01
Add your knowledge
Upload PDFs, DOCX, CSV or JSON, paste FAQs and policies, or point the crawler at your help center URL.
AI chat agent
An AI chatbot for customer service that
answers from your docs.
Upload your help center, product docs and policies. Softsasi Agent turns them into a branded chat widget that resolves repetitive questions, captures orders and leads, and escalates the rest to your team — on your website, WhatsApp and Messenger.
Try the chat agent right here
This is a live agent grounded in Softsasi's own knowledge base — ask it anything and see how your customers would experience it.
softsasi.chat
Hi — I read everything about Softsasi. Ask me anything.
Or see the pricing page
What the chat agent does
- Answers from your material only
- Every reply is retrieved from documents, FAQs, policies and tables you uploaded — so the agent can't invent a refund window you never published.
- Knows when to stop
- If your knowledge doesn't cover the question, the agent says so, collects the visitor's name and contact, and escalates the thread to a person.
- Collects structured data
- Orders, complaints, tracking lookups and leads are written into admin sheets you can search, filter and export as CSV.
- Lives where your customers are
- One agent serves the web chat widget, a full-page chat link, WhatsApp and Facebook Messenger — all threads land in the same inbox.
From documents to live chat
02
Set the personality
Name it, pick a brand colour, write the greeting, choose the languages and decide which tools it may use.
03
Publish and embed
Copy one script tag into your site. The floating chat launcher appears bottom-right, above every other element.
04
Watch the inbox
Read transcripts, take over any conversation as a human, review CSAT ratings and export the data you collected.
Embed it in 60 seconds
Paste one line before the closing </body> tag. The launcher inherits your brand colour and position from the studio.
<script async src="https://agent.softsasi.com/api/public/widget/your-agent.js"></script>
Prefer a link? Every published agent also gets a hosted full-page chat you can share in an email or QR code. Curious how retrieval works under the hood? Read about the RAG pipeline.
Questions teams ask first
How is this different from a scripted chatbot?
There are no decision trees to maintain. The agent reads your documents at question time and writes the answer, so updating a policy document updates every future reply.
Will it make things up?
It is constrained to the passages retrieved from your knowledge base. When nothing relevant is found it hands the conversation to a human instead of guessing.
How long does setup take?
Most teams publish a working agent in under an hour: one document, a greeting, and the embed snippet.
Can a human take over a live chat?
Yes. Any conversation in the inbox can be answered by a teammate, and the visitor sees the reply in the same chat window. The agent stays quiet after a handover.
Which channels are supported?
The embeddable web widget, a hosted full-page chat, WhatsApp via Twilio, and Facebook Messenger.
What does it cost to try?
The trial plan is $1 per month with one agent and 100 conversations. No card gymnastics, cancel any time.
Put it in front of customers today
One document is enough to start. Publish, embed, and let the agent take the first shift.