Plug InsiteChat into Freshchat and the AI takes first-line support off your agents' plate — answering FAQs from your own knowledge base in under 10 seconds, escalating only what actually needs a human.
Free on every plan · Works alongside Freshdesk · Grounded in your content, not hallucinated
Four steps. No engineer required.
In Freshchat admin, go to Settings → API Tokens and generate a new token. InsiteChat uses this to post AI replies to your live chat conversations.
In the InsiteChat dashboard, open Webhooks & Integrations → Freshchat. Enter your Freshworks domain (yourcompany.freshchat.com) and the API token from step 1.
Copy the webhook URL InsiteChat shows you, then paste it into your Freshchat webhook settings. This tells Freshchat to notify InsiteChat about new messages.
Done. The next time a visitor messages your Freshchat, InsiteChat reads the question, pulls the answer from your knowledge base, and posts a reply in under 10 seconds.
Real-world benefits from teams who switched from decision-tree bots to RAG-based AI.
InsiteChat reads every new Freshchat message and replies with an accurate answer pulled from your own docs, FAQs, and website content. No scripting, no decision trees, no maintenance burden.
Point InsiteChat at your help center, product docs, or website and it indexes the content. Answers are grounded in your material — no hallucinated features, no outdated pricing.
When a question is outside the AI's confidence or the visitor explicitly asks for a human, InsiteChat hands off with the full conversation context. Your agents don't have to say "can you repeat that?"
Keep Freshchat as the front door and Freshdesk as the backend ticketing system. InsiteChat slots in between, deflecting what's deflectable and ticketing what's not.
Teams that auto-resolve their top 20 FAQ questions typically see agent ticket volume drop by half, freeing them to handle the complex, high-value conversations.
Your median first response time drops from minutes to seconds. CSAT goes up, queue length goes down, and your on-call rotation finally sleeps through the night.
Three ways support teams use InsiteChat + Freshchat in production.
Most support queries arrive after 6pm local time. Let the AI handle FAQ deflection overnight and only wake a human when the visitor explicitly escalates.
Going from a 5-minute average first response to a 10-second AI reply is a CSAT lever your team can pull this week — no hiring, no new tools.
New agents watch the AI reply using grounded knowledge-base answers and learn your product voice by osmosis. Great for onboarding support teams.
The clear line between what the AI does and where a human takes over.
InsiteChat trains a private, per-workspace vector index. Your content is never mixed with other customers' data and is never used to train any underlying language model. Freshchat API tokens are encrypted at rest. You can delete your workspace, tokens, and indexed content at any time.
Everything you need to know about adding AI to your Freshchat.
When a visitor sends a message in Freshchat, Freshchat fires a webhook to InsiteChat. InsiteChat generates an answer using your trained knowledge base, then posts that reply back into the Freshchat conversation via the Freshchat API. The visitor sees it as a regular agent message.
InsiteChat is configured to say "let me get a human on this" and escalate — it won't hallucinate. The conversation stays in Freshchat for your human agent to pick up, with the AI's partial answer and the visitor's full message history preserved as context.
You need a Freshchat plan that supports API tokens and webhooks — on Freshworks' plans, that's Growth tier and above. InsiteChat itself works on the free plan including the free tier.
Yes. InsiteChat auto-detects the visitor's language and replies in the same language, as long as your knowledge base content or system prompt supports it. We support 95+ languages including RTL scripts like Arabic and Hebrew.
Not automatically — and that's intentional. AI training data should be curated, not scraped from live tickets. Feed it your docs, help center, and FAQs instead. You can manually add high-value past tickets to the knowledge base if they contain durable answers.
Yes. Once a human agent replies in a conversation, the AI backs off and stops auto-responding so agents aren't fighting the bot. You can also manually toggle the AI off per conversation from the InsiteChat dashboard.
Freddy is a decision-tree bot you script manually — it handles predefined flows. InsiteChat is a RAG-based AI that reads your content and generates answers to any question, including ones you didn't predict. You don't script flows; you point it at your docs.
The Freshchat integration is free on every InsiteChat plan. You pay only for InsiteChat message volume. On the Freshchat side, you pay whatever Freshworks charges for API access on your plan.
No. Once a human agent posts any message in a Freshchat conversation, InsiteChat marks that conversation as "human-handled" and auto-mutes the AI for 24 hours, refreshing the timer on every agent reply. Visitors see their support team answering, not a chatbot trying to outpace them.
Inbound visitor messages on the `conversation_user_created` and `message_create` webhook events. Status changes, agent-to-agent transfers, and internal notes do not trigger the AI — only what a customer actually sends to your team.
Yes. The InsiteChat dashboard lets you pick which Freshchat channels (e.g. "Website chat", "Mobile app", "WhatsApp via Freshchat") fire the AI. You might enable AI on the website widget but disable it on a high-touch enterprise support channel.
Each AI-generated reply counts as one message against your InsiteChat plan, same as a reply on our native widget. Freshchat counts inbound visitor messages toward whatever Freshchat's pricing model is on your plan — that's a Freshworks question, not ours.
Use Freshchat's test channel feature: connect a sandbox channel to InsiteChat, send messages from a tester, validate replies. Once you're happy, swap the production channel webhook to the InsiteChat URL. Most teams ship in under an hour from first connection.
Multi-turn. The full Freshchat conversation history is included in each AI request as context, so follow-up questions ("does that price include GST?") are answered correctly. The 24-hour mute on agent intervention is per-conversation, so even after a human picks up, the AI is silent until the conversation reopens or 24h passes since the last agent message.
Free on every plan. Set up in under 10 minutes. Agents thank you within the week.
Get started free