AI Agents · Overview
Docs / AI Agents

Overview

Deploy always-on AI agents that connect to messaging channels, keep persistent memory, and optionally search the web.

NevTan AI Agents are always-on assistants — like the Hermes Agent or OpenClaw — that NevTan runs for you, connected to messaging channels such as Telegram, WhatsApp, Discord, Slack, and Email, with persistent memory across conversations and optional web search.

Each agent runs against a model you choose. Browse the available agent types in the console, pick a plan, and deploy — the agent stays online, listening on its channels and drawing on its memory until you stop it.

Why run it as an agent instead of rolling your own

You could build a chatbot yourself using the Inference Playground and your own server to hold conversation state. An AI Agent is what you get when NevTan manages that whole loop for you as a hosted service:

A support bot on WhatsApp

Say you want customers to message a WhatsApp number and get help with orders, around the clock. Built yourself, that means standing up a server to hold the WhatsApp connection, storing each customer's conversation history somewhere, and watching your own logs when a reply goes wrong. As an AI Agent it's a managed deploy: create the agent with your AI provider key and turn on memory, then connect WhatsApp from the agent's own dashboard by pasting in the number's credential, and it's live — remembering a customer across messages. Pausing it overnight or restarting it after a config change is a button, not an on-call page.

A Slack assistant for a team

A Slack-facing assistant that answers questions in a channel needs the same pieces — a Slack app connection, a place to keep context between messages, somewhere to see what it's said. Connecting Slack works the same way as WhatsApp: deploy the agent, then paste the Slack bot token into the agent's own dashboard under Channels. Because channels live on the agent itself, adding Slack alongside an existing WhatsApp or Telegram channel (plan permitting) doesn't require a second deployment. (Slack is an OpenClaw channel — Hermes covers Telegram, Discord, and WhatsApp.)

Building it yourselfAI Agent
Where it runsInfrastructure you set up and maintainA hosted agent NevTan keeps online
ConfigurationRe-entered every time you change somethingSet once on a configuration screen — channels, credentials, model, memory, web search
Conversation stateYou build and store it yourselfBuilt-in persistent memory across conversations
Channel connectionsYou build the Telegram/Slack/etc. integrationBuilt-in channel connectors — just paste in your credentials
Uptime & restartsYou run and supervise the processPause, resume, or restart from a button in the console
VisibilityWhatever logging you add yourselfA built-in logs and activity view, plus credit usage

In short: an AI Agent gives you one console screen where configuration, channel connections, memory, and observability are all handled for you — no infrastructure of your own to run.

The lifecycle

Configure
name, type, channels, model
Deploy
launch the agent
Runs
listening on its channels
Logs & activity
watch what it's doing
Update or stop
restart, stop, delete
What AI Agents is (and isn't)
This is deployment-as-a-service for two pre-packaged, open-source agents — the Hermes Agent and OpenClaw — hosted for you as managed containers. It is not a custom agent-builder: bringing your own agent logic is not supported today.

Explore

Note
Agents are billed from your real-money credit balance — the signup bonus is reserved for App Platform & Databases and does not fund agents. See credits for how credits are metered.