Connect your data: files, API and automatic mapping
Which tables to load, what each one unlocks, how column mapping and quality reports work, and how to keep data fresh.
Agents work on your own products, stores, orders, stock and policies. You load them once into a standard data model. Your files can keep your own column names: the platform maps them for you.
What to load, and what it unlocks
| Table | Minimum columns | Unlocks |
|---|---|---|
| Products | sku, name, category | Personal Shopper, Shop the Look, merchandising agents |
| Stores | store_id, name, city (+ opening hours, time zone) | Store Locator, Store Staff Copilot |
| Customers | customer_id | Loyalty Assistant, Clienteling, segments |
| Orders + Order lines | order_id, customer_id, order_date / order_id, sku | order status, returns, sales analytics |
| Inventory | sku, location_id, on_hand | Stock-out Sentinel, availability |
| Promotions | promo_id | Promotion Analyst |
| Knowledge docs | .md / .txt files |
Customer Care, FAQ, HR assistant |
Formats: Parquet (recommended), CSV, TSV and JSON. Policies and FAQs: Markdown or plain text.
1. Upload a file
In the console, open Data, choose the table (for example Products), pick your file and click Upload & map columns.
2. Check the column mapping
The platform proposes a match for each standard field (for example item_code → sku, sale_price → price,
image_link → image_url) and shows sample values from your file. Required fields are marked; fix anything with the
dropdowns, then click Process.
3. Read the quality report
Processing takes seconds (products a little longer, because they're indexed for AI search). The report shows rows loaded, duplicates removed, empty or unreadable values per field, and columns that weren't used.
Keeping data fresh
Upload the same table again whenever it changes. Each upload is a new version; the newest successful one becomes active automatically and agents use it from the next question. A knowledge document with the same file name replaces the old one.
Automate it
Push files from your own systems with the API:
curl https://servicebuddy.com/api/v1/datasets -H "Authorization: Bearer $API_KEY" \
-F entity=product -F [email protected]
See API keys and integration. Connectors for e-commerce platforms such as Shopify and Magento are on our roadmap.
Popular AI Agent Buddies
Customer · Discovery & shopping
Personal ShopperFinds products by intent, budget, colour and occasion; compares options and recommends the best fits.
Customer · Service
Customer CareAnswers policy questions from the retailer's own documents, checks orders and return eligibility.
Customer · Service
Order Status (WISMO)"Where is my order?": status, dates and what happens next.
Customer · Service
Store LocatorNearest stores, live open/closed status in local time, hours and services.
Customer · Discovery & shopping
Shop the Look StylistBuilds a complete, coordinated outfit around any product or occasion using only real catalog items.
Merchandising & assortment
Merchandising CopilotAnswers assortment and performance questions with numbers, then recommends actions.
Inventory & supply chain
Stock-out SentinelFlags SKUs at risk of stock-out from on-hand stock vs recent sales velocity.
Store operations
Clienteling AssistantPrepares a client brief (tier, past orders) and suggests products to show a returning customer.
Marketing & CRM
Segment BuilderBuilds customer segments from tiers, cities and purchase behaviour, with counts.
Related articles
How onboarding works: from first call to live AI agents
The six steps from a first conversation to agents serving your customers: workspace, data, agents, testing, launch and team. Most retailers are live within days.
Install and configure agents for your business
Browse the agent catalog, install the agents your data supports, and tune them with your own instructions.
Test your agents in the Playground before going live
A private chat with all your agents that shows which agent answered and which tools it used, plus a pre-launch checklist.
Integrate with the REST API: agents and data in your own systems
Create an API key, call any agent (with streaming), push data automatically, and the full endpoint list.
Invite your team: roles and permissions
Give every colleague their own login with the right role: owner, admin, analyst or viewer.
Languages and international setup
Console, shopper and catalog languages explained, with recommended setups for Europe, the Gulf, India and South-East Asia.
Launch the AI shopper assistant on your website or a link
Choose languages and brand, then paste one line of code or share a hosted link. What shoppers see, and how to pause it.

















