Nexly Agents is an AI customer support agent, an AI chat agent for customer service, built by Nexly Lab in Warsaw, Poland. One agent answers your customers 24/7 in their language, only from knowledge your company approves, and shows the source of every answer. When it does not know, it says so and hands the conversation to your support team with full context. Analyst agents read every conversation and report what customers ask about.
What an AI customer support agent is
An AI customer support agent is software that answers customer questions in natural language, around the clock, using your company's knowledge: website, documents, manuals, product data and FAQ. It resolves the repeatable questions on its own and passes the rest to people.
It is different from a rule-based chatbot, which follows scripted flows and buttons, and from live chat, where every message is answered by a person. An AI support agent understands free-form questions, finds the answer in your sources and knows when to stop and call a human.
AI chat agent, chatbot or live chat
| Rule-based chatbot | AI customer support agent (Nexly Agents) | Live chat | |
|---|---|---|---|
| Who answers | Scripted flows and buttons | AI, from your approved knowledge | Your team |
| Free-form questions | Only the ones in the script | Yes, in 8 languages | Yes |
| Outside working hours | Yes, within the script | Yes, 24/7 | No, unless you staff it |
| When it does not know | Falls back to a menu | Says so and hands off to a person with context | Not applicable |
| Channels | Usually one | Website, email, WhatsApp, Messenger, Telegram, Signal, Teams, Slack | Usually website |
What Nexly Agents does in customer support
- One agent for every channel: website chat widget, email, WhatsApp, Facebook Messenger, Telegram, Signal, Microsoft Teams and Slack, with one knowledge base and one set of rules.
- Answers only from your knowledge, with sources. You switch documents, pages and Q&A on and off with one click.
- Human handoff with full context, for example as a ticket in Zoho Desk or Jira.
- Messages center for your support team: conversations from messengers in one inbox, queues (to answer, mine, waiting on customer, closed), a waiting-time counter, assignment to a team member and taking over a conversation with one click.
- Quality you can check: a set of control questions built from your documentation; every change of the agent is tested on it, with expected and actual answers side by side.
- Customer conversation insights: a daily report of top questions, problems, missing information and lost sales, with a recommendation of what to focus on today.
Who uses it
Online stores and sales networks. In one deployment the agent serves 40,000 users in 8 languages, and thousands of them start their day with content the agent prepares, such as a day plan or a commission calculator.
Manufacturers and technical support. A manufacturer of technical systems gave the agent 2,108 pages of technical documentation. Installers and service technicians get answers only from approved documents; questions the documentation does not cover go to the manufacturer's support team.
B2B support in Microsoft Teams and Slack. Partners, distributors and key accounts write to the agent where they already work. See AI agent for Microsoft Teams.
When Nexly Agents is not the right fit
If you want a self-serve tool you set up yourself in an afternoon, with public monthly plans, a self-serve AI chatbot or helpdesk will suit you better. Nexly Agents is set up for you by the Nexly Lab team, which prepares the knowledge, tests answers and keeps improving the agent. See our comparisons with Intercom Fin, Tidio and Chatbase.
How implementation works
You start with a free demo of the agent on your own website: you give us your website address and e-mail. A working agent on your knowledge is ready 2 to 3 weeks after you hand over the materials. Pricing is individual and depends on the scope of knowledge, integrations and hosting.
Data and GDPR
Personal data is processed under a GDPR data processing agreement (DPA). Personal data is masked in conversation analytics, technical logs are cleaned of e-mails and card numbers and deleted after 90 days, and the model API we use does not train on the data sent to it. You can run the agent on Nexly Lab infrastructure or self-hosted. More on security and GDPR.