VistasM AI
Support Initiative V2
Sign in
Customer-specific AI · not a generic chatbot

Build the factory that creates
your customer’s support agents.

VistasM AI ingests your product’s UI, docs, source code, support tickets, and screenshots, builds reviewed knowledge, and deploys a customer-specific Support Agent for your end users. Repeatable for every new product.

See the end-user widget →
Human review gates every artifact Read-only crawl by default Per-customer corpus
Workspace
NorthStar Transit
Sample customer onboarding
1
Workspace Createddone
2
Ingestiondone
3
Knowledge Builddone
4
Review317 items in queue
5
RAG Index
6
Agent Test
7
Deploy
2,563
sessions
20,232
messages
612
screens
How it works

From product knowledge to deployed agents

Every step is observable, reviewable, and repeatable for the next customer or product.

1. Ingest
Crawl product URL (read-only), ingest docs, support tickets, source code, and screenshots.
2. Build
Generate product map, workflows, golden QA, and RAG chunks.
3. Review
Human reviewers approve, edit, or reject every artifact before it can become production knowledge.
4. Deploy
Ship the Support Agent as an embeddable widget on your customer’s website or in-app.
Safety by default

Generated drafts are never production gold.

Every model-generated artifact lands in the Knowledge Review Queue as a draft. A human reviewer must approve before it joins the approved RAG index that powers the live agents.

Read-only crawl by default — no mutation of customer data
Per-customer corpus isolation
Approval gates between draft and live
Evaluation runs + regression tracking before deploy
Review queue snapshot
Promote Sandbox Route — WorkflowNeeds review
Why can’t I promote this sandbox route?Needs review
Escalation Rule — Map/Street WrongNeeds review
Screen Inventory — Sandbox > Routes > DetailApproved
Drop-in widget
<script
  src="https://cdn.vistasm.ai/widget.js"
  data-agent-id="support-agent-demo"
  data-workspace-id="pathwise-ezrouting-demo"
  async>
</script>
Allowed domains, theme, position, and welcome message are managed in the workspace.
Embed the Support Agent

A bubble on your customer’s website. Approved knowledge inside.

A small JavaScript snippet ships the Support Agent to your customer’s public site or authenticated SaaS app. End users get help in context, with screenshots and optional screen-share.

Stop building one-off chatbots.

Build the factory that creates customer-specific support agents — once.

© 2026 VistasM AI · Prototype|Flowise is an optional runtime adapter, not the core product.