---
locale: en
page: agents/platform
seo:
  title: "Platform — the AI agent that makes the changes you approve"
  description: "The Platform agent sets up your store, configures the other agents, and executes Strategist's recommendations — but only after you approve. Every action is logged with the recommendation, your approval, the change, and a revert link."
og:
  title: "Platform Agent — Burbuxa"
  description: "The agent that executes only what you approve, across your whole store — and logs every change with a way back."
  type: website
canonical: "https://burbuxa.com/en/agents/platform"
keywords:
  - ai configuration agent ecommerce
  - merchant approval workflow ai
  - ai store setup automation
  - audit trail revert ai changes
  - burbuxa platform agent
hero:
  eyebrow: "NOTHING SHIPS WITHOUT YOUR APPROVAL"
  title: "Platform"
  role: "It sets up your store, configures the other agents, and turns Strategist's findings into real changes — broadcasts, flows, review cadences, CRM rules. It executes only after you approve, and logs every action with a way back."
  layer: L3
  layerLabel: "Makes the changes you approve"
  cta: "Try free with your store"
trustStrip:
  variant: configurator-approval
  text: "Executes only with your approval. Every action is logged with the originating recommendation, your approval, the change it made, and a link to revert it. Approval is the price of trust."
identity:
  kpisLabel: "Owns"
  kpis:
    - "Recommendation execution rate"
    - "Configuration-change error rate"
    - "Time-to-first-action for new stores"
    - "Audit completeness"
lifecycle:
  title: "Nothing ships until you approve"
  intro: "The Strategist hands over a recommendation. From there, the Platform agent moves in four stages — and the first one is yours. No change reaches your store until you say so, and every change keeps a way back."
  steps:
    - label: "Approve"
      description: "Every change starts with a recommendation and your sign-off. The agent shows you exactly what it intends to do — the segment, the copy, the rule — and waits. Nothing happens until you approve."
    - label: "Execute"
      description: "On your approval, it makes the change across your store: sends the broadcast, builds the flow, sets the cadence, writes the CRM rule. The configuration work is the agent's job; the decision was yours."
    - label: "Audit"
      description: "Every action is logged with the recommendation it came from, who approved it, and exactly what changed. The trust contract is paperwork, not vibes — you can always see what happened and why."
    - label: "Revert"
      description: "Every logged action carries a link back. If a change isn't right, you undo it from the audit trail. No silent state changes, no one-way doors."
inventory:
  title: "What it configures"
  intro: "The Platform agent works across your whole store — not one surface. It writes, but only on your behalf, only with your approval."
  groups:
    - label: "Messaging"
      items:
        - "Sales & Support's brand voice and tone"
        - "Escalation and handoff routing"
        - "WhatsApp, Instagram, and email reply drafts"
        - "PDP copy and metadata (drafts for your review)"
    - label: "Automations & broadcasts"
      items:
        - "Cart-recovery and post-purchase flows"
        - "Broadcast copy and audience segments"
        - "Review-request cadences"
        - "Repurchase and reactivation triggers"
    - label: "CRM & reviews"
      items:
        - "Segmentation and tagging rules"
        - "Customer-history and VIP rules"
        - "Review collection and display config"
        - "Response and follow-up cadences"
    - label: "Integrations & agents"
      items:
        - "Catalog and order-data connections"
        - "Channel and storefront wiring"
        - "The other agents' setup and scope"
        - "First-run configuration for new stores"
comparison:
  title: "Not the automations you set up yourself"
  intro: "Automations you build are static rules in one surface, frozen until you go back and change them. The Platform agent works across the whole store and only ever acts on a recommendation you approved."
  aLabel: "Platform Agent"
  bLabel: "Static automations"
  rows:
    - label: "Scope"
      a: "Works across your whole store — messaging, automations, CRM, reviews, integrations, the other agents"
      b: "One surface, one rule at a time"
    - label: "Why it changes"
      a: "Acts on an evidence-backed recommendation you approved"
      b: "Fires on the trigger you wired, whatever the context"
    - label: "When it changes"
      a: "Adapts as the Strategist surfaces new findings"
      b: "Frozen until you go back in and edit it"
    - label: "Accountability"
      a: "Every action logged with approval and a revert link"
      b: "You reconstruct what changed from memory"
trust:
  title: "Every change keeps a way back"
  body: "Trust is the most expensive thing to rebuild, so the system never asks you to take a change on faith. Every action the Platform agent makes is recorded with the recommendation it came from, your approval, the exact change, and a link to revert it. Nothing is silent, and nothing is one-way. That audit trail is what lets the system do real work across your store without ever surprising you — approval is the price of trust, and the trail is the receipt."
tools:
  title: "Where it works"
  intro: "It writes across these surfaces — but only on your behalf, only with your approval."
  items:
    - name: "Automations"
      href: "/en/automations"
    - name: "Broadcasts"
      href: "/en/broadcasts"
    - name: "CRM"
      href: "/en/crm"
    - name: "Reviews"
      href: "/en/reviews"
    - name: "Integrations"
      href: "/en/integrations"
faq:
  - q: "Does it ever change anything without my approval?"
    a: "No. That's the whole contract. Every change the Platform agent makes starts with a recommendation and your explicit approval. It shows you exactly what it intends to do and waits — nothing reaches your store until you say yes."
  - q: "How is this different from the automations I set up myself?"
    a: "Automations you set up are static rules in one surface, frozen until you go back and edit them. The Platform agent works across your whole store — messaging, automation flows, broadcast copy and segments, CRM rules, review cadences, integrations, even drafts of PDP copy and metadata — and it only ever acts on a recommendation you approved. It does the configuration work; you decide if it's right."
  - q: "What if it executes something wrong?"
    a: "Every action carries a revert link in the audit trail. You undo it from there. No silent state changes, no one-way doors — that's exactly what the trail is for."
  - q: "What exactly does it log?"
    a: "For every action: the originating recommendation, who approved it, the precise change it made, and a link to revert. You can always see what happened and why, after the fact, not just in the moment."
  - q: "Does it write my product page copy?"
    a: "It can draft PDP copy and metadata for your review — it does not rewrite your live pages on its own. Like everything else, the draft is a recommendation you approve before anything ships."
  - q: "Does it set up my store, or only run changes later?"
    a: "Both. It does the first-run setup — connects your catalog, configures Sales & Support's voice and escalation routing, drafts your first automation flows — and then executes Strategist's recommendations over time, always with your approval."
related:
  title: "Where it connects"
  items:
    - href: "/en/automations"
      label: "Automations"
    - href: "/en/broadcasts"
      label: "Broadcasts"
    - href: "/en/agents/strategist"
      label: "Strategist Agent"
    - href: "/en/integrations"
      label: "Integrations"
close:
  recap: "The Platform agent is the operator that turns the Strategist's findings into real changes across your store — but only what you approve, and every action keeps a way back. It's how the system does real work without ever asking you to trust it blindly."
  cta: "Try free with your store"
---