Blog How It Works Contact Get Started

Resources

Why Authava A Strategy Partner
Use Cases
Team Can’t Find Answers Visitors Ignore Forms Answering Same Questions Website Has Hidden Gaps Traffic With No Conversions Convert Missed Calls to Bookings
Case Studies
Product
Chatbot Portal Bot Phone Bot Outreach Bot Email Bot
Guides
Make Bots Accurate Design Decisions Data Safety After Go-Live Full Integration DIY vs. Managed Does This Pay For Itself? Data Privacy Data Processing MCP Gateway
Reference
Affiliate Program Integrate with your Page Chatbot Page Performance Authava and Tech
FAQ Press Kit
explainer

Full Integration

Authava connects to your systems in both directions: you can manage it through an API, it can recognize your signed-in users, and it can call your systems before, during, and after a conversation.

Most chatbots are islands. You drop them on a page, they answer some questions, and that's the end of the relationship. They don't know who the current user is, so they can't look up your data. The bot is the same for everyone. They can't talk to your systems. They can't be managed programmatically. They may start fast for a demo, but hit a wall when you need to go further.

Authava was built by senior engineers with enterprise experience. It's API-first from the ground up — nearly anything you can do through the dashboard or chat widget, you can also do through an API call. Your bot isn't a bolt-on. It's part of your stack.

The Short Version

Here's what Authava integration covers:

  • Management API — nearly anything you can do through the dashboard or chat widget, you can also do through an API call: create bots, upload documents, pull transcripts, set messages
  • Full lifecycle webhooks — hook into before, during, and after the conversation
  • MCP Gateway — the bot can call your APIs, show embedded widgets, and give exact-wording answers mid-conversation
  • User authentication — your bot can inherit your app's security context and know who's logged in
  • Host page awareness — the bot knows what page it's on and can respond accordingly. The initial message can use querystring parameters to customize it.
  • UI integration — open it via JS, trigger JS from it, or embed it in your own widget (JavaScript guide)
  • CRM integration — contacts the bot collects can sync to your CRM (like HubSpot)

For how this works with the tools you already use, such as HubSpot, Salesforce, n8n, WordPress, Google, and Microsoft 365, see Authava and Technology.

Management API

The management API means you can treat Authava like infrastructure. Nearly everything available in the UI dashboard or chat widget is also available via API call.

Practical examples:

  • Create and configure a new bot entirely via API. We could bulk-create 100 bots.
  • Upload documents programmatically — call the API from n8n or your own scripts and document ingestion becomes a pipeline, not a manual task.
  • Pull all chat messages — feed transcripts into your analytics, compliance archive, or CRM
  • Set the greeting or tooltip dynamically — "Your account balance is $213", set at runtime

Application API: Webhooks at Every Stage of the Conversation

Authava hooks into the entire conversation lifecycle — before, during, and after.

Before the Chat Starts

Before a user types anything, you can call your own API and inject context:

  • Time-aware greetings: "We're closed right now, but your chatbot is ready to help"
  • Personalized openers for authenticated users: "Hi Marcus, your current balance is $1,232"
  • Campaign or referral context based on where the user came from

During the Chat

As the conversation progresses, Authava can call your systems mid-conversation. When a user asks "give me a quote" or "what's your top-selling product this week," you want a real and accurate answer from your real system — not an LLM guessing it. The bot does this through the MCP Gateway (below), or by matching against your configured Q&A logic.

The bot also knows which page it loaded on, so the same bot can show a different greeting or tooltip on your pricing page, support page, or product page — without any extra input from the user.

After a Lead is Captured

When a user provides their name and email, that contact can be sent to your CRM, either right away as part of the conversation flow or in a regular batch sync. We set up whichever fits your process.

After the Chat Ends

Post-chat analysis, summaries, and follow-up workflows can run once a conversation is finished. A conversation counts as finished after it has been quiet for about an hour, so these run shortly after, not the instant the visitor leaves. This is especially useful when the bot is part of your larger sales funnel.

A Note on n8n

Authava doesn't try to be a workflow orchestration tool. n8n is good at that. n8n can call Authava's API, and Authava can send webhooks to n8n. If you're already running automation through n8n, it plugs in cleanly.

Security: Your Data Stays Yours

Authava's webhook model means your bot calls out to your API to fetch data at runtime. You are not uploading your customer database to Authava. Sensitive data - account balances, order history, personal records - can stay in your environment. The bot requests it, your system returns it, it surfaces in the chat. The answer the bot gives becomes part of that conversation's transcript, like any other message.

User Authentication

Authava supports single sign-on integration. The bot can recognize a visitor who is already signed in to your site or app, without the user needing to re-enter their credentials.

How it works: your site tells the bot who is signed in, either from your AWS Cognito login or through a one-click sign-in handoff. Once authenticated, the bot knows who it's talking to and can serve account-specific responses.

This enables things like:

  • "What is my account balance?" - pulled from your live data, for the actual logged-in user
  • Per-user greeting messages set at session start
  • Role-aware responses

Our managed services team works with your developers to set this up. It requires real integration work, but it's well-trodden ground for us.

Query String Integration

Authava can read query string parameters from the host page URL and use them to set the welcome message or initial context. Useful for campaigns, multi-tenant portals, or anywhere you want to pass context without a full authentication flow.

CRM Integration

When a user provides contact information through the bot, Authava can create or update a contact record in your CRM (such as HubSpot). This is set up per bot: contacts can sync as part of the conversation flow, in a regular batch (within about 15 minutes), or on demand.

UI Integration

  • The host page can open the bot via JavaScript — trigger it from any element, not just the default launcher
  • The bot can trigger JavaScript on the host page — a user asks where something is, the bot highlights it directly on the page
  • Embed as your own widget — run the bot in full-screen mode with the close button hidden, drop it into any container, and it looks like a native part of your UI rather than a popup

JavaScript integration guide: your page can open and close the bot, send messages, show a tooltip, and register functions the bot can run. See Integrate with your Page for code samples and a live sample page.

Model Context Protocol (MCP)

Authava's MCP Gateway lets your bot use your systems in the middle of a conversation: look things up, calculate, and show interactive tools, then answer with the real result. It's built on the Model Context Protocol (MCP), an open standard for connecting AI models to tools.

Each connection is a tool the bot can use:

  • Call an API — look up an order, check inventory, calculate a quote, or update a CRM record
  • Show an embedded widget — a booking calendar, order tracker, or chart right inside the chat
  • Give an exact answer — your exact wording, like a policy statement, returned word for word

Why This Is More Than "Calling an API"

A traditional integration is a fixed pipe: a form sends exact fields to an endpoint in exactly the right format. Visitors don't talk that way. The MCP Gateway sits between how people talk and how your systems expect requests:

  • It maps plain language to the parameters your API needs. Each tool describes the inputs it expects. When a visitor says "where's the blue jacket I ordered last week," the bot works out which tool fits and fills in the parameters from what the visitor meant, not from a rigid form.
  • It works across several messages. If something required is missing, the bot can ask a follow-up question, then call the tool once it has the answer. Earlier details in the conversation stay in context, so the visitor doesn't have to repeat themselves, and the next question can build on the last result ("OK, and when will it arrive?").
  • It can chain tools. A single answer can use more than one tool, with one result informing the next: look up the customer, then their order, then show the tracking widget.
  • It knows who's asking. For signed-in visitors, the visitor's identity travels with each tool call, so "show my orders" works without asking for an account number. If a tool needs an identity the visitor doesn't have, the call doesn't run, rather than guessing.
  • It keeps credentials out of the conversation. Tools can connect through accounts like HubSpot, with access tokens stored encrypted and refreshed automatically. The visitor never sees them.
  • It can return exact output. A tool can return fixed, exact wording or an embedded widget instead of letting the AI rephrase the result.

What It Looks Like

A visitor asks, "Can I still return the boots I bought?" The bot recognizes the visitor, uses one tool to find their recent orders, asks which pair they mean if there's more than one, uses another tool to check the return window, and answers with the actual date, all inside one natural conversation.

Who Sets It Up

MCP Gateway is set up by our managed services team. It isn't a Studio setting. You tell us which systems and actions the bot should have, and we configure the tools, connect the accounts, and test them with you. You can use Authava's hosted gateway, where tools are defined in configuration, or your own MCP server.

MCP Gateway is about your bot reaching out to your systems. That's separate from outside AI assistants connecting in to your bot.

Read the full MCP Gateway guide.

Frequently Asked Questions

Can Authava's chatbot call my API during a conversation?

Yes. Through the MCP Gateway, the bot can call your APIs mid-conversation and use the live result in its reply, for example an order status, a price, or an account balance.

Can I open or control the chatbot from my own website's JavaScript?

Yes. Your page can open and close the bot, send messages, and show a tooltip. See Integrate with your Page.

Does the bot know who the logged-in user is?

It can. With single sign-on set up, through your AWS Cognito login or a one-click sign-in handoff, the bot recognizes the signed-in visitor and can give account-specific answers for that user.

Can I manage bots through an API instead of the dashboard?

Yes. Nearly everything available in the dashboard or chat widget is also available through the management API, including creating bots, uploading documents, and pulling transcripts.

Do I have to upload my customer database to Authava?

No. The bot calls your API at the moment it needs data, and your records stay in your systems. The answer the bot gives is saved in that conversation's transcript, like any other message.

The Bottom Line

Authava was built for teams who need a chatbot that fits into how their business actually works.

If you're evaluating integration depth, the questions to ask are: Can I manage it via API? Does it know who my users are? Can it call my systems mid-conversation? Can I embed it on my terms?

Authava is built to support all four.