If you run a store with real sales and real ad spend, "conversational AI agent" shows up in every vendor pitch in your inbox. The phrase covers three very different products at three very different prices. This page draws the lines, runs the cost math, and points at the part every vendor page skips: your margin.
What a conversational AI agent for ecommerce actually is
The dividing line in every serious definition is action-taking, not chatting. McKinsey describes agentic AI as "a system based on generative AI foundation models that can act in the real world and execute multistep processes" — distinguished from a chatbot precisely by the acting (Solo.io, quoting McKinsey).
Applied to a store: a support widget that can tell a customer how to request a refund is a chatbot. A system that can issue the refund inside Shopify is an agent. Same conversation, different power — and the gap decides what you can actually hand off.
The three layers of conversational AI your store already touches
Most operators are using layer one already without calling it AI.
Layer 1 — platform-native automation you already pay for
The platforms in your stack have conversational and automated AI baked in. Shopify's Sidekick can "handle tasks such as analyzing data, managing orders, or editing products" and works through text or voice chat (Shopify Help Center). Meta's Advantage+ automates audience, placement and budget inside Meta Ads. Klaviyo builds segments from a sentence and ships a customer agent for order tracking and returns (Klaviyo).
The catch: each is powerful inside its own walls and blind outside them. Sidekick can't see your Meta budget; Advantage+ can't read your Klaviyo flows.
Layer 2 — single-surface support agents
The most mature commercial category is customer support, and it's priced per outcome. Gorgias charges per resolved conversation: "Each resolved conversation costs $0.90 on most plans," with higher overage rates (Gorgias). Zendesk's AI agents are billed on "successful outcomes they deliver," with Suite plans starting at $55 per agent per month billed yearly (Zendesk).
Two structural tells: support AI is now priced like a result, not a seat, and both vendors build in a human handoff — an admission that these agents don't handle everything.
Layer 3 — cross-tool AI agents ("AI employees")
The newest layer works across your tools the way a hire would: read the ad accounts and the store and the email platform, reason about them together, and take multi-step action. This is where the profit lives, and also where the hype is thickest — more on that below.
Chatbot vs. AI employee: the distinction that decides your budget
The terms get used interchangeably in marketing copy. They name three different things, and the one you pick sets your cost structure.
The table below is framing built from the sourced capabilities above, not a sourced taxonomy — but each per-resolution figure traces to Gorgias and the VA rates to the sources in the worked example.
| Chatbot / single-surface agent | Virtual assistant (VA) | AI employee | |
|---|---|---|---|
| What it is | Software that resolves requests on one surface | A human contractor, hourly or retainer | Software taking multi-step actions across tools |
| Scope | One channel (the inbox, the widget) | Whatever you train them on | Every tool it integrates with |
| Pricing | Per resolution or per seat | Per hour or per month | Subscription, usually usage-based |
| Fails how | Confidently wrong answers | Slowly, visibly, recoverably | Wrong actions when not grounded in live data |
Three things worth spelling out:
- A chatbot answers; an agent acts. A support widget that narrates your refund policy is not the same product as one that issues the refund.
- A virtual assistant is a person. "Virtual assistant" predates AI and still overwhelmingly means a remote human. The honest comparison is economic, not categorical.
- "AI employee" is a scope claim, not a magic claim. What makes the framing real is cross-tool reach plus goal-direction. Gartner warns about "agent washing" — "the rebranding of existing products such as AI assistants, RPA and chatbots without substantial agentic capabilities" — and estimates only about 130 of thousands of self-described agentic vendors are real (Gartner). The same firm predicts over 40% of agentic AI projects will be canceled by the end of 2027.
For a deeper breakdown of this category, the AI employees for ecommerce guide maps the whole landscape.
What conversational AI automates well — and what it doesn't
Automates well today, because shipping products already do it:
- Tier-1 support. Order status, tracking, and returns questions resolve reliably from structured data. Gartner predicts agentic AI will "autonomously resolve 80% of common customer service issues without human intervention" by 2029 — note the "common" qualifier (Gartner).
- Ad delivery management. Meta and Google already automate bidding and budget inside their platforms; Meta claims businesses see "a 20% lower cost per result on average" with Advantage+ sales campaigns, a vendor figure, not a guarantee (Meta).
- Email flows and catalog edits. Rule-shaped, reversible, checkable work.
Automates poorly, with documented limits:
- High-stakes edge cases. Air Canada's chatbot invented a refund policy; a British Columbia tribunal found the airline liable for negligent misrepresentation and ordered it to pay CA$812.02, rejecting the "separate legal entity" defense (CBC). You own what your AI tells customers.
- Brand and creative judgment, and novel strategy. An agent can run a repricing playbook; deciding to reposition the store is your call.
The best coverage of the support-only slice is in the best AI chatbots for ecommerce customer service.
Worked example: the support-desk math
Say your store takes 300 support conversations a month — mostly order status, returns and product questions — and each human-handled ticket runs 8 minutes. Assume the AI fully resolves half (150), which is an assumption for the arithmetic, not a promised rate.
Option A — a human VA handles everything. At a mid-level offshore rate of about $8 per hour (DDIY), 300 × 8 min = 40 hours × $8 = $320/month. At a fully loaded US rate near $40 per hour (CallForce), the same 40 hours = $1,600/month.
Option B — AI resolves Tier-1, a human takes the rest. 150 resolutions × $0.90 (Gorgias) = $135, plus the helpdesk subscription. The remaining 150 conversations × 8 min = 20 hours → about $160 offshore or $800 US.
The honest reading: against a US cost baseline, per-resolution AI is dramatically cheaper on routine volume. Against a $6–$10 offshore VA, the dollar gap on 300 tickets is small — the stronger arguments there are instant response and zero management overhead, not price. And neither option removes you; Option B just concentrates your attention on the hard half.
The profit angle every vendor page skips
Here's what the conversational-commerce pages never get to: answering shoppers faster doesn't tell you whether a sale made money. That's a cross-tool question, and it's where an AI employee earns its name.
Victor, the AI employee from PodVector AI, works across the tools a human operator would. Victor sees every order, every ad dollar and every supplier cost together, so he computes the true per-order profit on each product. He pauses the losing Meta ad set, shifts budget in Google Ads, reprices the thin-margin SKUs in Shopify, builds and sends the Klaviyo flow, and drafts the support reply — then saves the report to your own Google Drive.
Say you sell a shirt at a $31 average order value across 340 orders a month, with $2,800 in monthly Meta spend. A chatbot can tell a customer where their shirt is. It cannot tell you that the shirt's Printify cost, Shopify fees and blended ad cost leave $4.10 of real profit per order on one SKU and a loss on another. Victor connects to Shopify, Meta Ads, Google Ads, Printify, Printful, Gelato and Klaviyo, so product cost is known before the first sale and the losing SKU gets fixed, not just reported. Victor is not a dashboard you read; he does the work.
That cross-tool reach is the whole point of layer three — and the defensible outcome is time, not a revenue promise. Structured, checkable work moves off your calendar; what you do with the reclaimed hours is where the P&L moves. When you're ready to see it on your own numbers, start with Victor.
FAQs
What is a conversational AI agent for ecommerce?
It's software that reads a natural-language message and takes an action in response — resolving a support ticket, updating an order, or adjusting a campaign — rather than only returning text. The action-taking is what separates an agent from a plain chatbot, per McKinsey's definition.
How much do conversational AI agents cost?
Support agents are mostly priced per resolved conversation — Gorgias lists $0.90 on most plans (Gorgias) — while helpdesk suites like Zendesk start around $55 per agent per month billed yearly (Zendesk). Cross-tool AI employees are subscriptions. Run the per-resolution math against your ticket volume and your current VA cost before you commit.
Is a chatbot the same as an AI agent?
No. A chatbot converses on one surface; an agent takes multi-step actions across tools. Gartner calls the practice of relabeling chatbots as agents "agent washing" and estimates only about 130 of thousands of self-described agentic vendors are real (Gartner).
Will a conversational AI agent run my store for me?
No shipping product claims that. Support agents hand off what they can't resolve, platform tools stage changes for your review, and you stay the decision-maker of record. The useful question is which repetitive work you can hand off, not whether the whole operation runs itself.
Who is liable when the AI gives a wrong answer?
Your store is. The Air Canada tribunal held the company — not its chatbot — liable for the bot's misinformation and ordered it to pay CA$812.02 (CBC). Budget review time; it's the new cost that replaces execution time.
What should I automate first?
Start with the work that's high-volume, low-judgment and reversible: Tier-1 support, email flows, catalog edits, and routine ad delivery checks. Keep brand voice, novel strategy and high-stakes edge cases on your own desk, and prefer tools whose output lives in your own Shopify, Klaviyo and Drive so the work survives the vendor. For the full operator's view, see the AI employees for ecommerce guide.