03Orchestration Layer

AI agents that run
your retail workflows.

Every recommendation becomes an action. Agents execute, notify, and iterate — inside Solya today, across your tools tomorrow.

The orchestration problem

Good decisions die waiting for someone to act.

The hardest part of retail isn’t deciding — it’s getting the decision into the system that executes it, week after week.

Dashboards don’t do the work.

A chart that says “transfer 38 units to Lyon” doesn’t move 38 units. Someone still has to open the ERP and type.

Every task is repeated every week.

Same query, same export, same email, same approval, same write-back. Five hours of one buyer’s week, every week, forever.

Nothing closes the loop.

The action gets taken (sometimes). The outcome shows up (eventually). Nobody connects the two. The model never learns.

How it works

Inside the Orchestration Layer.

Three scopes. Agents act where your retail runs.

01Agents in SolyaActive

Agents that run inside Solya.

Pre-built agents for replenishment, allocation, markdown and ops. They run on your metrics and rules, and you stay in control of what auto-executes vs what needs approval.

  • Run on your metrics, signals, and constraints
  • Human approval where it matters, auto-execute where it doesn’t
  • Every cycle logged, every action reversible
01 · Agents in Solyarunning
AReplenishment
142 cycles · 24h
AAllocation
38 cycles · 24h
AMarkdown
17 cycles · 24h

Awaiting approval

Markdown −20% · Trail Boot · 142 SKUs

Impact: +€3.7k

02Agents in your appsAvailable soon

Agents that act across your tools.

Extend agents into the tools your team already uses — Slack, email, Gmail, Zendesk, Klaviyo. The agent does the work; the team sees it happen where they already are.

  • Plug-and-play with your existing stack
  • Same logic, same audit trail, new surface
  • Customer Care, Marketing Ops, Store Ops
02 · Agents in your appssoon
Sl

Slack

Posted summary to #ops

Gm

Gmail

Drafted supplier email

Ze

Zendesk

Replied 12 tickets

Same logic · same audit · new surface

03Agents in core systemsAvailable soon

Agents that write to your ERP, OMS, POS.

The final mile. Agents that take action in the system of record — orders, transfers, price changes — with full two-way sync and rollback.

  • Full two-way sync with SAP, Oracle, Cegid, Ginkoia, Lightspeed
  • Actions executed at the system of record
  • Per-action audit trail and one-click rollback
03 · Agents in core systemssoon
POSTSAP · /transfers38 units
PATCHCegid · /prices−20%
POSTOracle · /ordersMOQ × 6
SAPOracleCegidManhattan
Capabilities

The automation engine for every retail workflow.

Six capabilities to deploy, supervise and scale agents safely.

Pre-built agent library

Replenishment, allocation, markdown, ops. Ready to run on your data.

Custom agent builder

Describe the workflow, ship the agent. No code, full audit trail.

Human-in-the-loop controls

Per-workflow thresholds for what auto-runs vs what needs approval.

Full audit trail

Every proposal, override, approval and write-back, with reasoning.

Expanding integrations

Slack, Gmail, Zendesk today. SAP, Oracle, Cegid, Manhattan next.

Continuous feedback loop

Outcomes feed back into the model. Agents improve every cycle.

Fit into the runtime

The automation engine of the Solya runtime.

Decisions executed where work actually happens — your systems, your tools, your team.

01

Data Layer

Unified retail model

02

Intelligence Layer

Decisions, your rules

03

Orchestration Layer

Agents that act

You are here
04

Application Layer

Role-specific apps

Retail was never built for AI.
We’re rebuilding it.

I had one person whose actual job was to copy data from one tool to another. That role is gone. They’re running a category now.
TB

Thomas Baudouin

COO · 12 stores

Solya

Stop watching the dashboard.
Start running the decisions.

A 30-minute walkthrough on your own data shape. We’ll show what Solya would decide for your network this week.