Live
Shopify e-commerce, built and managed end to end · Premium apparel from India
- E-commerce
- Shopify
- D2C
Retail and e-commerce · Customer conversations
AI shopping concierge for web and WhatsApp
A Claude shopping assistant on your Shopify store and WhatsApp that answers size, stock, delivery and returns questions from live store data, then hands off to your team.
- ClaudeAnthropic
- GeminiGoogle
- Typical timeline: 3–10 weeks
The problem
Shoppers leave when they can't get a fast answer on sizing, stock, delivery or returns, and your team answers the same questions across chat, email and WhatsApp.
- The same handful of questions arrives on every channel, all day.
- Stock and delivery answers go stale the moment they are copied into a FAQ.
- Out-of-hours questions wait until morning, and some of those carts never come back.
How we solve it
A concierge on your site and WhatsApp that reads live catalogue, cart and order data from your store, answers in your brand voice and passes complex cases to a person.
It plugs into Shopify, WhatsApp Business, HubSpot or Zoho CRM and GA4 or BigQuery, so nothing you rely on today is replaced. The architecture below shows each step and which model does the work.
Architecture
From your stack, through the AI layer, to numbers you can track
Each step names the model or system doing the work. Writes to your systems wait for a person to approve them.
Your stack
- Shopify
- WhatsApp Business
- HubSpot or Zoho CRM
- GA4 or BigQuery
AI layer
Softpiper AI layer
- Claude
- Gemini
- Questions arrive, using Web chat · WhatsApp Cloud API
Shoppers write in the website widget or on WhatsApp through the WhatsApp Business Cloud API.
- Triage and quick answers, using Claude Haiku
A small, fast model classifies each question and answers routine ones from your store policies, held in a cached system prompt.
- Live store tools, using Shopify UCP MCP
Catalogue, cart and checkout come from Shopify's Universal Commerce Protocol MCP endpoint, policies from the store's own policy tool, and order status from the commerce API.
- Multi-step help, using Claude Sonnet
Sizing advice, comparisons and returns move to a larger model that can chain several tool calls.
- Search by photo (optional), using Gemini Embedding
Catalogue images are embedded with a multimodal model, so a shopper's photo finds similar products.
- Hand-off and CRM, using HubSpot / Zoho MCP
Conversations a person should take go to your team with the transcript; contacts and outcomes are written to your CRM through its MCP server.
Outcomes you measure
- Containment (deflection) rate
- Assisted conversion rate
- First-response time
- Cart-recovery rate
Why this model
Claude and Gemini, each where it leads
We choose per task on capability, cost and where your data must stay. Final versions are picked at kickoff.
ClaudeAnthropic
Claude follows a detailed brand and policy brief closely and calls strict, schema-checked tools, so answers come from your store's systems rather than from memory. A small, fast Claude model handles routine questions and a larger one takes over multi-step help.
GeminiGoogle
Optional: Gemini's multimodal embeddings let shoppers search by photo (“something like this, in blue”) across your catalogue images.
Delivery
What you get, how long it takes, and how we keep it safe
Guardrails
- Stock, prices and delivery dates are always read from the store, never generated.
- Refunds, discounts and order changes need staff approval.
- WhatsApp messages follow Meta's opt-in and template rules.
- The assistant says it is an AI, and a person can take over any conversation.
Plus the guardrails on every build: facts from your systems, approvals for writes and money, a human one step away, audit logs and evals. See all
Deliverables
- Concierge agent with Shopify, order and CRM tools
- Website widget and WhatsApp channel
- Hand-off queue for your team
- Evaluation set built from your real customer questions
- Dashboard for containment, conversion and cost
Typical timeline
- Pilot on web chat3–5 weeks
- Production with approved WhatsApp templates6–10 weeks
A planning range for a first production release. We confirm scope, price and dates in a tailored proposal.
KPIs to track
- Containment (deflection) rate
- Assisted conversion rate
- First-response time
- Cart-recovery rate
- Customer satisfaction (CSAT)
- Escalation accuracy
- Cost per resolved conversation
Goals we agree with you and measure from a baseline. They are what we track, not results we promise.
Related services
The services behind this playbook
AI agents
Agents that answer, act and hand off, on Claude and Gemini, inside the tools you already run.
View serviceE-commerce
Storefronts with character that load fast, convert on mobile and run on payment rails your customers trust.
View serviceIntegrations
Connect your CRM, ERP, commerce, payments and messaging so data moves once, correctly.
View service
Our work in this space
What we've built for retail and e-commerce
Sites and products we've shipped for businesses in this sector. They show how we work; they are not deployments of this playbook.
Archived preview
More playbooks
Related playbooks
Retail and e-commerce
Catalogue and content factory
Thousands of SKUs with thin descriptions, missing attributes, no translations and inconsistent imagery hold back search visibility and conversion.
- Claude
- Gemini
- Typical timeline
- 2–4 weeks
- Plugs into
- Shopify · WordPress / WooCommerce · Google Drive +1 more
Real estate
Lead qualification and property-matching concierge
Portals and ads bring high lead volumes, follow-up is slow, and agents spend their time on enquiries that were never going to buy.
- Claude
- Gemini
- Typical timeline
- 4–10 weeks
- Plugs into
- HubSpot or Zoho CRM · Salesforce · Listings database +2 more
Healthcare
Patient front desk and intake assistant
Clinics lose hours of staff time to appointment calls, pre-visit forms, multilingual questions and reminder follow-ups.
- Claude
- Gemini
- Typical timeline
- 6–10 weeks
- Plugs into
- EHR / HIS scheduling · WhatsApp Business · Google Workspace or Microsoft 365 +1 more
Pip · Softpiper's AI
Ask Pip about this playbook
Ask Pip how this would work for you
Pip is Softpiper's AI. Tell it what you run today and it will walk through how this playbook fits, what we would build first and a tailored price range. Or book a call with the team.