Meta Global AI Developer Hackathon. $1M in prizes.Get notified
LLAMA 3.2 Vision 90B, LLAMA 3.3 70B
Exati
39%
faster customer service handling times2,209
hours of customer support time saved over four months26%
faster onboarding of new support agents*All results are self-reported and not identifiably repeatable. Generally expected individual results will differ.
CASE STUDY
Transforming support for smart city platform clients
At a glance
Industry
SaaSUse case
Enhancing customer support with an intelligent assistantGoal
Resolve customer issues and onboard new staff more efficientlyLlama versions
Llama 3.2 Vision 90B, Llama 3.3 70BDeployment
Oracle Cloud Infrastructure (OCI)THEIR STORY
Improving quality of life with software that makes cities smart
Exati is Latin America’s leading provider of smart city operations software serving over 700 municipalities in six countries and in three languages. The Exati ecosystem helps oversee streetlights, roads, trees and waste from 10 million assets and 300,000 IoT sensors across urban operations like road maintenance, street cleaning and forestation. With Exacti, cities and their contractors can manage urban infrastructure and deliver services efficiently.THEIR GOAL
Making it faster for customer support to access information and resolve requests
The Exati smart city platform ingests vast streams of unstructured video, audio, photos and text data from IoT sensors embedded across urban environments. It transforms this raw data into actionable intelligence for city governments and contractors — supporting a wide range of roles.Exati customer support agents answer over 1,000 requests a month, many of which require them to search for information through fragmented data sources, including lengthy video recordings and dense documentation with visual elements. Support agents often need to sift through these resources during live customer calls, causing delays in case resolution and resulting in a suboptimal customer experience.
THEIR SOLUTION
A Llama-based intelligent customer support assistant
Teaming up with Oracle, Exati implemented a Llama-backed AI solution that processes information about smart cities and company software and provides live assistance to customer support agents. Support agents can ask questions in natural language and receive succinct syntheses of multimodal information — text documents, video feeds, still images, graphs and diagrams — in Spanish, Portuguese and English.
The Llama-enabled AI assistant surfaces relevant information to support agents quickly.
THEIR APPROACH
Llama provides multimodal understanding and real-time agent support
The solution uses Llama 3.2 Vision 90B and Llama 3.3 70B to process documents and extract key information. The Llama 3.2 Vision 90B model generates detailed, context-aware image descriptions to infuse responses with a more nuanced understanding of documents with illustrations. The Llama 3.3 70B model creates concise, insight-rich text summaries and reports from text and visual data sources, helping support agents get the information they need to facilitate real-time customer interactions.When a support agent submits a query, the system enhances it through prompt engineering and single-shot learning, then retrieves the most relevant information using Sentence Transformers and Facebook AI Similarity Search (FAISS). This context is combined with the original question and sent to the large language model (LLM) to generate precise, real-time responses. Visual data is handled similarly, with Llama 3.2 Vision producing descriptive embeddings that can be integrated directly into the retrieval-augmented generation (RAG) pipeline.
The Llama-enabled RAG workflow leverages multiple data sources, including visual data.
THEIR SUCCESS
Saving time and elevating trust with streamlined customer support
The solution significantly reduces the average response time for support agents and enhances the accuracy of information passed on to customers. This saves time and money while creating more satisfied users, ultimately enhancing Exati’s reputation and improving customer trust.Before implementing the AI solution, handling a support case took an average of 63 minutes — now it takes just 38, saving 25 minutes per case. Onboarding time also dropped from 12 weeks to under 9 weeks.
- 39% faster customer service handling times
- 2,209 hours of customer support time saved over four months
- 26% faster onboarding of new support agents
“Llama’s strong multilingual capabilities and multimodal performance made it a clear choice for our diverse linguistic requirements. The fact that open-source Llama was pre-integrated with our existing OCI technology stack simplified development and production.”
Ulisses Casemiro - AI Analyst, Exati
“Open-source Llama is transparent, integrates easily and can deploy across platforms. All this, plus lower operational costs, makes Llama essential for our strategy. Most importantly, we are confident that our data is not used for training, ensuring the privacy and security of our information.”
Diego Roessle - VP of Engineering, Exati
“Llama is a critical tool that is flattening the learning curve for new support agents and improving the support, service and information we provide our customers.”
Diego Roessle - VP of Engineering, Exati
Previous slide
Next slide
Start building
Explore more
ConsumerBusiness
Shopify | Llama case studiesShopify uses Llama to generate product pages, localize content, and automate support, helping developers scale workflows and save time.Read more
ConsumerTech
Scribd, INC | Llama case studiesDelivering faster, cheaper and more accurate results with a Llama-powered AI content discovery assistant.Read more
Stay up-to-date
Our latest updates delivered to your inbox
Subscribe to our newsletter to keep up with the latest AI updates, releases and more.