Features

Every capability, by tier — and what ships where

The three tiers of the platform, USP by USP, with the availability of each on Point of Contact — Hosted and Point of Contact — Self-Hosted.

One Teams-native chat platform for AI and human agents across every domain: Foundry agents you own, handoff to Teams users and back, registered users, secure file spaces, Hosted or Self-Hosted.

Availability, capability by capability

Which capabilities ship on which product today
Capability Point of Contact — Hosted Point of Contact — Self-Hosted
Multi-domain widgets Included with Point of Contact — Hosted Coming to Self-Hosted — today: One widget today
Your Foundry, your tokens Included with Point of Contact — Hosted (Your agents, your tokens) Self-Hosted today; Multi-agent registry and model choice Coming to Self-Hosted
Multi-directional handoff Included with Point of Contact — Hosted Coming to Self-Hosted — today: Take over and close in Teams
Hours, presence, fallback Included with Point of Contact — Hosted Coming to Self-Hosted
Registration via your Entra External ID Included with Point of Contact — Hosted Coming to Self-Hosted
Secure file spaces Included with Point of Contact — Hosted Coming to Self-Hosted
Topic Analysis Included with Point of Contact — Hosted Coming to Self-Hosted
Plain-language Assistant Included with Point of Contact — Hosted Coming to Self-Hosted
One Teams tab Included with Point of Contact — Hosted Teams tab today; transfer, files, analysis and alerts Coming to Self-Hosted
Signed updates you apply — (we push updates) Self-Hosted today
Your Azure at cost — (we run the infrastructure) Self-Hosted today

Tier 1 — Widgets

One platform, a widget for every domain

Sales site, support portal, partner app — each gets a widget with its own look, greeting, allowed origins, agent routing, sign-in mode, hours and file rules. Agents see every domain in the same Teams queue; the closed-chat report shows the source domain. Add or change a widget from the dashboard, no redeploy.

Self-Hosted today

Today the Self-Hosted runtime ships one website widget (Growth+) on the origins you set at deploy time, answered by the same Teams tab.

  • Each widget carries a list of allowed origins, and that list drives the Content-Security-Policy frame-ancestors header on the widget host, so a browser refuses to render the widget on a domain you did not list.
  • The widget is mobile-first: it respects the host page's safe areas, follows the on-screen keyboard as it opens and closes, and behaves with the browser back button on Android.
One platform, a widget for every domainFour sites and apps each embed their own Point of Contact widget with its own look, allowed origins, agent and sign-in mode. Every widget lands in one platform, one Teams queue with groups and routing, and the Foundry agents you own. The closed-chat report shows the source domain.YOUR MICROSOFT TEAMSOne queuefor every domainGroups and routingset per widgetB2C sales siteRetail support siteB2B partner portalInternal helpdesk appWidget - salesown look, origins,agent, sign-in modeWidget - retailown look, origins,agent, sign-in modeWidget - partnersown look, origins,agent, sign-in modeWidget - helpdeskown look, origins,agent, sign-in modePoint of Contactone platform, oneTeams tabFoundry agentsyours, in Azure AIFoundryClosed-chat reportwith the source domain
Included with Point of Contact — Hosted

Tier 2 — Flexible AI & human interaction

Your Foundry. Your tokens. Our platform.

Self-Hosted creates an Azure AI Foundry resource and project in your subscription with a gpt-5-mini deployment — or runs on a Foundry project you already own. The assistant's system prompt is yours to edit (Growth+); inference is billed by Microsoft to your subscription; the cost dashboard shows token usage and its estimated cost, with a projection on Professional. We never run the model.

Included with Point of Contact — Hosted

Register one or more Foundry agents from your project, pick who greets first, route to specialists, and let a receiving agent read a summary of what was said so far before it continues. Full cost control: inference is a line on your Microsoft bill at Microsoft's rates; change model or agent in Foundry without asking us. We publish no inference estimate because the choice is yours.

  • Model choice, a registry of several Foundry agents with routing, and a summary handed to the receiving agent.
  • No cap on AI conversations and no per-resolution fee on any plan; your single variable cost is inference on your Foundry, billed by Microsoft at Microsoft's rates.
  • The AI never trains on your data, on any product: inference runs on Azure AI Foundry you own, under your own Azure agreement, and we never run the model.
Your Azure AI Foundry, your tokens, our platformA visitor chats through Point of Contact. The platform sends the conversation to Azure AI Foundry in your own Azure - your project and your model deployment - and shows the answers. Microsoft bills the inference tokens to you. Your team in Teams takes the handoff.asksanswerstokens billed to youhandoffYOUR AZUREVisitorson your websitePoint of Contactwidget, Teams tab, handoffAzure AI Foundryyour project, your modeldeploymentMicrosoftbills your inferenceYour team in Teamsclaim, reply, close
Included with Point of Contact — Hosted

Tier 2 — Flexible AI & human interaction

AI first or human first. Hand off as often as it takes.

Handoff runs in every direction, any number of times: "Chat with a Human" or your agent brings a person in; claim, transfer to a colleague or group, take over, hand back to AI, reopen a closed chat. Hours or live Teams presence decide when a human can be offered — the AI is never gated — and presence filters who is notified. When nobody is available, the widget shows your fallback contact.

Self-Hosted today

Today on Self-Hosted the AI answers first; your team sees the live chats in the Teams tab, takes over a chat the AI is answering and closes it there.

  • An unanswered request for a person returns to the AI after 5 minutes, an unaccepted transfer returns to the original agent after 2, and a chat with no activity closes itself after 10 minutes.
Hours, presence, and always another point of contact

Availability has three layers. Business hours — per widget or tenant-wide — decide when a human can be offered; the AI is never gated by them. Each agent's live Teams presence filters who is notified about a new chat or a human request. When nobody is available, the widget shows the fallback contact you configure — a phone number, an email, a booking link.

Multi-handoff between the AI agent and your team in TeamsA visitor chats with the Foundry agent. When the visitor taps Chat with a Human (or the agent calls for a person), the business-hours schedule - or live presence when no schedule is set - decides whether a Teams user takes the chat or the visitor gets a fallback contact. The Teams user can transfer to a colleague or group, hand back to the AI with a summary, close, and reopen.Chat with a Humanavailableotherwiseclosereopentransferhand back with summaryYOUR MICROSOFT TEAMSAVAILABILITYVisitoron your websiteFoundry agentyour Azure AI FoundryBusiness hoursor live presenceFallback contactwhen no one is availableTeams userclaims and repliesColleague or grouptakes the transferClosedreopen when needed
Included with Point of Contact — Hosted

Tier 3 — External users & files

Anonymous by default. Registered when it matters.

Per widget you choose anonymous, optional or required sign-in. Registration and profile happen in the widget, against an External ID tenant you own; the service validates a real OIDC token. Signed-in users open their Past conversations, download a PDF transcript and unlock the file space. Admins do not manage external users in the app: the directory is yours; the widget is the front door.

Files where necessary — a secure Azure space, for review

A per-user file space with quotas: draft, submit for review, request deletion — submitted files are kept in a 28-day backup — with per-agent permissions deciding who sees what, a short-lived link issued per read, and files shown as chat bubbles in the thread. Anonymous visitors cannot upload, by design. Files are reviewed by people; the AI never reads them.

  • 15 MB per file by default, 100 files and 500 MB per registered user, with the allowed file types set per widget; submitted files are kept in a 28-day backup.
  • Where a registered user can upload a file, it goes to a private Azure Blob space for your team to review. The AI never reads it.
Sign-in through your Entra External ID tenant and file review by your teamA visitor signs in from the chat widget through the customer's own Entra External ID tenant. Point of Contact validates the token, returns the profile and past conversations, keeps drafts and submitted files in Azure Storage and notifies the team in Teams, who review the files. Files are reviewed by people, not read by the AI.Your teamin TeamsAzure StoragePoint of ContactYour EntraExternal ID tenantChat widgetVisitorAnonymous by default.Sign-in optional orrequired per widgetValidates issuer,audience and tenantFiles are reviewed by people,not read by the AISign in (popup)Register or sign inID tokenTokenProfile and past conversationsUpload draftDraft in the user's spaceSubmit for reviewBackup snapshot, lockActivity-feed bellReview in the Teams tabRequest deletion (optional)
Included with Point of Contact — Hosted

Tier 3 — External users & files

See what people actually contact you about

Topic Analysis runs across your closed chats for the range you pick, groups them by topic, shows the domains behind each, and writes per-topic recommendations for your human agents and your AI agent. The closed-chat report adds source domain, agents, transfers and handling time. For continuous improvement: no sentiment scores, no trend charts, no per-user profiles.

Self-Hosted today

The cost dashboard shows token usage and its estimated cost; the median projection is a Professional entitlement, and infrastructure spend stays in Azure Cost Management.

  • Filter by date, agent and widget domain; each run classifies up to 200 new closed chats and reuses the classifications it already has, so a large backlog is worked through in predictable, priced steps.
  • A read-only view of closed conversations, filtered by date, agent type, specific agent and source domain: chats resolved, time to resolution, human handling time, transfers per chat and who closed each one.
Topic Analysis over closed chatsClosed chats filtered by date, agent, widget or domain are classified by your own Foundry agent. Topic Analysis groups them into topics with a resolution rate and keywords, and turns them into coaching for your team, instruction changes for your AI agents and a per-domain view. Act on it, then run it again.Closed chatsby date, agent, widget or domainYour Foundry agentclassifies each chatTopic Analysistopics, resolution rate, keywordsCoaching for your teamcommon resolutionsInstruction changesfor your AI agentsPer-domain viewwhich domains drive each topicContinuous improvementact, then run it again
Included with Point of Contact — Hosted

Foundation — Teams and Azure

Agents and admins, one Teams tab

No second console. Human agents claim, transfer, take over, reopen and close chats in a Teams personal tab; admins manage widgets, Foundry agents, groups, hours, sign-in, files and analysis in the same tab. Sign-in is Teams SSO; new chats, human requests, transfers and file submissions arrive as Teams activity-feed alerts, filtered by each agent's live presence.

Self-Hosted today

Today the Self-Hosted Teams tab lists the live chats, takes over a chat the AI is answering, replies and closes it — with Teams SSO — from an app package built for your tenant.

  • The Settings tab has an Assistant. Say what you want in plain language; it reads the current state, applies the change through the same endpoints the forms use — same validation, your admin permissions, your Azure visibility — and shows every call it made. Nothing it does could not be done in the forms; it is faster.
  • Widgets, Foundry agents, groups, hours, sign-in, files and notifications are settings in the Teams tab — changed live, no redeploy — or ask the Assistant in plain language.

What you get, and what you need

What you get — Point of Contact — Hosted

A Teams personal tab for agents and admins (Teams SSO, activity-feed alerts); a website widget per domain through one embed tag; Azure AI Foundry agents registered from your own project; multi-directional handoff with business hours, presence and a fallback contact; visitor sign-in through your Entra External ID tenant with history and PDF transcripts; per-user file spaces for review; Topic Analysis and the closed-chat report; the Assistant for plain-language configuration.

What you get — Point of Contact — Self-Hosted

A managed application in your subscription — Function App, Cosmos DB, SignalR, Key Vault, a storage account and an Azure AI Foundry resource and project with a gpt-5-mini deployment (or a project you connect); a Teams tab where agents list, claim, reply to and close chats; one website widget on the origins you set (Growth+); an editable system prompt (Growth+); seat metering, a cost dashboard, signed customer-initiated updates and just-in-time publisher access. The full platform is coming to Self-Hosted.

What you need — Point of Contact — Hosted

Microsoft 365 with Teams (agents sign in with Teams SSO); an Azure subscription holding the Azure AI Foundry project you bring — inference is billed to it by Microsoft; an account that can grant admin consent in your Entra directory; a website or app to carry the widget.

What you need — Point of Contact — Self-Hosted

Microsoft 365 with Teams; an Azure subscription you can deploy a managed application into; Microsoft Entra ID P2 for the one to five people who will approve just-in-time publisher access; optionally an existing Azure AI Foundry project to connect instead of the one the deployment creates.

What you need — Private plan

Microsoft 365 with Teams; an Azure subscription you can deploy a managed application into; optionally an existing Azure AI Foundry project to connect. No Entra ID P2 is needed: the Private plan has no publisher access to approve.

Book a 20-minute demo Read the docs Compare hosting

Every domain. Every agent. One point of contact.

Hosted — we operate it. Self-Hosted — isolated in your subscription. Same Teams tab, same widget, Azure AI Foundry you own.

Book a 20-minute demo Read the docs

Coming to the Azure Marketplace · Microsoft Teams + Azure required · Two ways to run it