**How is Salesforce’s Agentforce transforming customer service, and why are 90% of deployments starting with Service Cloud?** David Brown, Chief Customer Officer for Service Cloud at Salesforce, shares insights on autonomous agents, multilingual support, and the future of service teams.

In this interview from the Agentforce World Tour in Amsterdam, David Brown discusses real-world implementations, from Heathrow Airport’s WhatsApp agent to Simply Health’s strategic email automation. Learn about the dramatic channel shift happening in customer service, how **AI-assisted agents can easily take the burden off your customer service team**, and how to implement Agentforce successfully in your organization.

![Why 90% of AI agent deployments start with customer service](https://www.youtube.com/embed/HrQeM_-8Q2k)

Discover the practical strategies companies are using to achieve measurable ROI, the critical importance of trust and accuracy in AI deployments, and why service teams are evolving from cost centers to revenue generators.

## Some insights on why AI dominates customer service in Agentforce deployments

**Question:** Why is service the area where we see most AI agent implementations, with almost all customer references being service-oriented?

**David Brown:** In Salesforce, if you look at our Agentforce deployments globally, depending on what month you look at, around **90% of the go-lives of Agentforce use cases start with Service Cloud**. It’s so tangible, isn’t it? People can tend to prove the cost savings quite quickly with service use cases, so we’ve tended to sign off on way more projects and get more go-lives in the area of service than anywhere else.

**Question:** Are there specific types of service use cases that stand out in terms of adoption?

**David Brown:** Most people try an agent in a chatbot to start with, within the chat channel. Most people try it as well, with fairly basic requirements, like an FAQ knowledge article being exposed to an agent. But there are many more advanced and exciting use cases now where agents are taking real actions, completing processes, and doing more complex work.

## Moving beyond traditional chatbots

**Question:** How do modern conversational AI agents differ from the earlier chatbots that people generally disliked?

**David Brown:** I think the only thing disliked more than a chatbot is IVR (Interactive Voice Response). If you were pushed into an IVR channel, it’s probably the worst of all. But these conversational experiences you can drive now through agents, it’s a dramatic difference to a static “if yes, no, go here.” Those kind of static chatbot experiences are becoming a thing of the past and very obsolete.

**Question:** Salesforce put help.salesforce.com entirely on Agentforce. What were the results?

**David Brown:** Help.salesforce.com. I think the stat now is that **86% of cases created are resolved with an autonomous agent**. Customer satisfaction has gone up as well because they’re getting answers faster. To resolve 86% without diverting to another channel or a human is pretty amazing, and it’s driven some incredible cost savings.

## Productivity gains from basic AI capabilities

**Question:** Beyond autonomous agents, what productivity improvements are companies seeing from basic AI features?

**David Brown:** Just the most basic of AI deployments in service, like a summarization of a call, can save companies minutes after a call finishes with note-taking. The Service Reply capability we’ve got, where you just take in your knowledge base and recommend things for the service agent to say, is **improving average handling time for some customers by 45%**. So, transcripts being summarized and recommendations being prompted during a call, which are just generative AI capabilities, are really supercharging service agents today.

## Multilingual support and translation

**Question:** What’s the status of language support for Agentforce across Europe’s diverse markets?

**David Brown:** Unfortunately, the Dutch language support comes a little bit after Spanish, French, and Portuguese support. English clearly is the first language we support; we GA’d (general available) that at Dreamforce. But French, Spanish, German, and Portuguese will come in the next round of languages, and Dutch will come very soon after.

**Question:** How will translation capabilities impact the service industry?

**David Brown:** We’re excited about the translation capability and the impact that’s going to have on the service industry. With outsourced contact centers, if I can now handle a customer conversation with an agent where you can speak in any language, then the days of me having to take up my contact centers to South Africa, Manila or India have changed. I think it’s going to have a bigger impact on interpersonal communication than we now think.

## Real-world example: Heathrow Airport’s Haley

**Question:** Can you share an example of an autonomous agent already live?

**David Brown:** I flew here through Heathrow Airport, and they have an autonomous agent called Haley. She can do wayfinding, find the route to your gate, the gate number, and how to get there. Obviously, the most important use case for me was finding out what’s on the British Airways menu. She handles menus, lounges, and routes. Haley sits within WhatsApp, which is an autonomous agent powered by Agentforce.

**Question:** What impact has Haley had on Heathrow’s operations?

**David Brown:** It’s only been live for a few months, but it’s taken the number of chats that a human now has to interact with to answer the same questions down by 40%. High-volume, low-value use cases. I think they’re trying to do two things, which are to improve the experience of me as a customer coming through Heathrow and also to take away high-volume, low-value work from service agents.

## The dramatic channel shift in customer service

**Question:** What changes are you seeing in how customers prefer to communicate with service teams?

**David Brown:** We did the seventh edition of the State of Service report, which interviewed 6,500 customers. Less than 10% of customers have a channel strategy. Most customers just think it’s great to provide support via all channels; they think the more channels the better. But what we found is that the customers who are delivering the highest customer satisfaction statistics tend to focus on the channel that’s right for the use case.

**Question:** Can you give an example of this strategic channel shift?

**David Brown:** Simply Health has this strategy to move from 70% of their inbound contact being on a phone to about 10% over the next three years, with WhatsApp going from 5% to maybe 50% of interactions. I think that’s having the biggest impact. **Channel shift is one of the biggest things that’s happening now because of the advent of AI**.

## From cost center to revenue generator

**Question:** Where do you think customers are moving next after implementing service agents?

**David Brown:** I met an electronics retailer in Germany who are using Marketing Cloud with Salesforce to send campaigns out. What they used to send at the beginning of every email is “do not reply.” But what they’ve done is they’ve put their service agent, a product expert, at the front of a marketing campaign. So if you wanted to talk about features and functions, and do some more due diligence on a product, use the product knowledge base and a support agent at the front of a marketing campaign.

**Question:** How does that change the role of service teams?

**David Brown:** That is turning your service team into a lead-nurturing role. And that’s the way service teams start to become revenue generators rather than just cost centers. So I think the promise of service centers not just being about cost, but being about profit.

Salesforce's David Brown reveals why 90% of Agentforce deployments start with service, how companies achieve 86% autonomous resolution.

AI Agents Are Revolutionizing Customer Service: Why 90% of Deployments Start Here

Amazon has announced a major expansion of its Nova AI portfolio, introducing four new models and innovative services designed to help companies implement AI at scale with greater control. This includes **Nova Forge**, a pioneering service for building custom AI models, and **Nova Act**, a service for creating highly reliable AI agents.

## Key Takeaways
- **Nova 2 models** deliver industry-leading price-performance across reasoning, multimodal processing, conversational AI, code generation, and agentic tasks.
- **Nova Forge** enables companies to build their own optimized variants of Nova by infusing proprietary data early in the training process through a unique "open training" approach.
- **Nova Act** achieves breakthrough 90% reliability for browser-based UI automation workflows built by early customers.

## Nova 2 Model Family
Tens of thousands of companies are using Nova for diverse applications, such as producing high-quality content, automating multi-step tasks, and accelerating development of AI agents. The new Nova 2 model family balances speed, cost, and intelligence across multiple AI domains.

### Nova 2 Lite
**Nova 2 Lite** is a fast, cost-effective reasoning model for everyday workloads that can process text, images, and videos to generate text. Customers can adjust how much step-by-step "thinking" the model performs before responding, balancing intelligence depth with speed and cost—ideal for **customer service chatbots**, document processing, and business automation. Nova 2 Lite delivers industry-leading price performance in its intelligence class.

![Benchmark comparison chart of AI models: Nova 2 Lite, Claude, GPT, Gemini](https://assets.aboutamazon.com/dims4/default/96af8d4/2147483647/strip/true/crop/1694x2990+0+0/resize/421x743!/quality/90/?url=https%3A%2F%2Famazon-blogs-brightspot.s3.amazonaws.com%2Fee%2F02%2F87e2494b433ca1a87bf33cbde2eb%2Fnova-2-lite-bolded-compressed-final-footnote.png)

### Nova 2 Pro
**Nova 2 Pro** is Amazon's most intelligent reasoning model that can process text, images, video, and speech to generate text. It’s ideal for highly complex tasks like agentic coding, long-range planning, and sophisticated problem-solving—where the highest accuracy is essential. The model can also serve as a "teacher" for knowledge distillation—transferring its capabilities into smaller, more efficient "student" models for specific domains and use cases.

![Benchmark comparison chart for AI models: Nova 2 Pro, Claude, GPT, and Gemini](https://assets.aboutamazon.com/dims4/default/d390210/2147483647/strip/true/crop/2062x3022+0+0/resize/507x743!/quality/90/?url=https%3A%2F%2Famazon-blogs-brightspot.s3.amazonaws.com%2F82%2Fad%2F1fa86b6d4633ad828df5e2719609%2Fnova-2-pro-bolded-compressed-final-footnote.png)

Both Nova 2 Lite and Nova 2 Pro have built-in web grounding and code execution capabilities, meaning they can search the web for current information and run code directly—ensuring responses stay anchored in up-to-date facts rather than relying solely on training data.

### Nova 2 Sonic
**Nova 2 Sonic** is Amazon's speech-to-speech model that unifies text and speech understanding and generation for real-time, human-like conversational AI. It features expanded multilingual support with expressive voices, higher accuracy, and a one-million token context window for sustained interactions, while enabling seamless switching between voice and text. The model handles tasks asynchronously, letting users continue natural conversations—even switching topics—while actions like booking flights complete in the background. It also seamlessly integrates with **Amazon Connect**, telephony providers (Vonage, Twilio, AudioCodes), and conversational AI frameworks (LiveKit, Pipecat), making it ideal for **customer service applications**, AI assistants, and interactive voice experiences.

### Nova 2 Omni
**Nova 2 Omni** is a unified multimodal reasoning and generation model that can process text, images, video, and speech inputs while generating both text and images—an industry first. It handles up to 750,000 words, hours of audio, long videos, and hundred-page documents, simultaneously analyzing entire product catalogs, testimonials, brand guidelines, and video libraries at once. This eliminates the cost and complexity of connecting multiple specialized models.

Organizations like Cisco, Siemens, Sumo Logic, and Trellix are using Nova 2 models for applications ranging from agentic threat detection to video understanding and voice AI assistants.

## Nova Forge: First-of-its-kind service for building your own frontier AI models
Organizations embedding proprietary knowledge into AI applications currently face three compromises: customizing proprietary models in ways that only scratch the surface for integrating an organization’s expertise, continuing to train open-weights models without access to the original training data, which risks the model regressing on foundational capabilities like instruction following, or building from scratch at enormous expense. What organizations need is access to both frontier model capabilities and the ability to deeply integrate their expertise.

**Nova Forge** empowers organizations to build their own optimized variants of Nova—we call them “Novellas”—by blending their proprietary data with Nova’s frontier capabilities. The service pioneers "open training"—giving exclusive access to pre-trained, mid-trained, and post-trained Nova model checkpoints so customers can mix their proprietary data with Amazon Nova-curated datasets at every stage of model training. The result is a customized model that combines Nova's full knowledge and reasoning power with deep understanding of each organization's specific business. Customers can start building their own Novellas with Nova 2 Lite today.

Beyond model checkpoints and data-mixing capabilities, Nova Forge offers three additional powerful capabilities: First, the ability to train AI using your own environments, which are referred to as reinforcement learning “gyms.” These gyms are synthetic environments where the models learn from simulated scenarios that reflect their real-world use cases. Second, option to create smaller, faster models that maintain their intelligence at a lower cost—trained on AI-generated examples from larger models through a process called synthetic data-based distillation. Third, access to a responsible AI toolkit that allows them to implement safety controls.

Organizations like Booking.com, Cosine AI, Nimbus Therapeutics, Nomura Research Institute, OpenBabylon, Reddit, and Sony are building their own models with Nova Forge to better serve their unique requirements.

Once customers create their own frontier model with Nova Forge, they can deploy it on Amazon Bedrock with the same enterprise-grade security, scalability, and data privacy as all other Bedrock models. This complete solution—from building their own frontier model to production deployment—ensures organizations achieve optimal AI performance tailored to their specific business needs, with exclusive use of their model securely hosted on AWS.

## Nova Act: New AWS service for building and managing reliable AI agents for UI-based workflows
**Nova Act** is now available as a service on AWS for building and deploying highly reliable AI agents that can take actions in web browsers. Powered by a custom Nova 2 Lite model, Nova Act provides the fastest and easiest path to build and manage fleets of agents that automate browser-based tasks. Nova Act delivers 90% reliability on early customer workflows and outperforms competing models on relevant benchmarks.

Nova Act achieves breakthrough reliability by training a custom Nova 2 Lite model through reinforcement learning, running thousands of tasks on hundreds of simulated web environments. This style of training allows Nova Act to excel on UI-based workflows like updating data in a customer relationship management (CRM) system, testing website functionality, or submitting health insurance claims.

![gif of Nova](https://assets.aboutamazon.com/dims4/default/e7b12e2/2147483647/strip/true/crop/1600x900+0+0/resize/1320x743!/quality/90/?url=https%3A%2F%2Famazon-blogs-brightspot.s3.amazonaws.com%2F53%2F02%2F0323944346649b5c1b0817fc7f35%2Fgif-rl-gyms.gif)

With Nova Act, developers can start prototyping an agent in minutes with a no-code playground using natural language prompts, refine that Nova Act agent in familiar IDEs like VS Code, and then deploy to AWS. What customers build and test locally scales in production, with comprehensive management tools and monitoring through the Nova Act AWS console.

Organizations across sectors are already seeing results with Nova Act:
- Startup Sola Systems integrated Nova Act to automate hundreds of thousands of workflows per month for their clients across business-critical tasks like reconciling payments, coordinating shipments, and updating medical records.
- 1Password used Nova Act to enable users to access their logins with fewer manual steps, and works automatically with a single simple prompt that works across hundreds of different websites.
- Hertz accelerated its software delivery by 5x and eliminated its Quality Assurance (QA) bottleneck by using Nova Act to automate end-to-end testing across its rental platform—which processes millions in daily bookings—transforming what used to take weeks into hours.
- Amazon Leo eliminated its QA constraint ahead of its satellite internet launch by using Nova Act to write test scenarios in natural language that automatically execute and adapt across thousands of web and mobile test cases, reducing what previously took weeks of engineering effort to minutes while running three times faster with zero AI costs after initial runs.

## Getting started with Amazon Nova
To learn more about Amazon Nova, visit [aws.amazon.com/bedrock/nova](https://aws.amazon.com/nova/). Start building for free at [nova.amazon.com/dev](https://nova.amazon.com/ap/signin?clientContext=132-7278988-8454135&openid.return_to=https%3A%2F%2Fnova.amazon.com%2Fdev&openid.identity=http%3A%2F%2Fspecs.openid.net%2Fauth%2F2.0%2Fidentifier_select&openid.assoc_handle=amzn_agi_hyperion_us&openid.mode=checkid_setup&language=en_US&openid.claimed_id=http%3A%2F%2Fspecs.openid.net%2Fauth%2F2.0%2Fidentifier_select&openid.ns=http%3A%2F%2Fspecs.openid.net%2Fauth%2F2.0).

Amazon Nova introduces four new AI models, pioneering open training service, and AI agents enabling companies to implement AI at scale with control.

Amazon Unveils Nova 2 AI Models: Customizable AI for Businesses and 90% Reliable Agents

On a dark winter’s night, what could be more engrossing than the latest tragifarce about energy firms, guaranteed to set spines tingling?

## Act I: The Elderly Pensioner Left Without Heating

The scene is Stafford where an elderly mother lives alone in a council house fitted with a smart prepayment meter. She was persuaded to switch to **British Gas** with the promise of “preferential rates,” whereupon the meter stopped working.

Because she can no longer monitor her credit or consumption, her electricity supply keeps cutting out. And since there is no gas supply to the house, that leaves her with no heating. She can’t pay on the app because that informs her that her account is locked.

“She gets different advice every time she rings,” says her daughter. One agent wants the pensioner to climb a ladder to photograph the meter number already on its records (she’s recently returned from a hospital stay). Another tells her to use her old top-up card from her previous supplier. She does, after putting more money on it. Unsurprisingly it has no effect and she now faces clawing the money back from a supplier she no longer has an account with.

Eighteen days after the switch she can’t check what she’s paid for and she can’t keep warm.

![British Gas admitted an unexplained ‘delay’ in transferring a customer account.](https://i.guim.co.uk/img/media/daea66aab1348b0b09b7b1916b0381e42040dac6/215_0_4170_3336/master/4170.jpg?width=445&dpr=1&s=none&crop=none)

British Gas’s conscience is always tenderised by a headline. It admits an unexplained “delay” in transferring her account which left her without heating three times in one chilly week. It also admits a shortfall in customer service when she complained. It is now switching her to a credit meter, addressing the app issue and paying her goodwill for the “inconvenience.”

## Act II: The Glasgow Man Overcharged by Thousands

Up in Glasgow, a customer has spent seven years telling his supplier that there is something wrong with his bills. Ever since a new meter was fitted in 2018 they have been enormous. So enormous that he has been left with a four-figure debit several years on the trot because his ever-increasing direct debits can’t keep up. In vain, he has sent photos of his meter first to **Bulb**, then when it collapsed, **Octopus Energy**.

Both agreed there was something amiss, but since they couldn’t work out what they kept on helping themselves to his money and he kept getting into debt.

![A meter incorrectly recording energy use in cubic feet rather than cubic metres led to astronomical bills and debt.](https://i.guim.co.uk/img/media/05622dc953f3bc411881e39c7d87e2f3279a7cfb/362_0_3625_2900/master/3625.jpg?width=445&dpr=1&s=none&crop=none)

It was when he tried to switch supplier this year that the new company spotted the hitch: his meter readings had been recorded in cubic feet rather than cubic metres which means he has potentially been billed more than three times what he should have been. Octopus promised to sort it out but the only contact he has received in the weeks since has been more huge bills.

“I feel financially at the mercy of a company that really doesn’t care about me,” he laments.

Enter Guardian Money, and Octopus leaps into action. It explains that the switch from an imperial meter to a metric one in 2018 was not recorded in the energy database, so his bills continued to be calculated in cubic feet. This has now been corrected and his account has been reviewed and backdated.

The result? It was discovered he had overpaid more than **£8,000**. He’s since been refunded this plus interest calculated at 8% and compensation worth 10% – a total just short of **£12,000**.

## Act III: The Teenage Schoolgirl Threatened Over a £20 Debt

We’re in Dorset where a teenage schoolgirl is being threatened with a trashed credit rating by **Ovo**. The utilities company has decided that she owes it £20. The fact that the charges are for a rental property her family left a year ago and cover a period after they’d moved out is of no interest to Ovo. Its debt collectors inform her that she faces legal action if she doesn’t pay up.

This plot is spookily similar to a previous drama in which Ovo billed a teenage boy for a supply at an address he’d long since left.

![Debt collectors told one school girl she faces legal action for a £20 energy bill debt.](https://i.guim.co.uk/img/media/70d86f585be955b16725a7b6633ea20383d417ad/155_0_3351_2682/master/3351.jpg?width=445&dpr=1&s=none&crop=none)

Ovo won’t talk to the girl’s mother, who was the account holder during the tenancy because of GDPR. It doesn’t think it’s odd that she has a paid-up Ovo account for the family’s current home which was opened when they moved.

“How did our society get to the point where big businesses can threaten children over debts that aren’t theirs?” she asks. Ovo blames tracing agents, who match debts with creditors, who somehow got hold of the student’s details and matched them to a debt accrued between tenancies. It has now cancelled the debt, deleted her details and dispatched a box of White Company lotions in contrition.

Weird tales of meter mix-ups, incomprehensible bills, and to foment the drama, a teenager threatened with a trashed credit rating

Energy Giants' Customer Service Nightmares: Elderly Left in the Cold, Teenagers Threatened Over Debts

## Lyft's AI Revolution in Customer Support

Lyft has partnered with the **AWS Generative AI Innovation Center** to develop an innovative "intent agent" that is transforming customer service for both riders and drivers. This **agentic AI system** has achieved remarkable results, including an **87% reduction in average resolution time** for customer support calls.

![Arriving passengers wait to board Lyft vehicles at Los Angeles International Airport](https://imgproxy.divecdn.com/vof6TDbPH9eez2DE9A04uj_v-2N0lgdfiwV2C8EQT0I/g:ce/rs:fill:1200:675:1/Z3M6Ly9kaXZlc2l0ZS1zdG9yYWdlL2RpdmVpbWFnZS9HZXR0eUltYWdlcy0xMTg1OTk1NjY5LmpwZw==.webp)

*Arriving passengers wait to board Lyft vehicles at the new "LAX-it" ride-hail passenger pickup lot at Los Angeles International Airport*

## The Breakthrough: Natural Dialogue Instead of Menus

Unlike traditional support systems that force users through frustrating menu trees, Lyft's **intent agent** uses **natural dialogue** to understand and act on behalf of customers. The system can ask clarifying questions, search Lyft's database for context, and handle multi-turn conversations to accurately identify user needs.

**Key capabilities include:**
- **Disambiguating vague requests** that would confuse traditional systems
- **Accessing driver information** including ride history, account status, and previous interactions
- **Determining automatically** whether an issue can be resolved by AI or requires human expertise

## 24/7 Availability with Impressive Results

The AI agent is available **24/7 in both English and Spanish**, providing round-the-clock support that wasn't previously possible. According to Lyft's data:

- **More than half** of customer and driver support requests are now handled in **under three minutes**
- The system immediately solves **over 50% of issues** without human intervention
- Remaining cases are efficiently forwarded to specialized human agents

## Transforming the Driver Experience

Ameena Gill, VP of Safety and Customer Care at Lyft, explains the transformative impact: "Drivers are now spending **less time on the phone with customer support** and more time helping riders. This creates a positive cycle: supported drivers are more engaged and available to deliver great services."

Before implementing the AI system, Lyft faced significant challenges. The company's growth and customer support costs moved in lockstep, creating operational constraints that reduced service availability for both riders and drivers.

"Drivers were waiting for help on issues affecting their livelihood, while riders weren't getting the fast, personalized support they deserved," Gill noted.

## Strategic Implementation of Generative AI

Lyft identified customer service as the **ideal starting point** for deploying AI in its operations. As generative AI technology matured, the company recognized that **agentic AI was uniquely suited** to customer service challenges.

The current approach creates an optimal division of labor:
- **AI handles high-volume routine issues** efficiently
- **Human agents serve as true specialists** for complex inquiries requiring empathy and accountability

This combination of **conversational intelligence** and backend workflows allows Lyft to provide personalized responses while maintaining the human touch where it matters most.

Lyft and the AWS Generative AI Innovation Center built an “intent agent” that helped reduce average resolution time for customer support calls by 87%.

Customer Services Articles Published 2 days ago

Customer Services Articles Published 3 days ago