Your customers write at 11:47pm.HelpIQ replies at 11:47:18pm.While you sleep.

HelpIQ is the front-office AI agent that reads your knowledge base, your Shopify and your Stripe orders to reply directly to your customers, by email, WhatsApp, chat or Instagram. No transfer. No wait. No copy-paste from a dusty Notion.

auto-resolved tickets
83%
median first response
28s
average CSAT
4.7/5
cost per ticket
-67%

Orders of magnitude the installation aims for; they are measured on your premises during the audit, on your volumes.

HelpIQ, Offer illustration
HelpIQ. Front-office AI for e-commerce and B2C SaaS
00Three conversations

The customer writes at 11:47 pm. The reply leaves at 11:47 pm and 18 seconds.

Every reply is composed with the real order status, your size guide and your return policy, then checked by a guardrail before sending. These three exchanges are fictional; the mechanism is the one we install.

WhatsApp
My dress arrived in size 38 instead of 40. Wedding on Saturday, I'm filing a complaint!
Fictional datareply in 18 s
Email
Hello, I still haven't received parcel 4589, it has been pending for 6 days.
Fictional datareply in 24 s
Chat
What size should I take for this sweater if I usually wear M elsewhere?
Fictional datareply in 11 s
What a ticket costs, with and without HelpIQ

Starting assumption, to be replaced with your own figures during the audit: a ticket handled by a person costs about €7.80; a ticket handled by HelpIQ costs the infrastructure, €0.11.

Monthly difference
€76,900
Installation paid once, then €0 per ticket. The calculation runs in your browser.
01What the agent does

Fully automated, nothing to manage.

01

Answers, doesn't route

Unlike chatbots that say 'an agent will get back to you', HelpIQ writes and sends the full reply: tracking number, return link, refund policy tailored to that specific customer case.

02

Connected to your real stack

Native webhooks for Shopify, WooCommerce, Stripe, Zendesk, Front, Gorgias, WhatsApp Business. HelpIQ checks order #4589 status in real time before replying, no generic small talk.

03

Brand & legal guardrail

Every reply passes a Python policy guardrail that blocks off-brand promises (refund >X EUR, impossible delays) and enforces the tone defined by your Head of Brand.

02How it works

Live in days, not months.

  1. 01

    Week 1-2, Audit & ingestion

    Audit of your last 5,000 tickets, recurring-intent extraction, ingestion of T&Cs / FAQ / guides into pgvector.

  2. 02

    Week 3-4, Wiring & shadow mode

    Shopify / Stripe / Zendesk webhooks live, policy guardrail calibrated with your Head of Brand, shadow mode: HelpIQ proposes, human sends.

  3. 03

    Week 5-6, Auto-send in production

    Switch to auto-send on high-confidence intents (tracking, simple returns, FAQ). Real-time CSAT monitoring, weekly fine-tuning.

03Under the hood

The pipeline behind this offer.

Support & Voice, Triage & EscalationVoice + text, 100% self-hosted open-source, no data leaves, 6 stages
Channels
WhatsAppGmailSlackTéléphone (SIP)
SIP / Webhook
Voice agent, SIP
TwilioRingoverAsteriskDétection de voixTranscriptionSynthèse vocale
VAD → STT → TTS
Understanding
FastAPILLM localTraduction
Hybrid search
KB search
Vector
Embeddingspgvector
Graph
Apache AGE
Reranking
Reranker
Context + SLA
Decision & SLA
FastAPIPostgreSQLRedis
Routing
Response
Auto-resolve
VoixGmailWhatsApp
Human escalation
Slack
04Tools we operate

We plug the agent into the tools you already use.

  • Shopify
  • WooCommerce
  • Stripe
  • WhatsApp
  • Zendesk
  • n8n

Nothing to learn: we configure and operate these connections for you.

A true story

11:47pm on a Sunday. Marie opens her parcel for Saturday's wedding.

What it costs you today

Marie just received her dress for next Saturday's wedding, but it's a size 38 instead of the 40 she ordered. She fires an angry message into your chat: 'This is unacceptable, I'm filing a complaint tomorrow.' Monday morning, your CX agent finds 47 similar tickets piled up. Three customers have already posted on Trustpilot. One cancelled a 340 EUR order. With HelpIQ, at 11:47:18pm, Marie got: 'Hi Marie, I'm so sorry about the size 38. I've generated your pre-paid return label, and your size 40 ships tomorrow on free express, you'll have it Wednesday before the wedding.' CSAT: 5 stars. Trustpilot: saved.

The facts

A level-1 ticket handled by a person costs the time of a read, a search and a reply. HelpIQ does all three in under thirty seconds, at any hour, on a model installed on your premises. What is still billed: the installation, then the maintenance. Never the ticket.

Why us

Wikolabs installs AI systems built on open-source models, on your premises, paid once. No black box: your data, your prompts and your history stay in your infrastructure. We quote at a fixed price or on time and materials, never per ticket or per token, and we say before you sign what works and what does not yet.

What we put in place

Concretely: 6 weeks from signature to auto-send. Week 1-2 audit + pgvector ingestion. Week 3-4 webhook wiring + shadow mode (human validates every send). Week 5-6 auto-send on high-confidence intents, weekly fine-tuning. Deliverable: dashboard with resolution rate, CSAT, cumulative savings, visible to your COO.

Support & Customer Service

24/7 customer service that knows your business rules by heart

E-commerce customer service is a sink of time and costs: order status, delivery times, return procedures, product questions. Hundreds of repetitive tickets your human agents answer every day. The RAG SAV Agent ingests your business rules (Notion base, FAQ, return policy), keeps them in a vector index and replies to every customer via email or WhatsApp in under 30 seconds.

The problem

A human agent handles 40–80 tickets per day. Replies are inconsistent across agents. The cost of quality customer service is prohibitive for SMB e-commerce. And long response times degrade satisfaction and increase negative reviews.

Our answer

The agent is connected to your knowledge base (Notion, Confluence or simple PDF), your e-commerce CMS (Shopify or WooCommerce) for real-time order status, and your contact channels (email via FastAPI, WhatsApp Business). It responds in under 30 seconds, automatically escalates complex cases to a human and documents every interaction.

How we deploy

  1. 01
    Knowledge base ingestion

    Import of your FAQ, return policy, product sheets and business rules into a vector index (Pinecone or Weaviate).

  2. 02
    E-commerce connection

    Shopify or WooCommerce integration for real-time access to order status, deliveries and stock.

  3. 03
    Channel deployment

    Email configuration (IMAP/SMTP), WhatsApp Business API and/or chat widget on your site. Testing on 200+ scenarios.

  4. 04
    Monitoring & continuous improvement

    Dashboard of processed tickets, resolution rate, unanswered questions. Weekly knowledge base updates.

Concrete benefits

80% of tickets resolved without humans

Frequent questions (order status, delivery time, return procedure) are handled automatically 24/7.

Response time < 30 seconds

Every customer receives a precise and consistent response in seconds, even at night and on weekends.

Customer service cost divided by 5

Your human agents only handle complex cases and escalations. Cost per processed ticket drops dramatically.

Frequently asked questions

Can the agent handle multiple languages?
Yes. The agent detects the client's language and replies in the same language, provided your knowledge base is multilingual or auto-translation is enabled.
How are business rule updates managed?
You update your Notion or FAQ, and the agent automatically re-indexes overnight. New rules are active the next morning.
What happens when the agent doesn't know the answer?
It automatically escalates to a human agent with the full conversation context and a suggested response.
Which e-commerce CMS is compatible?
Shopify and WooCommerce natively. PrestaShop, Magento and any CMS with a REST API on request.

Plug HelpIQ in before the next Black Friday

6 weeks from signature to auto-mode. No per-ticket billing. Free audit on your last 5,000 tickets.