Publicado el Deja un comentario

La IA en el trabajo: cómo la IA nos ayudará a sortear la incertidumbre

enero 24, 2025

La IA en el trabajo: cómo la IA nos ayudará a sortear la incertidumbre

Ilustración de un telescopio

Por Jared Spataro, CMO de IA en el Trabajo de Microsoft

No hace mucho tiempo, la gente se maravillaba de la capacidad de la IA para, por ejemplo, escribir un soneto de Shakespeare sobre los Seattle Seahawks. Este año han ido más allá de los trucos de fiesta, ya que la IA se ha convertido en una realidad cotidiana en el trabajo, mezclándose con los hábitos de los empleados y aportando un valor real. A medida que concluye 2024, he reflexionado sobre la rapidez con la que se ha afianzado este cambio. El impulso ha sido notable, y en 2025 la IA será esencial para todos los procesos empresariales, estrategias e innovaciones.

Esta es mi opinión sobre los desarrollos de IA más innovadores de 2024, y lo que podemos esperar en 2025, ya que la IA marca el comienzo de nuevas oportunidades en un contexto de incertidumbre geopolítica y económica.

2024 fue el año que aumentó las expectativas sobre lo que puede hacer la IA, en especial para las empresas. Lo más destacado:

El razonamiento avanzado abrió nuevas y asombrosas posibilidades.  
Los grandes modelos de lenguaje, a pesar de todas sus capacidades, han tenido durante mucho tiempo un talón de Aquiles: se quedan cortos a la hora de responder a preguntas que requieren un razonamiento complejo, en especial en matemáticas y ciencias. Pero con el modelo o1 de OpenAI, dimos un gran paso adelante con sus refinadas capacidades de «pensamiento» y su capacidad para manejar tareas complejas con mayor precisión. 

Para los líderes empresariales, o1 crea nuevas oportunidades para aplicar la IA para una resolución de problemas más precisa y sofisticada. Una empresa minorista puede utilizarlo para predecir el comportamiento del consumidor y optimizar la gestion del inventario. Las instituciones financieras pueden aprovecharlo para realizar cálculos complejos, evaluar riesgos y pronosticar las tendencias del mercado. OpenAI lanzó la versión completa de o1 este mes, pero la compañía ya ha insinuado avances en el próximo modelo de o2.

La elección de varios modelos marcó el comienzo de una nueva flexibilidad.  
Este año trajo una ola de innovación en modelos de IA, incluidos desarrollos en sistemas lingüísticos pequeños y grandes. Los usuarios avanzados quieren tener la capacidad de alternar entre modelos, por lo que GitHub Copilot comenzó a ofrecer acceso a Claude 3.5 Sonnet de Anthropic, Gemini 1.5 Pro de Google y o1 y o1-mini de OpenAI. Dado que cada uno de los modelos tiene sus propios puntos fuertes, los desarrolladores tienen entre manos la elección para la tarea en cuestión. Un modelo puede sobresalir en codificación compleja, mientras que otro puede procesar código, imágenes, audio, video y texto de manera simultánea. Un modelo más pequeño y menos intensivo en recursos podría ser adecuado para tareas más sencillas. Este nuevo capítulo de flexibilidad y agilidad conducirá a casos de uso más refinados y a la resolución de problemas con IA. 

Los agentes de IA ofrecieron un nuevo y poderoso giro en la IA para las empresas.
Si bien la mayoría de las personas utilizan la IA para mejorar su productividad personal, muchos han empezado a explorar cómo  los agentes de IA pueden ejecutar procesos empresariales. Los agentes pueden pensar o razonar, recordar, recibir formación e incluso saber cuándo pedir ayuda. Un agente, por ejemplo, puede encargarse del cumplimiento de los pedidos de principio a fin: tomar el pedido, procesarlo, hacer sustituciones inteligentes de los artículos agotados y, a continuación, entregarlo para su envío. Los agentes ganaron atención e impulso en 2024, y son la próxima frontera para la IA en el trabajo (más sobre esto a continuación).   

Ahora miremos hacia el futuro. En 2025, predigo que la IA será más intuitiva e inventiva, ya que ayudará a las empresas a enfrentarse a nuevos retos en tiempos de incertidumbre. A continuación, te explicamos cómo hacerlo:

La IA ayudará a las empresas a navegar por la incertidumbre geopolítica y económica.  
Todos los ejecutivos de alto nivel con los que hablo se han comenzado a preparar para un período de inestabilidad política y económica global. Su enfoque: cómo posicionar mejor a sus organizaciones para enfrentar un período de incertidumbre. La buena noticia para ellos es que nunca ha habido una herramienta más preparada para enfrentar el momento que la IA.  

En este momento, por ejemplo, las empresas de todo el mundo sopesan cómo los posibles nuevos aranceles estadounidenses podrían sacudir sus cadenas de suministro. La IA puede ser bastante útil para ejecutar escenarios hipotéticos. Puede analizar los datos de la cadena de suministro e identificar proveedores y rutas alternativas, al tiempo que ayuda a las empresas a asegurarse de que cumplen con las nuevas regulaciones. La IA es la herramienta a la que las empresas recurrirán primero a medida que se adapten a las nuevas realidades.

La transformación de los procesos de negocio será clave para desbloquear el valor empresarial de la IA.
Si 2024 fue el año que demostró las ganancias de productividad personal de la IA (algunos usuarios ahorran más de 10 horas cada mes con Copilot, según nuestra investigación), 2025 se tratará de convertir esas ganancias en dólares y centavos. Y eso no va a venir a través de minutos u horas individuales ahorradas. Vendrá de la transformación de los procesos de negocio, es decir, de una mejora notable a procesos repetibles y estandarizables.  

¿La clave para llegar allí? Agentes. Los agentes cambiarán la forma en que gestionamos los procesos de principio a fin, desde las ventas hasta las cadenas de suministro. Anticipo que cada organización pronto tendrá una constelación de agentes, donde cada uno trabaja en segundo plano las 24 horas del día, los 7 días de la semana para ejecutar o automatizar procesos comerciales completos.

La empresa de ciencia de materiales Dow se ha mostrado prometedora en este frente. Los empleados utilizaron agentes para analizar miles de facturas de envíos en América del Norte, para señalar anomalías y destacando los posibles ahorros. Una vez que el sistema esté escalado por completo a nivel mundial, Dow anticipa una mayor precisión en las tarifas logísticas y la facturación, lo que en el primer año debería resultar en que la empresa ahorre millones de dólares en costos de envío. Este tipo de transformaciones de procesos se extenderán a las empresas globales en el próximo año.  

Veremos las primeras chispas de una nueva ola de invención.
Una explosión de organizaciones nativas de IA ofrecerá nuevos bienes y servicios innovadores, para alcanzar una gran relevancia y una gran escala. Estas ingeniosas startups incorporan la IA en todos los aspectos de su negocio, desde los procesos internos y la forma en que interactúan con los clientes hasta los productos que venden. Los nativos de IA han comenzado a aparecer en casi todas las industrias, en campos tan variados como la publicidad y la atención médica. Provocan nuevos niveles de competencia y siempre están en la búsqueda de cómo la IA puede resolver desafíos o crear nuevas posibilidades. Eso es algo de lo que todas las empresas (incluso las grandes y de larga data) deberían aprender.  

En los próximos meses y años, la IA nos llevará a nuevos descubrimientos de todo tipo. Su capacidad para analizar grandes cantidades de datos, identificar patrones y descubrir información significa que puede señalar soluciones y oportunidades que antes eran inimaginables. En la atención médica, por ejemplo, la IA puede analizar los registros médicos para predecir los resultados de los pacientes y sugerir tratamientos personalizados, lo que mejora la atención y reduce los costos. En el comercio minorista, puede analizar el comportamiento de los clientes para optimizar el inventario y personalizar las experiencias de compra. Y en la fabricación, la IA puede optimizar los procesos de producción mediante la predicción de fallos de equipos y la programación del mantenimiento para reducir el tiempo de inactividad.

La IA impulsará los avances al ayudarnos a crear materiales más resistentes, desarrollar soluciones más sostenibles y mejorar los productos de consumo, descubrimientos revolucionarios que se traducirán en resultados valiosos para las empresas de todos los sectores.

Para obtener más información sobre la IA y el futuro del trabajo, suscríbanse a este boletín.

The post La IA en el trabajo: cómo la IA nos ayudará a sortear la incertidumbre appeared first on Source LATAM.

 

​The post La IA en el trabajo: cómo la IA nos ayudará a sortear la incertidumbre appeared first on Source LATAM.  

Publicado el Deja un comentario

AWS Elastic Beanstalk improves scaling and deployment speeds for Windows instance with EC2 Fast-launch

With AWS Elastic Beanstalk you can easily deploy and manage applications in AWS without worrying about the infrastructure that runs those applications. Now, Elastic Beanstalk automatically launches Windows instances with EC2 Fast Launch enabled on currently supported Windows platform versions. This new functionality helps in timely provisioning of Windows instances at scale reducing downtime and improving operational costs during deployment, saving time for development and operational teams.

Elastic Beanstalk support for EC2 Fast Launch for Windows Instances is generally available in commercial regions where Elastic Beanstalk is available including the AWS GovCloud (US) Regions. For a complete list of regions and service offerings, see AWS Regions.

For more information about Windows Fast Launch support please read documentation. To learn more about Elastic Beanstalk, visit the Elastic Beanstalk product page.

 

​With AWS Elastic Beanstalk you can easily deploy and manage applications in AWS without worrying about the infrastructure that runs those applications. Now, Elastic Beanstalk automatically launches Windows instances with EC2 Fast Launch enabled on currently supported Windows platform versions. This new functionality helps in timely provisioning of Windows instances at scale reducing downtime and improving operational costs during deployment, saving time for development and operational teams. Elastic Beanstalk support for EC2 Fast Launch for Windows Instances is generally available in commercial regions where Elastic Beanstalk is available including the AWS GovCloud (US) Regions. For a complete list of regions and service offerings, see AWS Regions. For more information about Windows Fast Launch support please read documentation. To learn more about Elastic Beanstalk, visit the Elastic Beanstalk product page.  

Publicado el Deja un comentario

Luma AI’s Ray2 visual AI model Amazon Bedrock

Luma AI’s new video-generating AI foundation model (FM), Ray2, is now available in Amazon Bedrock. AWS is the first and only cloud provider to offer fully managed models from Luma AI, expanding creative possibilities for developers and businesses using AWS services.

Luma Ray2 is a large-scale video-generation model capable of creating realistic visuals with fluid, natural movement. With Luma Ray2 in Amazon Bedrock, you can generate production-ready video clips with seamless animations, ultrarealistic details, and logical event sequences with natural language prompts, removing the need for technical prompt engineering. Ray2 currently supports 5- and 9-second video generations with 540p and 720p resolution, making the model ideal for a variety of professional creative applications. Streamline the creative process from concept to execution by using Ray2 video generations for content creation, entertainment, advertising, and media use cases. Content creators can swiftly generate video clips for product promotion and storytelling. Product teams can rapidly visualize concepts and create video mockups for market testing. Production studios can create previsualizations, generate realistic backgrounds, and produce initial versions of special effects sequences.

Luma AI’s Ray2 model is now available in the US West (Oregon) AWS Region. To learn more about Ray2 and how to use it in your projects, read the AWS News Blog, visit the Luma AI in Amazon Bedrock page, the Amazon Bedrock console, or check out the Amazon Bedrock documentation.

 

​Luma AI’s new video-generating AI foundation model (FM), Ray2, is now available in Amazon Bedrock. AWS is the first and only cloud provider to offer fully managed models from Luma AI, expanding creative possibilities for developers and businesses using AWS services.
Luma Ray2 is a large-scale video-generation model capable of creating realistic visuals with fluid, natural movement. With Luma Ray2 in Amazon Bedrock, you can generate production-ready video clips with seamless animations, ultrarealistic details, and logical event sequences with natural language prompts, removing the need for technical prompt engineering. Ray2 currently supports 5- and 9-second video generations with 540p and 720p resolution, making the model ideal for a variety of professional creative applications. Streamline the creative process from concept to execution by using Ray2 video generations for content creation, entertainment, advertising, and media use cases. Content creators can swiftly generate video clips for product promotion and storytelling. Product teams can rapidly visualize concepts and create video mockups for market testing. Production studios can create previsualizations, generate realistic backgrounds, and produce initial versions of special effects sequences.
Luma AI’s Ray2 model is now available in the US West (Oregon) AWS Region. To learn more about Ray2 and how to use it in your projects, read the AWS News Blog, visit the Luma AI in Amazon Bedrock page, the Amazon Bedrock console, or check out the Amazon Bedrock documentation.  

Publicado el Deja un comentario

Amazon Bedrock Flows announces preview of multi-turn conversation support

Amazon Bedrock Flows enables you to link foundation models (FMs), Amazon Bedrock Prompts, Amazon Bedrock Agents, Amazon Bedrock Knowledge Bases, Amazon Bedrock Guardrails and other AWS services together to build and scale pre-defined generative AI workflows. Today, we announce preview of multi-turn conversation support for agent nodes in Flows. This feature enables dynamic, back-and-forth conversations between users and flows, similar to a natural dialogue.

Customers who use agents to execute steps within flows have indicated that sometimes the agents don’t have all the context required to successfully complete the action. With this preview launch, an agent node can intelligently pause the flow’s execution and request user- specific information. After the user sends the requested information, the flow seamlessly resumes the execution with the additional inputs. This feature enables a more interactive and context-aware experience, because the flow can adapt its behavior based on user responses.

The preview of multi-turn conversation support in Flows is now available in all regions that Flows is available. To get started, see this blog and the AWS user guide.

 

​Amazon Bedrock Flows enables you to link foundation models (FMs), Amazon Bedrock Prompts, Amazon Bedrock Agents, Amazon Bedrock Knowledge Bases, Amazon Bedrock Guardrails and other AWS services together to build and scale pre-defined generative AI workflows. Today, we announce preview of multi-turn conversation support for agent nodes in Flows. This feature enables dynamic, back-and-forth conversations between users and flows, similar to a natural dialogue. Customers who use agents to execute steps within flows have indicated that sometimes the agents don’t have all the context required to successfully complete the action. With this preview launch, an agent node can intelligently pause the flow’s execution and request user- specific information. After the user sends the requested information, the flow seamlessly resumes the execution with the additional inputs. This feature enables a more interactive and context-aware experience, because the flow can adapt its behavior based on user responses. The preview of multi-turn conversation support in Flows is now available in all regions that Flows is available. To get started, see this blog and the AWS user guide.  

Publicado el Deja un comentario

Amazon CloudWatch allows alarming on data up to 7 days old

Amazon CloudWatch now lets customers evaluate metrics data for an extended duration of up to 7 days, a 7x increase from the previous limit of 24 hours.

This enhancement empowers customers to monitor the health of longer-running or infrequent processes, such as daily data loading jobs or day-over-day performance trends, offering deeper insights into their resources and applications. Customers now have the flexibility to leverage CloudWatch alarms for use cases like near-real-time metric tracking as well as for monitoring patterns that span multiple days.

Amazon CloudWatch alarming on multi-day data is now available in all AWS Regions including the AWS GovCloud (US) Regions.

To alarm on multiple days of data, create or update an alarm using the CloudWatch console or Command Line Interface (CLI), specify a period of at least 3,600 (1 hour) and the number of datapoints that you want to compare against the threshold. For more information, visit the CloudWatch alarms documentation section.

 

​Amazon CloudWatch now lets customers evaluate metrics data for an extended duration of up to 7 days, a 7x increase from the previous limit of 24 hours. This enhancement empowers customers to monitor the health of longer-running or infrequent processes, such as daily data loading jobs or day-over-day performance trends, offering deeper insights into their resources and applications. Customers now have the flexibility to leverage CloudWatch alarms for use cases like near-real-time metric tracking as well as for monitoring patterns that span multiple days. Amazon CloudWatch alarming on multi-day data is now available in all AWS Regions including the AWS GovCloud (US) Regions. To alarm on multiple days of data, create or update an alarm using the CloudWatch console or Command Line Interface (CLI), specify a period of at least 3,600 (1 hour) and the number of datapoints that you want to compare against the threshold. For more information, visit the CloudWatch alarms documentation section.  

Publicado el Deja un comentario

AWS Marketplace launches automated version archiving for AMI and container products

Today, AWS Marketplace announces the availability of automated archival of old, unused product versions that are no longer available publicly for subscription (restricted). This feature is available for Amazon Machine Image (AMI), AMI with CloudFormation templates, and container products.

With this release, AWS Marketplace is streamlining the version management experience for AWS customers and sellers. With automated version archival, any product version that has already been restricted by a seller for longer than two years will be archived. Archived versions will no longer be available to launch from AWS Marketplace for new customers; however, existing users can continue to use them through launch templates and Amazon EC2 Auto Scaling groups by specifying the AMI ID. Any archived version that has not been used to launch any new instances in the previous 13 months will be deleted. Once an archived version is deleted, it is no longer available to launch for new or existing users.

Now AWS customers see only the latest versions of the products in AWS Marketplace, reducing the risk of using outdated versions. For sellers, it simplifies product management by automatically removing unused older versions.

This capability is enabled for all AMI and container products, and no additional action is needed from sellers. To learn more about this feature, see the AWS Marketplace Seller Guide.
 

 

​Today, AWS Marketplace announces the availability of automated archival of old, unused product versions that are no longer available publicly for subscription (restricted). This feature is available for Amazon Machine Image (AMI), AMI with CloudFormation templates, and container products. With this release, AWS Marketplace is streamlining the version management experience for AWS customers and sellers. With automated version archival, any product version that has already been restricted by a seller for longer than two years will be archived. Archived versions will no longer be available to launch from AWS Marketplace for new customers; however, existing users can continue to use them through launch templates and Amazon EC2 Auto Scaling groups by specifying the AMI ID. Any archived version that has not been used to launch any new instances in the previous 13 months will be deleted. Once an archived version is deleted, it is no longer available to launch for new or existing users. Now AWS customers see only the latest versions of the products in AWS Marketplace, reducing the risk of using outdated versions. For sellers, it simplifies product management by automatically removing unused older versions. This capability is enabled for all AMI and container products, and no additional action is needed from sellers. To learn more about this feature, see the AWS Marketplace Seller Guide.    

Publicado el Deja un comentario

AWS Elastic Beanstalk adds default support of EC2 Launch Template when creating new environments

With AWS Elastic Beanstalk, you can easily deploy and manage applications in AWS without worrying about configuring the infrastructure that runs those applications. Now, AWS Elastic Beanstalk will support EC2 Launch Template in Elastic Beanstalk environments by default after the deprecation of EC2 Launch Configuration. With this feature, you will not need to manually set certain configuration options in any AWS accounts, as described in documentation, to direct AWS Elastic Beanstalk to use the EC2 Launch Template when creating new environments.

Elastic Beanstalk’s support for EC2 Launch Template is generally available in commercial regions where Elastic Beanstalk is available including the AWS GovCloud (US) Regions. For a complete list of regions and service offerings, see AWS Regions.

For more information about with Launch Template support in Elastic Beanstalk please read our developer guide. To learn more about Elastic Beanstalk, visit the Elastic Beanstalk product page.

 

​With AWS Elastic Beanstalk, you can easily deploy and manage applications in AWS without worrying about configuring the infrastructure that runs those applications. Now, AWS Elastic Beanstalk will support EC2 Launch Template in Elastic Beanstalk environments by default after the deprecation of EC2 Launch Configuration. With this feature, you will not need to manually set certain configuration options in any AWS accounts, as described in documentation, to direct AWS Elastic Beanstalk to use the EC2 Launch Template when creating new environments.
Elastic Beanstalk’s support for EC2 Launch Template is generally available in commercial regions where Elastic Beanstalk is available including the AWS GovCloud (US) Regions. For a complete list of regions and service offerings, see AWS Regions.
For more information about with Launch Template support in Elastic Beanstalk please read our developer guide. To learn more about Elastic Beanstalk, visit the Elastic Beanstalk product page.  

Publicado el Deja un comentario

AWS Resource Groups now supports 172 more resource types

Today, AWS Resource Groups is adding support for an additional 172 resource types for tag-based Resource Groups. Customers can now use Resource Groups to group and manage resources from services such as AWS Entity Resolution, Amazon Personalize, and Amazon Q Apps.

AWS Resource Groups enables you to model, manage and automate tasks on large numbers of AWS resources by logically grouping your resources. You can create collections of resources such as applications, projects, or workloads, and manage them on dimensions such as cost, performance, and compliance in AWS services such as myApplications, AWS Systems Manager and Amazon CloudWatch.

AWS Resource Groups expanded resource type coverage is available in all AWS Regions, including the AWS GovCloud (US) Regions. You can access AWS Resource Groups through the AWS Management Console, the AWS SDK APIs, and the AWS CLI.

For more information about grouping resources, see the AWS Resource Groups user guide and the list of supported resource types. To get started, visit AWS Resource Groups console.

 

​Today, AWS Resource Groups is adding support for an additional 172 resource types for tag-based Resource Groups. Customers can now use Resource Groups to group and manage resources from services such as AWS Entity Resolution, Amazon Personalize, and Amazon Q Apps. AWS Resource Groups enables you to model, manage and automate tasks on large numbers of AWS resources by logically grouping your resources. You can create collections of resources such as applications, projects, or workloads, and manage them on dimensions such as cost, performance, and compliance in AWS services such as myApplications, AWS Systems Manager and Amazon CloudWatch. AWS Resource Groups expanded resource type coverage is available in all AWS Regions, including the AWS GovCloud (US) Regions. You can access AWS Resource Groups through the AWS Management Console, the AWS SDK APIs, and the AWS CLI. For more information about grouping resources, see the AWS Resource Groups user guide and the list of supported resource types. To get started, visit AWS Resource Groups console.  

Publicado el Deja un comentario

Amazon ElastiCache now supports 1-click connectivity setup between EC2 and your cache

Starting today, you can easily connect Amazon ElastiCache clusters to an Amazon Elastic Compute Cloud (Amazon EC2) instance directly from the AWS Management Console. You can also connect to your cache using AWS CloudShell to execute commands without additional setup.

With a single click, you can establish secure connectivity between your cache and an EC2 instance, following AWS recommended best practices. ElastiCache automatically configures VPCs, security groups, and network settings, eliminating the need for manual tasks like setting up subnets and ingress/egress rules. This streamlines the process for new users and developers, enabling them to launch a cache instance and connect it to an application within minutes.

You can also choose to connect to your cache from the Console using AWS CloudShell. Just click the “Connect to cache” button in the new “Connectivity and Security” tab. This will open a new AWS CloudShell session and connect to your cache using the valkey-cli tool. Once connected, you can execute common Valkey commands, including reading data (e.g. GET <key>) and writing data (e.g. SET <key> <value>). This lets you test cache functionality directly in the console, without needing to connect to it from an EC2 instance.

Learn more about setting up connectivity to a compute resource from your ElastiCache cluster in the Amazon ElastiCache user guide.

 

​Starting today, you can easily connect Amazon ElastiCache clusters to an Amazon Elastic Compute Cloud (Amazon EC2) instance directly from the AWS Management Console. You can also connect to your cache using AWS CloudShell to execute commands without additional setup. With a single click, you can establish secure connectivity between your cache and an EC2 instance, following AWS recommended best practices. ElastiCache automatically configures VPCs, security groups, and network settings, eliminating the need for manual tasks like setting up subnets and ingress/egress rules. This streamlines the process for new users and developers, enabling them to launch a cache instance and connect it to an application within minutes. You can also choose to connect to your cache from the Console using AWS CloudShell. Just click the “Connect to cache” button in the new “Connectivity and Security” tab. This will open a new AWS CloudShell session and connect to your cache using the valkey-cli tool. Once connected, you can execute common Valkey commands, including reading data (e.g. GET <key>) and writing data (e.g. SET <key> <value>). This lets you test cache functionality directly in the console, without needing to connect to it from an EC2 instance. Learn more about setting up connectivity to a compute resource from your ElastiCache cluster in the Amazon ElastiCache user guide.