AI customer support.
Your questions, answered.

Supraway is a collaborative AI agent platform being built for customer support and customer care teams. It brings workflow observation, shared knowledge, agent building, testing and human oversight into one approach: Build, Optimize and Scale. The goal is to turn your team’s expertise into agents that follow your business rules and work alongside your people.

An AI customer support agent uses instructions, knowledge and connected tools to handle a customer request. Depending on its permissions, it can retrieve context, prepare an answer or carry out a defined action. For example, a returns agent could check an order and the returns policy, then prepare the next step or ask a person to review an exception.

A chatbot primarily provides a conversational interface. An AI agent can also work through a task using context, business rules and authorized tools. The distinction is what the system is allowed to do, rather than whether it has a chat window. Supraway’s approach combines conversational support with defined workflows, team review and clear limits on actions.

Supraway is designed for CX and customer support teams that want to build agents from their own operational knowledge. Support agents contribute real cases, team leaders define priorities, quality reviewers evaluate behavior, and operations or IT teams review connections and permissions. The shared workspace is intended to bring those perspectives together throughout the agent lifecycle.

Supraway’s intended role is to work with your existing systems of record. Your helpdesk, CRM or commerce platform remains the source of customer and order information. Supraway’s product previews show how an agent could use that context within an approved workflow. The exact connections and supported actions need to be confirmed for your setup.

The proposed process starts with one customer workflow and examples of how your team handles it. Review the actions and decisions, connect the relevant policies, then write the agent’s instructions in a shared workspace. Add tool permissions and handoff rules, test both routine cases and exceptions, and review the scope before considering a live rollout.

The Observe Agent concept captures actions together with their context: the application, the target, the interaction and its observed result. It helps a team examine how a request was handled and why a step may matter. Interpretations stay linked to evidence for review. The current platform walkthrough uses an illustrative session; it does not record your screen or generate a live workflow.

The Learning Agent is designed to turn reviewed evidence into a proposed skill: when to use it, what to check and when to involve a person. That knowledge stays connected to its source and reviewer. In the platform example, a damaged-item workflow becomes reusable guidance for a returns agent. The team reviews the proposed guidance before using it in shared instructions.

Start with approved policies, business rules, representative customer conversations and examples of successful resolutions. Include exceptions, escalation rules and information the agent must verify before acting. Supraway’s shared knowledge concept keeps the source, context and reviewer alongside that guidance. Supported file formats and ingestion methods will be clarified as the platform’s launch scope is finalized.

Supraway’s builder is designed around editable instructions, shared knowledge and team review, so operational experts can contribute directly. The preview lets you edit a sample prompt and apply a teammate’s suggestion. Connecting business systems, defining custom actions or meeting deployment requirements may still need an administrator or developer. The level of technical work depends on the workflow and its tools.

Multiplayer means building an agent with the people who understand different parts of the work. Instead of keeping instructions in one person’s chat, the team works from a common prompt, knowledge base and review process. In Supraway’s platform preview, a team leader suggests an approval rule while a CX specialist contributes a relevant customer example.

A shared prompt library is a common collection of instructions for areas such as brand voice, policies and human handoffs. Supraway’s proposed library gives the team a shared foundation to review and reuse across agents. An owner and a version make changes easier to discuss. A policy update should be reviewed and tested with the agents that rely on it.

The platform preview shows a shared prompt with teammate comments and suggested instructions that can be applied to the draft. The intended workflow moves from drafting to team review and testing before release. This gives subject-matter experts a place to explain a change and reviewers a way to assess its effect. The public demo illustrates this process; it is not a live multi-user session.

Supras are the AI teammates used to illustrate Supraway’s collaborative approach. In the platform walkthrough, Nova represents customer care, Orbit helps with agent building, and Atlas supports quality and testing. They make the roles easier to understand. Your people remain responsible for defining the work, reviewing suggestions and deciding what an agent should be allowed to do.

People define policies, contribute examples, review quality and handle cases that need judgment. Team leaders also set the scope and approval rules for a workflow. Supraway’s approach is to make that expertise part of the agent’s design and improvement process. Automating a repeated task does not remove the need for an accountable person or a clear escalation path.

Test the routine path, policy exceptions and requests that require human approval. Check whether the agent uses the right information, explains the next step and respects its permissions. Supraway’s preview includes an eligible return, an expired return window and a refund above the approval limit. These are sample scenarios; production testing should use your own reviewed cases and success criteria.

The Quality Agent concept evaluates conversations against your team’s standards, including policy accuracy, clarity, empathy and human handoffs. The aim is to make review criteria explicit and turn findings into changes the team can discuss. In the preview, a response needs review because the approval owner is unclear. A human reviewer decides how the instructions should change.

That is part of Supraway’s intended quality approach. A common framework can compare whether human and AI responses follow the policy, communicate clearly and escalate appropriately. Comparisons also need context, such as case complexity and the agent’s permitted scope. The platform currently illustrates a shared review framework; it does not publish measured human-versus-AI performance results.

The Insights Agent concept turns recurring review themes into specific improvements. For example, an unclear approval owner suggests an instruction change, a missing exception suggests new knowledge, and an updated policy suggests rerunning affected tests. The purpose is to connect observations to practical next steps in Build. Your team decides which changes to make and validates the result.

Useful measures include resolution quality, customer satisfaction, appropriate escalations and cost per resolution. Review policy adherence and incorrect actions alongside speed or automation volume. Define what counts as a resolved case before comparing results. Supraway’s approach emphasizes shared quality standards; the public preview does not provide verified performance benchmarks or guaranteed savings.

The platform’s rollout preview covers Chat & Messaging, Email and Voice. The intended approach is to choose channels around a reviewed workflow and keep the same policies and permission boundaries across them. The public walkthrough does not connect or activate any channel. Channel availability and provider requirements need to be confirmed for the scope of your deployment.

Zendesk and Shopify appear in Supraway’s integration examples. They illustrate how helpdesk context and order information could support a returns workflow: reading a ticket, checking an order and preparing an approved next step. The public examples do not connect to live accounts. Connector availability, permissions and supported operations should be confirmed with the team before planning an integration.

A Resolution Agent is designed to move a customer request toward the next approved action, such as preparing a reply or a return. A Routing Agent is designed to send a request or exception to the right person or agent with the relevant context. They serve complementary roles: completing work within scope and involving the right owner when the case needs something else.

That depends on the permissions and rules established for the workflow; it should never be assumed. Supraway’s example requires a person to approve refunds above a defined threshold. The agent prepares the request and hands it to the approval owner. The amount shown in the demo is illustrative, and the demo itself cannot issue a refund or change a real order.

Define separately what an agent can read, what it can prepare and what it can execute. Also define who can edit instructions, review changes and release a version. Supraway’s control concept uses these boundaries to scope each workflow. Production access controls, data handling and deployment requirements must be reviewed for your environment before customer systems or data are connected.

The website currently presents Supraway’s product direction through interactive previews. You can explore sample workflows, edit an example prompt and run illustrative tests. Those previews do not launch a production agent, connect customer accounts or demonstrate a live multi-user service. Contact the team through the demo page to discuss current availability and the scope of a potential first project.

Choose a repeated request with a clear policy, reliable information and a named owner. Returns, delivery questions or account guidance can be useful candidates when their scope is well defined. Prepare examples of successful resolutions and exceptions, decide when a person should step in, and agree on how to evaluate quality. Start small enough that the team can review the outcome.

The current public platform overview does not specify a finalized list of supported model providers or a customer-selectable model catalog. Model choice should be discussed alongside your workflow, language needs, data requirements, latency and cost expectations. Ask the team about the options available for your project; the FAQ will be updated as the launch offering is confirmed.

No. The public product walkthroughs use illustrative cases and sample data to explain the workflow. They do not connect to your helpdesk, access a customer account or carry out a real refund. For an initial discussion, use anonymized examples rather than sensitive customer records. Any future production use needs a separate review of access, data handling and the agreed workflow scope.

The current pricing page starts with your project’s scope rather than a published fixed price. Relevant factors include the people involved, the workflows and channels, expected volumes, and the systems an agent would use. Pricing, usage limits and any pilot terms need to be confirmed with the team. No per-ticket rate or free-trial duration is promised in the current public offering.