How do I train an AI agent on 1CS?

Connect your website URL, upload documents, or add FAQ entries to the knowledge base. The AI indexes your content automatically using vector search (RAG) and is ready to answer in minutes.

How does 1CS bill for AI agent usage?

1CS uses a credit-based, pay-as-you-go model. You pay per AI resolution — there are no monthly seat fees for AI. Credits never expire.

Can the AI agent analyse images sent by customers?

Yes. The AI uses vision models to analyse photos uploaded by customers — useful for product troubleshooting, damage assessments, size comparisons, and visual FAQs.

What happens when the AI cannot answer a question?

You configure escalation rules that trigger when the AI is not confident. Options include: leaving the conversation unassigned, routing to a specific agent, or distributing via round-robin to the team.

Can I test the AI agent before going live?

Yes. You can preview and interact with the AI agent inside the 1CS dashboard to verify accuracy before activating it for customers.

Does the AI retain conversation history when handing off?

Yes. When the AI escalates to a human agent, the full conversation history — including what the AI said and the reason for escalation — is passed along so agents can pick up without asking the customer to repeat themselves.

AI Agents

Let AI handle the routine. You handle the rest.

Deploy AI agents trained on your knowledge base. They answer questions, analyse images, and hand off to a human agent — with full context — the moment it matters.

Start free → See pricing

Northpine Support · AI Agent

Trained on 248 articles · gpt-4o-mini

Try it live

My package is 3 days late and I need it for a wedding tomorrow.

I'm so sorry about the delay — that's stressful timing! I've flagged your order for expedited review. Our team will contact you within the hour with a resolution or a same-day alternative.

↗ Escalated to human · Maria · context attached

Does this come in size M?Do you ship to Norway?I need a refund of $80.

AI online · responds in ~3s · 1 credit per resolution

Most customer support questions are repetitive: order status, return policies, pricing, delivery timelines. These questions cost your team time but rarely require human judgement. 1CS AI Agents answer them automatically — trained on your actual business content, available 24 hours a day, and billed only when they successfully resolve a query. When a question exceeds the AI's confidence, it hands off to a human agent with the full conversation history attached, so the customer never has to repeat themselves.

Key capabilities

✓ RAG knowledge base with semantic vector search
✓ Image analysis via vision models
✓ Configurable escalation: unassign, assign, or round-robin
✓ Credit-based pay-as-you-go — no monthly seat fees

How AI Agents works

Four steps from setup to first resolution.

01

Build your knowledge base

Paste your website URL (auto-crawled), upload PDF documents, or type FAQ entries manually. The system indexes all content using vector embeddings so the AI can search semantically — finding relevant answers even when the customer words their question differently from how you wrote the content.
02

Configure the AI agent's brain

In the Edit Brain panel, define the AI's name, tone (formal, casual, friendly), objectives, and the exact conditions under which it should escalate to a human. You can also restrict the AI to specific inboxes or channels.
03

Test thoroughly before going live

Use the Chat Preview panel to simulate real customer questions — pricing, refund policies, edge cases, and scenarios that should trigger escalation. Adjust the Brain until every answer is accurate and every escalation fires correctly.
04

Go live, monitor, and improve

Activate the agent on your chosen inboxes. Monitor live conversations from the admin dashboard, track credit usage per resolution, and refine the Brain as new question patterns emerge. The AI improves as your knowledge base grows.

How 1CS AI Agents differ from simple chatbots

Traditional chatbots follow rigid decision trees — if the customer deviates from the script, the bot breaks. 1CS AI Agents use large language models combined with retrieval-augmented generation (RAG), meaning they understand natural language, handle follow-up questions, and generate contextually appropriate responses rather than picking from a list of pre-written answers. The result is a conversation that feels significantly more natural than a typical support bot.

Credit-based pricing: what it means in practice

Unlike per-seat pricing where you pay for every agent regardless of how much they work, 1CS charges one credit per AI resolution. A resolution is counted when the AI successfully handles a complete customer interaction — not per message. This means high-volume months cost more, but quiet periods cost less. Credits never expire, so you can buy in advance during sale periods and use them at your own pace.

Image analysis with vision models

Customers frequently send photos: a damaged product, a receipt, a screenshot of an error, or a photo of what they want to order. 1CS AI Agents use OpenAI vision models to analyse these images and respond accordingly — describing damage for a warranty claim, reading receipt details to verify a purchase, or identifying a product from a photo. This capability eliminates a category of questions that previously required human review.

Human handoff: what happens and what the agent sees

When the AI decides to escalate — either because it's not confident, the customer requests a human, or an escalation rule triggers — it assigns the conversation to the appropriate agent or team according to your configured routing rules. The human agent opens the conversation and sees the complete message history, including every question the AI answered and the reason for escalation. No context is lost.

Who uses this

Teams that get the most value from AI Agents:

→ After-hours support that resolves tier-1 questions without a human on duty
→ High-volume e-commerce shops deflecting order-status and return queries automatically
→ SaaS companies reducing ticket volume before it reaches the human support queue
→ Marketplaces triaging and routing conversations to specialist teams at scale

Set up AI Agents

Step-by-step guides from the 1CS Help Center:

Add content to knowledge base

Upload documents, paste URLs, or write FAQ entries

Configure AI Agent Brain

Set tone, objectives, and escalation rules

Write escalation rules

Define when the AI should hand off to a human

Set up human handoff

Configure handoff behaviour and agent routing

Set up conversation routing

Route escalated conversations to the right team

Frequently asked questions

Everything you need to know about AI Agents in 1CS.

How do I train an AI agent on 1CS?: Connect your website URL, upload documents, or add FAQ entries to the knowledge base. The AI indexes your content automatically using vector search (RAG) and is ready to answer in minutes.
How does 1CS bill for AI agent usage?: 1CS uses a credit-based, pay-as-you-go model. You pay per AI resolution — there are no monthly seat fees for AI. Credits never expire.
Can the AI agent analyse images sent by customers?: Yes. The AI uses vision models to analyse photos uploaded by customers — useful for product troubleshooting, damage assessments, size comparisons, and visual FAQs.
What happens when the AI cannot answer a question?: You configure escalation rules that trigger when the AI is not confident. Options include: leaving the conversation unassigned, routing to a specific agent, or distributing via round-robin to the team.
Can I test the AI agent before going live?: Yes. You can preview and interact with the AI agent inside the 1CS dashboard to verify accuracy before activating it for customers.
Does the AI retain conversation history when handing off?: Yes. When the AI escalates to a human agent, the full conversation history — including what the AI said and the reason for escalation — is passed along so agents can pick up without asking the customer to repeat themselves.