In-App Chat Support
that resolves, not just replies
Help users on the screen where they got stuck. IrisGPT answers from your verified knowledge and live system data, runs the workflow when the answer is an action, and hands off to your help desk with the full conversation when a human should take over.
What is in-app chat support?
Most support journeys still push users out of the product. A "Contact us" link opens an email form. A help icon opens a help center in another tab. Each step away from the product costs context: the user has to describe where they were, what they clicked, and what went wrong, and an agent has to ask follow-up questions before anyone can help.
In-app chat keeps the conversation next to the problem. With grounded AI on the front line, the common questions (how do I do this, where is my order, why can I not log in) resolve in the conversation, and the rest reach your team with the transcript already attached.
Why in-app chat beats sending users to email or a help center
How IrisAgent works inside your app
One widget, grounded in the same knowledge and procedures your support team already trusts.
Embed with one script tag
Paste the IrisAgent script into your web app. The widget appears as a launcher bubble with your logo, colors, starter prompts, and greeting. Everything after that is configured in the dashboard, not in your codebase.
Grounded answers with sources
IrisGPT answers from your knowledge base, past tickets, and procedures, and cites what it used. The Hallucination Removal Engine stops answers your content does not support. No source, no guess.
Live data instead of generic replies
When the answer depends on the user's account, a SmartOps procedure asks only for the field it needs, calls your Custom API, and replies with the real result: the order status, the plan, the reset link.
Actions, not just answers
Procedures handle the workflows behind common in-app questions, like password resets and product how-to steps, end to end. Chain APIs so one result feeds the next step.
Clean handoff to your existing help desk
Open a ticket in Zendesk, Freshdesk, or Salesforce with the transcript attached, or move to live chat in Salesforce Enhanced Chat or the IrisAgent inbox while the user stays in the same widget.
Feedback that improves the next answer
Users rate answers with thumbs up or down. Review weak answers, fix the source or revise the reply once, and the correction carries across every channel.
See it applied to real workflows: automate password resets and answer product how-to questions, both built on SmartOps procedures.
Context that carries through the whole conversation
The worst in-app experience is a bot that repeats questions and then hands the user to an agent who asks them all again. IrisAgent is built to avoid both.
- Ask only what the workflow needs. A procedure requests a field, such as an order ID, only when it needs it to call your API, then answers from the live result.
- Answer in the user's language. Set a fixed language or let IrisGPT translate dynamically, using the same knowledge base.
- Stay scoped to the right product. Product segmentation keeps retrieval focused when you support more than one product.
- Hand off with the transcript. Tickets created in Zendesk, Freshdesk, or Salesforce include the conversation. Salesforce Enhanced Chat receives the prior conversation, so the agent picks up where the AI left off.
- Keep the user in one window. With live chat through Salesforce Enhanced Chat or the IrisAgent inbox, the user keeps chatting in the same widget after the handoff.
Once a human takes over, Agent Assist drafts the reply inside the help desk from the same knowledge and procedures.
Built for resolution, not for product tours
Messaging suites like Intercom bundle in-app support with product tours, onboarding checklists, and marketing messages. IrisAgent does not try to replace those. Its job is narrower and harder: give the right answer from verified sources, run the support workflow with live data, and hand off cleanly when the AI should not be the one answering.
If your messenger already runs on Intercom, Zendesk Messaging, or Freshchat, IrisGPT can run inside it. If you want a dedicated support widget, embed the IrisAgent widget directly.
What about mobile?
The IrisAgent widget is responsive and works in mobile browsers, with its own setting for auto-opening on mobile. IrisAgent does not ship a native iOS or Android SDK. Teams whose native apps use Zendesk Messaging can run IrisGPT as the AI agent behind it.
One answer, every channel
The in-app widget runs on the same IrisGPT engine as your help center, Slack, Teams, WhatsApp, and email. Users get the same grounded answer wherever they ask, and a correction made once applies everywhere.
Set up in-app chat support in five steps
- 1
Connect your knowledge
Connect your help center, docs, and ticketing system. IrisGPT learns from articles and resolved tickets you already have.
- 2
Configure the widget
Set branding, launcher position, starter prompts, language, response style, and an optional proactive greeting. Scope answers by product if you support more than one.
- 3
Add procedures and APIs
Write SmartOps procedures for your top in-app workflows and connect the Custom APIs they need for live account data.
- 4
Choose the handoff
Pick contact capture, ticket creation in your help desk, or live chat, and decide which tags or routing attributes go with it.
- 5
Paste the script and go live
Add the script tag to your app. Start with a few pages or user segments, review conversations and feedback, then expand.
Security for chat inside a logged-in product
- SOC 2 Type II and GDPR compliant
- Data encrypted at rest and in transit
- PII redaction before data is processed
- Answers limited to the knowledge sources and APIs you connect
- Private deployment options for teams that need their own environment
Read more on the IrisAgent security page.
operations
Any questions?
We got you.
Explore related topics
Go deeper on grounded AI chat, procedures, and the workflows users ask about most.
Calculate Your Savings
See what resolving questions in-app could save your team.
Try the ROI Calculator →AI Chatbot Guide
Types, deployment, and how to measure an AI support chatbot.
Read the chatbot guide →