Rezolve

Multimodal AI in Action: Voice, Email, Search & Service Catalog

Shano K. Sam

Senior Editor

Created on:

February 27, 2026

5 min read

Last updated on:

February 27, 2026

AI & Automation

What to Expect in This Session

A live, unscripted voice AI demo resolving a password reset end-to-end — with no human agent

Real-time multilingual detection: the same AI switching to Tagalog mid-call without configuration

Email-to-service-catalog automation: a plain-language email becoming a fully populated service request

Vision AI in Teams: reading a BSOD screenshot and providing targeted troubleshooting steps

Conversational service catalog: pasting an invoice image and having the AI extract and submit the request

Creator Studio AI Flow Builder: generating a production-ready Microsoft Entra group creation workflow from a description in minutes

INTRODUCTION

Employees don't interact with IT through a single channel — and a modern AI service desk shouldn't require them to. In this session from Rezolve Connect 2026, Chris Martin (Senior Sales Leader), Josh Carrero (Solutions Architect), and Andrew Richardson (Senior Automation Engineer) demonstrated Rezolve.ai's full multimodal stack in live action across voice, email, Microsoft Teams, vision AI, and an AI-powered workflow builder.

Nothing in this session was pre-recorded or scripted. Every demo ran live — including a phone call to the AI voice attendant, a real email sent and processed, a Teams conversation with image analysis, and a workflow built from scratch and triggered through the bot before the session ended.

➜ Access all Rezolve Connect 2026 sessions on demand

What Is Multimodal AI?

In the context of enterprise IT service, multimodal AI means the ability to accept and intelligently process inputs across different formats and channels — voice calls, emails, chat messages, screenshots, document images — and take action based on what's understood from each. It is the difference between an AI that works only inside a specific interface and one that meets employees wherever they are.

Rezolve.ai's multimodal capabilities include voice (Rezolve VoiceIQ), chat, email processing, vision AI (image reading), and automated workflow generation. This session demonstrated all of them.

Voice AI: A Full Resolution Over the Phone

The session opened with a live call into Rezolve.ai's AI voice attendant. The scenario was a locked-out employee needing a password reset — one of the most common Level 1 support requests in any IT environment.

LIVE DEMO WALKTHROUGH

The caller described being locked out and needing a password reset. The AI located the account, dispatched a reset link to the registered email address, waited on the line while the caller completed multifactor authentication (demonstrated using Okta, though Entra and Duo are also supported), confirmed the password had been successfully changed, and generated a support ticket with a ticket number read back to the caller — all without any human agent involvement.

BEHIND THE DEMO: WHAT THE DASHBOARD SHOWS

The Voice AI admin dashboard surfaces call volume, average call duration, and sentiment analysis for every interaction. Supervisors can review full call transcripts and see the AI's real-time sentiment read. In this demo, the system flagged that the caller hesitated frequently, indicating potential stress or uncertainty — context that travels with the ticket and can inform how the follow-up is handled.

"The ability to detect sentiment in real time — not just resolve the issue, but understand how the employee experienced the interaction — is what makes this more than a phone tree replacement." — Chris Martin, Senior Sales Leader, Rezolve.ai

Multilingual Support: Dynamic Language Detection

Voice AI in Rezolve.ai is not limited to English. The session demonstrated real-time multilingual detection with a Tagalog-speaking caller. Without any configuration change or language selection step, the AI identified the language mid-call, switched to Tagalog, continued the conversation, asked clarifying questions to understand the issue, and proceeded toward ticket creation.

"The AI detected that the caller was speaking Tagalog and immediately switched languages. It was able to have a full interactive conversation, understand the issue, and ask clarifying questions — all leading into ticket creation." — Chris Martin, Senior Sales Leader, Rezolve.ai

This works across the full supported language set with no per-language setup required — making it particularly valuable for global organizations supporting multilingual workforces.

Email AI: From Inbox to Fully Formed Service Request

The second channel demonstrated was email. A support request for a new MacBook Air was sent to Rezolve.ai's support inbox with casual, plain-language content: hardware preference, color, and storage size mentioned in the body.

WHAT HAPPENED AUTOMATICALLY

When the email arrived in the platform, the AI parsed the content, matched it to the correct service catalog item (hardware request), identified the mandatory fields for that service type, and populated them directly from the email — including the specific MacBook Air model, preferred color, and storage size. The result was a fully formed, categorized, and subcategorized service request, ready for fulfillment, with no manual data entry.

This demonstrates the connection between the email channel and the service catalog's process definition layer. The AI doesn't just log an email as a generic request — it understands intent, maps it to a defined process, and captures the required data.

USE CASE SNAPSHOT

Any employee request that arrives via email — hardware, software, access, onboarding — can be automatically converted into a properly categorized, properly routed, properly populated service request. The service desk team receives a ticket that is already complete.

Vision AI in Microsoft Teams: Troubleshooting with Image Intelligence

The third channel was Microsoft Teams, where Rezolve.ai's virtual support agent demonstrated a capability that goes beyond keyword matching: the ability to read and reason about images.

THE SCENARIO

An employee reported that their computer was behaving abnormally. Rather than describing the error in text, they attached a screenshot of the screen showing a blue screen stop code. The AI read the image, identified the specific stop code, and provided targeted troubleshooting steps for that exact error — not a generic 'computer not working' response.

HOW THE CONVERSATION PROGRESSED

The conversation continued naturally. The employee asked how to boot in safe mode. The AI provided instructions. When the employee decided to escalate, the AI generated a complete ticket: subject, full description, image attachment, and correct domain, category, and subcategory — all mapped automatically from the conversation context.

"Vision AI is not just helpful for troubleshooting and triaging. It's also how we deliver services — reading documents, invoices, forms — and acting on what's in them." — Josh Carrero, Solutions Architect, Rezolve.ai

Conversational Service Catalog: Vision AI for Service Fulfillment

Josh Carrero demonstrated vision AI applied to a different problem: service fulfillment via document image reading.

THE SCENARIO

An employee needs an urgent invoice payment. Instead of filling out a form, they tell the bot what they need. When the bot asks for the invoice number, they paste a screenshot of the actual invoice into the Teams conversation.

WHAT THE AI DID

The AI read the image and extracted everything relevant: invoice number, vendor name (Verizon Wireless), billing address, and total amount due. The employee added a brief explanation of the urgency. The AI mapped all of that into the correct service catalog fields automatically and presented a summary for confirmation before submitting.

"I didn't have to provide any of that manually. The AI pulled the invoice number, the vendor name, the address, and the total amount directly from the image I shared. All I had to do was explain the urgency." — Josh Carrero, Solutions Architect, Rezolve.ai

The employee never typed an invoice number. They never filled out a form. They had a conversation, shared a document, and the AI handled everything else. This is what conversational service fulfillment looks like at the intersection of intent-driven ITSM and vision AI.

See how this connects to broader service catalog design: AI Help Desk Software for ITSM, HR & Enterprise Support

Creator Studio AI Flow Builder: Automations Built in Conversation

Andrew Richardson demonstrated the newest capability in Rezolve.ai's Creator Studio: an AI flow builder that generates production-ready automation workflows from plain-language descriptions.

WHAT IS THE AI FLOW BUILDER?

The AI flow builder is a conversational interface for building automations. A team member describes what they want to automate. The AI asks clarifying questions to understand the requirements — execution type, data sources, field collection, error handling — and generates a complete, working workflow: API server configuration, UI cards for user input, confirmation steps, and error handling.

Workflows that previously required weeks of API research, script writing, and platform training can now be produced in minutes by someone who has never opened Creator Studio before.

"Instead of weeks of complex configuration, users can now build, test, and deploy intelligent workflows in minutes — simply by describing what they need. The platform translates natural language into production-ready flows, handles logic and integrations automatically, and adapts in real time through conversation." — Chris Martin, Senior Sales Leader, Rezolve.ai

LIVE BUILD: CREATE A SECURITY GROUP IN MICROSOFT ENTRA

Andrew started with a blank canvas and described his intent: create a security group in Microsoft Entra. The AI asked: Should group creation require approval or execute immediately? What type of group — security or Microsoft 365? What information should be collected from the requester? Who should be set as the group owner automatically?

Each question was derived from the AI's knowledge of what the Microsoft Graph API requires and what constitutes best practice for group creation flows. Andrew answered, the AI generated the complete workflow — API server, UI cards, confirmation steps, error handling — and streamed it directly into the Creator Studio canvas.

FROM FLOW TO LIVE BOT

Once saved, Andrew added a trigger: a natural language description of when the bot should invoke the flow, plus sample queries to help the AI match user intent. Audience scoping limited who could trigger the flow. After publishing, he opened the Teams bot, typed 'I want to create a new group,' and the bot matched it to the flow, collected the group name, confirmed the details, and created the security group — live, in the Entra tenant, visible in real time.

The AI also automatically included a confirmation step before execution — something Andrew hadn't specifically asked for, but the AI included as a best practice for any action that creates a resource.

APPROVAL WORKFLOWS

The AI flow builder also supports approval workflows, where a request creates a ticket, routes to an approver via the ticketing system or a Teams notification, and only executes the downstream automation once approved. Every step is documented, creating a complete audit trail — end-to-end documentation that something happened, built entirely with AI.

The Full Multimodal Picture: One Intelligence, Every Channel

Across voice, email, Teams chat, vision AI, and conversational automation, the session demonstrated a consistent principle: employees should be able to get help the way they naturally work — and the AI should handle the work of understanding, categorizing, routing, and acting.

The channels are different. The intelligence is unified. Whether a request arrives as a Tagalog phone call, a screenshot of a blue screen error, a pasted invoice image, or a casual message about needing a new Entra group, Rezolve.ai interprets intent, maps it to the right process, and executes — with full auditability and human oversight wherever the workflow requires it.

Frequently Asked Questions

1. What is multimodal AI in the context of IT service management?

A. Multimodal AI refers to the ability to process inputs across different formats and channels — voice, text, email, images, documents — and take intelligent action based on what is understood from each. In IT service, it means employees can request help through any channel they naturally use, and the AI handles interpretation, routing, and resolution.

2. Does Rezolve.ai's Voice AI support languages other than English?

A. Yes. The voice AI detects language dynamically in real time and adapts the conversation accordingly — no per-language configuration is required. The session demonstrated this live with a Tagalog-speaking caller who had a complete, unscripted AI conversation with no setup in advance.

3. What is Vision AI and how does it work in the service catalog?

A. Vision AI is the ability for the AI to read and extract information from images — screenshots, documents, invoices, forms. In the service catalog, this means employees can share a document image in a conversation and have the AI automatically populate the relevant service request fields from what it reads, without the employee manually entering any data.

4. How does the AI Flow Builder differ from traditional workflow automation tools?

A. Traditional workflow automation requires technical knowledge of APIs, scripting languages, and platform-specific configuration. The AI Flow Builder lets teams describe what they want to automate in plain language, asks clarifying questions, and generates a production-ready workflow — including API integrations, UI cards, confirmation steps, and error handling — in minutes. Workflows that previously took weeks now take minutes.

5. Can automations built with the AI Flow Builder include approval steps?

Yes. The AI Flow Builder supports both simple flows (immediate execution) and approval workflows, where a request creates a ticket, routes to an approver, and executes only after approval. Every step is documented, creating a full audit trail.

6. How does email processing connect to the service catalog?

When a support email arrives, Rezolve.ai's AI parses the content, identifies the intent, matches it to the appropriate service catalog item, and populates the required fields automatically — including any specific details mentioned in the email body. The result is a fully formed, categorized service request with no manual data entry.

7. Where can I watch the full session?

Watch on YouTube or access all sessions at Rezolve Connect 2026 On Demand