · Trend business run
Sell · 2Marketing and cataloging.
BuildingL4
Marketing Agent
Goal in plain English → audience, creative, channel mix, send time, measurement — autonomously, with humans only on guardrails.
11 agents · Agent topology
LiveL4
AI Cataloging Agent
Mannequin or product image → enriched catalog page with model-on photo, attribute set, and motion video. 5 hours, not 15 days.
15 days → 5 hr · Cycle compression
03 · The autonomy matrix · where each agent sits

6 of 21 workflows reach L4 in the 90-day plan. The remaining 16 sit at L2-L3 deliberately — strategy stays human, brand stays human-curated, compliance stays human-accountable.

Autonomy matrix · 21 retail workflows × L0-L5
Canonical home: /autonomous
04 · The 10 building blocks · how every agent is constructed

Verbatim from the Agentisation note · Appendix B. Every agent on this catalog is built from some combination of these primitives.

Tool Calling
Agent invokes external tools/functions to read state and write back to systems.
Standard Tool Interfaces (MCP)
Common contract for tool discovery, schemas, auth — agent and tool teams decoupled.
Context Engineering
Choosing, structuring, and compressing what the agent sees per turn.
Agent Memory
Short-term · episodic · semantic · procedural memory across runs.
Context Graphs
Entities and relationships for multi-hop reasoning over retail decisions.
Wide Research
Generate many candidate hypotheses, score broadly, then converge.
Deep / Hierarchical
Planner → specialist sub-agents → integrator.
Multi-Agent Orchestration
Peer agents propose · critique · enforce · arbitrate.
Observability
Inputs · context · tool calls · outputs · errors captured as traces.
Evals
Golden tests · simulations · replay · adversarial · policy compliance.