● Production · 3 DACH instances · client under NDA
An intelligent customer service hub for e-commerce.
A CRM with built-in AI that brings orders from Magento and PrestaShop, emails from Office 365, customer history and automated replies together in one place. The consultant sees everything within 5 seconds — order history, latest messages, fulfilment status, team notes. The AI drafts a ready reply to every customer question. Average response time drops from 4 hours to 15 minutes.
Status
● Production live
Client
E-commerce shop, 3 countries · NDA
Markets
PL · DE · AT
Response time
4h → 15 min
Stack
Next.js · Supabase · Claude API · M365
CRM table
Every order. Every channel. One table.
Top: filters STATUS / ACTION / STAGE / LAST EMAIL / LAST CALL / CONTACT / DATE. Main table: order number, last contact, B2B customer (Firma Alpha / Beta Solutions / Gamma Group), contact person's email and phone, product, pallets/shipped/amount, status, stage, deadline, commitments. The consultant filters "Urgent + unanswered + DACH" in a single click.
crm.studioapps.pl/zamowieniaOrders · Table
SAMPLE DATA
Order table · 7 filters · status + stage + deadline · B2B customers anonymised (Firma Alpha / Beta Solutions / Gamma)
Product numbers
What Customer Service AI actually does.
Figures from a genuine deployment. The client is an e-commerce shop with a product portfolio across 3 markets — it requires an NDA. We will show the originals on a call.
4h → 15min
Average response time
-94% time
3
M365 mailboxes integrated
PL · DE · AT
2
AI modes
semi-auto + auto
2
E-commerce platforms
Magento + PrestaShop
∞
Contact history per customer
timeline · chat · notes
AI Concierge
A message from a customer. An AI reply in 3 seconds.
The consultant opens the order. On the right they see the customer's message: "please email me today the exact delivery date for the two orders 50000123456 and 50000123457". They type a hint for the AI: "Tell the customer that delivery will be next week". They click "Ask the Agent for a reply" → Claude generates a complete, professional email with the order history, the company context and the right tone of voice. The consultant reads it, edits a sentence if needed, and sends.
SAMPLE DATA
Correspondence drawer · customer message · hint for the AI (optional) · "Ask the Agent for a reply" · full reply in the company's tone of voice
Two working modes
Semi-automatic for control. Fully automatic for scale.
The company decides how much responsibility it hands over to the AI. Most clients start in semi-automatic mode (the consultant approves every reply); after 2 months, once the AI is at 95%+ accuracy, they switch routine questions over to full automation.
01
Semi-automatic mode
The AI prepares a ready reply; the consultant reads it, edits it and approves it. Every message passes through a human before it goes out. Default for new clients — 2 months to build trust in the model.
Customer emailAI draftsConsultant approvesSent
02
Automatic mode
The AI answers predefined categories of question on its own ("when is delivery?", "can I get a receipt?", "change of address"). The consultant steps in only for edge cases. 95%+ accuracy required before it is switched on.
Customer emailAI classifiesAuto-replyAudit log
Customer history
Every contact. Every order. All in one timeline.
The "History" drawer shows the full timeline: messages synced from Office 365, phone calls (logged), chat, consultants' notes, agreements. Every action carries a timestamp, an author and a status. A consultant sees straight away what a colleague agreed back in March and picks the case up without asking the customer "what did we agree back then".
crm.studioapps.pl/zamowienie/5000087421/historiaDrawer · History · Timeline
Instead of 5 tools — one screen for the consultant.
Before the rollout, a consultant switched between the mailbox, the Magento panel, the PrestaShop panel, notes in Excel and the company messenger. After the CRM went live — everything in one browser window.
M
Magento 2.4
REST · orders · customers · status
P
PrestaShop
webhook · orders · customers
M365
Microsoft 365
Graph API · 3 company mailboxes
C
Claude API
Opus 4.6 · auto-reply + classification
Customer data + statuses
Everything about the customer in 5 seconds.
The "Data" drawer: name, email, phone, postcode/city, product, amount. Status section: case status (Standard / Urgent / Critical), service stage (New message / Replied / Follow-up / Closed), action (To ship / To contact / Done), commitment status (Awaiting decision / Approved / Cancelled). 4 dimensions of state — no customer is ever left hanging.
crm.studioapps.pl/zamowienie/50000123482Drawer · Customer data
SAMPLE DATA
Customer data · 6 fields · status section (4 dimensions of state) · Case status + Service stage + Action + Commitments
9 modules
What Customer Service AI brings together on one screen.
Designed together with the ops manager of a real e-commerce shop. Every feature came out of a concrete blocker in daily work, not out of an abstract roadmap.
AI Concierge — reply generator
Claude analyses the order history, the correspondence, the status and the company's procedures → and generates a complete reply in the company's tone of voice. The consultant edits a line if needed and sends it.
Intelligent inbox
Every email from the 3 M365 mailboxes lands in the CRM. Automatic matching to a customer or order by email address, company name or order number in the text. Unrecognised messages go to a "shared inbox" for manual assignment.
Full contact history
A timeline per customer: every email, phone call, chat, consultants' notes and agreements. Every action tagged with a timestamp and an author. Nobody asks "what did we agree back then" — it is all in one view.
4 status dimensions
Case status (Standard/Urgent/Critical) · Service stage (New/Replied/Follow-up) · Action (To ship/To contact/Done) · Commitments (Awaiting decision/Approved). Filtering on any combination.
Deadline tracker
Every order has a fulfilment deadline. A red highlight 48h before the date, "overdue" marked in red. Sort by deadline → in the morning you see what has to be done today.
Custom procedures per company
The company defines its own communication rules ("never promise a specific delivery date without confirmation from the warehouse", "always offer a 5% discount on complaints up to PLN 200"). The AI learns those rules and respects them.
Audit trail for compliance
Every status change, every message sent, every note — stored with an author and a timestamp. Who did what, and when. A customer can request an export of their history (GDPR).
Multi-language communication
3 mailboxes: PL/DE/AT. The AI writes replies in the customer's language (recognised from the mailbox). A customer writes in German — the reply comes back in German, in DACH style.
Multi-tenant with RLS
Every client's data lives in an isolated space (Postgres Row Level Security). Zero chance of crosstalk between e-commerce business A and B. Anonymised fake brands in the UI (BuildFlow / StudioDemo / StudioCRM).
8 benefits of the rollout
What changes in the team once it is live.
From the ops manager's own product description — what genuinely improved in customer service after Customer Service AI went live.
01
Faster customer service
Access to all the information without digging through mailboxes and sales systems — instantly.
02
Better organisation
All communication and the whole customer history in one place. No jumping between tools.
03
Process automation
The AI takes over repetitive tasks (when is delivery, can I get a receipt, change of address) — consultants get on with the important work.
04
Personalisation
Every reply is generated from the customer's history and the current situation. Not a template, not generic.
05
Quality control
The company defines its own communication rules and how the AI operates. The manager supervises, the AI executes.
06
Fewer mistakes
The system keeps an eye on statuses, deadlines and history — nothing is left hanging for a week.
07
Business scaling
Serving more customers without growing the team proportionally. An AI multiplier.
08
Team efficiency
All the work in one environment — no switching between 5 tools. Less fatigue, fewer mistakes.
"This is not just a CRM. It is an intelligent operations hub for customer service — sales, communication, automation and AI in one place."
Customer Service AI · production · client under NDA · 3 DACH markets
Landscape
Customer Service AI vs. global CRM tools.
Zendesk, Intercom and Freshdesk are the market leaders — mostly EN, SaaS subscription from €30/month/agent. Our CRM is for operators in PL/DE/AT who want custom development around their own workflow + AI in Polish and German + integration with Magento + M365 with 3 company mailboxes.
Feature
Customer Service AI
custom dev · NDA
Zendesk
$55/agent/mo
Intercom
$74/seat/mo
Freshdesk
$15-79/agent/mo
Native Magento integration
yes · REST
app marketplace
app marketplace
app marketplace
Office 365 multi-mailbox
3 mailboxes · Graph API
yes
yes
yes
AI auto-reply in Polish
Claude · natural PL
EN-first
EN-first
EN-first
Custom procedures per company
in the system prompt
macro + rules
workflows
automations
Multi-tenant with RLS
Postgres RLS
enterprise only
enterprise only
enterprise only
Pricing model
custom dev
per agent SaaS
per seat SaaS
per agent SaaS
The code is yours
yes · self-hosted
vendor lock-in
vendor lock-in
vendor lock-in
Stack
What Customer Service AI is built from.
The standard StudioApps stack + Microsoft Graph API for M365 + Magento REST + PrestaShop webhooks. Everything self-hosted, and the code belongs to the client.
Next.js 16
App router
React 19
UI
TypeScript
strict
Supabase
Auth + RLS
PostgreSQL
Multi-tenant
Tailwind 4
Styling
Claude API
AI auto-reply
MS Graph API
Office 365 · 3 mailboxes
Magento REST
Orders + customers
PrestaShop
Webhook + REST
pg_cron
Email sync 10 min
Vercel
Hosting EU
Honestly
What Customer Service AI does not have yet.
The product has been in production for months, used inside a real company. It works, it is good — but it is not everything. Here is what sits on the backlog, so you know what you are getting.
RoadmapVoIP telephony — calls are currently logged by hand after the conversation. Integration with a PBX (Zadarma / 3CX) is on the roadmap.
RoadmapWhatsApp Business / on-site chat — currently email and phone only. Multi-channel in Q3 2026.
LimitNo Shopify / WooCommerce integration — Magento and PrestaShop only. Shopify would be 5 days of work.
By designNo public SaaS plan. We build per client — the quote depends on the scope; hosting and the AI model are billed at supplier cost.
UI gapReporting — currently basic metrics (how many messages, average time). A "manager view" dashboard with trends and alerts is on the roadmap.
DeliberateNo voice AI for phone calls — text replies only. An ElevenLabs integration like the one in opiekunek is not in scope for this CRM (a different use case).
Complementary products
After customer service — outbound communication.
Customer Service AI handles incoming emails with complaints and questions. Mailing Suite puts the same set of customers to work in outbound campaigns — a lead universe of 1.48M, behavioural segmentation across 9 dimensions, cohort tracking from engaged to customer. The same CRM data, two directions of communication.
Complementary product
Mailing Suite — CRM + Email Marketing
A CRM connected to an email marketing platform. 9-dimension behavioural segmentation, real-time analytics, drip sequences, 2-way inbox with M365. Flagship campaign: 53% open rate · 44% CTR.
An e-commerce shop with 500+ orders a month? We will show you it live.
A no-obligation conversation: 30 minutes, a screenshare from a DACH client's production system (under NDA), and you will see the AI Concierge write a reply to a real message in 3 seconds.
