Documentation
Agents
Build, connect and operate voice and text agents.
Basics
- What is an agentAn agent is one saved configuration that answers a caller, holding its persona, its model, its voice, its knowledge, its tools and the channels it is reachable on.
- Create an agentPick one of the three build routes, name the agent, pick its voice, and land on its detail page with every other setting on a default you can change.
- Build an agent with AIAnswer six questions, hand over the documents the agent should know, and a model designs the whole agent while you carry on working.
- The agent overview tabEvery card, number and header control on an agent's Overview tab, and where each one comes from.
- Write the personaSet the agent's identity, write the personality prompt that decides how it speaks, and give it a greeting, all saved as you type.
- How the prompt is assembledWhat the model actually receives at the start of a conversation, which parts you write, which parts the platform appends, and how to write a prompt that survives both.
- Choose a model and reasoning effortPick the model that answers for an agent, set how hard it reasons before it replies, and understand the latency and cost that comes with each choice.
- Voices and languageGive an agent one voice or several, set the default, tune the speech engine per voice, and turn diacritization on for a Fusha voice.
Knowledge
- How knowledge worksKnowledge sources belong to the workspace, each agent picks the ones it may use, and retrieval pulls matching passages into the turn.
- Add documents to the libraryUpload a file to the workspace knowledge library and follow it through indexing until the agent can use it.
- Add text directlyPaste text into the library as a knowledge source when the content has no file behind it.
- Add a web page or crawl a sitePoint the library at a public HTTPS URL, choose a single page or a crawl, and re-crawl it when the content changes.
- Organize knowledge in foldersGroup sources into folders, reuse a folder across agents, and understand what removing and deleting each do.
- The agent did not use my documentWork through the five things that stop a knowledge source reaching a turn, in the order they fail most often.
Tools
- Tools overviewThe four kinds of tool an agent can call, where each one is defined, and how the agent decides to use it during a conversation.
- System toolsThe three built-in tools every agent has, their controls, and how enabling and describing them works.
- Web searchThe built-in tool that lets an agent look something up online during a conversation, and what you can configure about it.
- Dialect detectionThe built-in tool that recognises the caller's Arabic dialect so the agent can adapt to it, and what you can configure about it.
- End conversationThe built-in tool that lets an agent close a conversation itself, and what you can configure about it.
- Webhook toolsGive an agent one HTTP endpoint of your own to call mid-conversation, with the parameters the model fills in and the credentials the portal injects.
- OpenAPI toolsImport a Swagger or OpenAPI document once so the agent can reach a whole API, and follow the import through its three states.
- OpenAPI tool operationsThe operation list stored for an imported API, the per-operation auth modes and fixed values, the enable switch, and the path a call takes through the portal.
- Connect a databaseRegister a read-only PostgreSQL, MySQL or MariaDB connection so an agent can answer callers from live data, and follow it from Reading schema to Ready.
- Review and approve database questionsApprove the questions EchoNova drafts from your schema, test what a caller's wording would reach, and resync when the schema changes.
- Database tool securityHow your connection string is protected, why database tools are read-only, what the agent can see, and what is stored and logged.
- Workspace secretsStore an API key once, reference it from a tool header, and understand why it is never shown again and what blocks a delete.
- Auth connectionsSave a basic, bearer or OAuth2 credential once, attach it to the tools that need it, and know exactly when EchoNova fetches a new token.
- The retrieval benchCheck whether an agent would reach an imported API operation for something a caller might say, and understand what the bench deliberately does not do.
Channels
- Channels overviewWhat a channel is, every channel an agent can answer on, what each one supports, and which one to launch first.
- WhatsApp prerequisitesWhat to have ready in Meta before you import a WhatsApp number, what signing in with Meta asks for, and what every connect message means.
- Connect WhatsAppImport a WhatsApp Business number and connect it to an agent from the Channels tab, send it a first message, read the reply in the portal, and disconnect it when you are done.
- WhatsApp limitsWhat the agent receives, what it sends, how conversations are grouped and what one number can be connected to, as it works on WhatsApp today.
- MessengerImport a Facebook Page and connect it to an agent so it answers the messages people send the Page on Messenger.
- InstagramImport an Instagram professional account, through the Facebook Page it is linked to, and connect it to an agent so it answers the direct messages people send it.
- SlackInstall the EchoNova app in a Slack workspace so people there can talk to an agent by direct message, by mentioning it in one channel, or both.
- SMS with TwilioWhat a connected Twilio number gives an agent on the text side, what the Send SMS action does, and what happens to a text message sent to the number.
- Voice with TwilioPut an agent on a Twilio phone number so that callers reach it, and know what the caller hears, which settings apply on the call, and what ends it.
- Genesys Cloud chatConnect a Genesys Cloud organization so an agent can answer as a bot inside your message flows, and hand the conversation back to a queue when a person is needed.
- Genesys Cloud voiceWhat the Audio Connector trigger on a Genesys Cloud connection is for, what a phone call routed through it does today, and where to put an agent on the phone instead.
- ZendeskConnect a Zendesk account so an agent answers new tickets and customer replies as public comments, and hands a ticket to your team by tagging it.
- Zoho DeskSign in with Zoho so an agent answers new tickets and incoming replies in a Zoho Desk organization as email replies, and hands a ticket to your team by tagging it.
- IntercomConnect an Intercom workspace with the credentials of your own Intercom app so an agent answers conversations as a teammate you choose, and hands a conversation to your team by tagging it.
- Human handoffWhat handing a conversation to a person means on each channel that supports it, what the customer and your team see, how to write the persona so it hands off at the right moment, and what to do on the channels with no handoff.
Integrations
Operate
- Find and read a conversationOpen the Conversations tab, narrow the list to the call or chat you want, and read what the agent and the caller said.
- Transcripts and summariesWhat a conversation records for every turn, when the summary above the transcript appears, and why a summary is sometimes missing.
- Tag a conversationCreate the workspace's conversation tags, put them on a conversation, rename them and delete them.
- Agent analyticsEvery metric the portal shows for an agent, its exact definition, and what each one leaves out.
- Test your agentOpen the playground, talk to your agent by chat or by call, and understand what the test does and does not prove.
- Agent statusThe two statuses an agent can hold, what each one allows, and exactly what archiving does and does not stop.
- What a turn costsWhat a voice and a text conversation are charged for, how AI usage becomes credits, and a worked voice turn and text turn at today's rates.
- Working hoursSet an agent's time zone and answering window on the Settings tab, and know which control stops an agent answering.