The Week in AI: IBM's Concert, Fireworks AI's Firefunction v2, Gretel's Synthetic Data Set, LogicMonitor's Edwin AI, More

Last week's list of product and services announcements in the AI space is a long one. It includes the GA release of IBM's Concert tool for leveraging GenAI to deal with app sprawl, Solventum's new AI-driven payment integrity and revenue cycle solution, the launch of LogicMonitor's AI copilot for IT operations, Target's newly released GenAI tool for employees at 2,000 stores, and more!

IBM announced the general availability of IBM Concert, a new tool designed to leverage generative AI to address application sprawl and enhance business efficiency. Unveiled at the IBM Think 2024 conference in Boston, IBM Concert aims to help developers and site reliability engineers manage growing application installations across cloud and on-premises environments. IBM pointed to the urgent need to manage the rapid growth of cloud-native apps, projected to double by 2028. IBM Concert offers insights into connected apps, simplifies compliance, and automates application management through the watsonx platform. Its capabilities include analysis, visualization, and recommendations to speed up processes, reduce complexity, and improve resiliency.

Fireworks AI has launched Firefunction-v2, an open-source function-calling model optimized for real-world scenarios, including multi-turn conversation, instruction following ,and parallel function calling. It retains Llama 3’s multi-turn instruction capability (0.84 vs 0.89 on MT bench) while consistently outscoring Llama 3 on function calling tasks (0.51 vs 0.30 on Nexus parallel multi-function eval), the company said. The model is designed to have "the intelligence and generalizability to handle real-world agentic use cases," the company said, including things like chat assistants, which involve function-calling capabilities, alongside general chat and instruction following. Firefunction-v2 is fast and economical, the company said, promising seamless integration into existing systems, supported by an OpenAI-compatible API, and a user-friendly demo app.

Gretel, a provider of tools for synthetic data generation, introduced Gretel Navigator, an agent-based, compound generative AI system built to automate data creation and curation processes for AI development. With simple natural language or SQL prompts, Gretel Navigator enables users to create, edit, and augment tabular data, and design realistic, high-quality test and training datasets from scratch. Developers can also leverage existing datasets to generate insight-rich synthetic data on demand. Gretel Navigator addresses traditional challenges with data acquisition head-on, the company said, by enabling developers to generate customizable, realistic synthetic datasets that mimic real-world patterns without compromising individual privacy. Navigator supports a wide range of data formats, modalities, and context-specific optimizations to streamline workflows and expedite AI projects.

LogicMonitor Unveils AI Copilot for IT Operations
LogicMonitor launched Edwin AI, a generative AI copilot designed to aid IT operations staff. Aimed at reducing "alert fatigue," diagnosing incidents swiftly, and simplifying observability, Edwin AI leverages retrieval-augmented generation and pre-trained large language models (LLMs) connected to customer data. Developed without a GPT wrapper, Edwin AI can process data from more than 4,000 integrations, the company said, offering readable summaries of complex alerts, troubleshooting suggestions, and step-by-step remediation. Early tests indicate it can reduce alerts by up to 95%, according to the company. Edwin AI will soon be available as an add-on to LogicMonitor’s core service.

Researchers from Stanford, UC Berkeley, Google DeepMind, MIT, and others launched OpenVLA, an open-source visual-language-action (VLA) AI model for robot guidance. Designed to enable robots to understand scenes and execute tasks from plain language prompts, OpenVLA is built on a pre-trained Prismatic-7B VLM and a Llama 2 7B large language model (LLM). It uses a visual encoder to process images into embeddings, facilitating task execution based on natural language inputs. Trained with more than 970,000 robot manipulation indexes from the OpenX dataset on 64 Nvidia A100 GPUs, OpenVLA aims to advance robotics by simplifying task commands into robotic actions, the researchers said. It significantly outperformed the closed-source RT-2-X model in various tests. OpenVLA’s code is available on GitHub, with model checkpoints on Hugging Face, promoting broader research in embodied AI.

Solventum, previously known as 3M Health Care, launched an AI-driven payment integrity and revenue cycle solution, the Solventum Revenue Integrity System, in collaboration with Sift Healthcare. The new tool aims to help health systems reduce and prevent claim denials, ensuring timely and accurate reimbursement. With the healthcare industry facing tight margins and increased administrative costs due to claim denials, this solution targets clinical documentation integrity, coding, and utilization review workflows. By integrating machine learning interventions, Solventum's system offers real-time insights and predictive analytics, shifting healthcare providers from a reactionary to a proactive approach. This collaboration supports health systems in optimizing payment outcomes throughout the patient's clinical journey.

Data security startup Immuta unveiled new data governance and audit features for retrieval-augmented generation (RAG) AI solutions across multiple cloud platforms. RAG applications enhance large language models (LLMs) with external data for improved content accuracy. Immuta's latest release provides a multilayer architecture for securing, monitoring, and auditing sensitive data accessed by RAG-based AI applications. This development aims to address the growing challenge of AI data security. A recent survey found 80% of data experts see AI complicating data security, while 88% note employee use of AI regardless of official company adoption. Immuta's solution focuses on the storage and data layers, offering fine-grained access control in collaboration with Amazon Web Services and enhancing data security through comprehensive policy enforcement.

Target announced plans to roll out a new generative AI technology for employees at nearly 2,000 stores nationwide by August. The company is introducing Store Companion, an AI-powered chatbot app for store associates’ handheld devices, designed to assist with process and procedure-related queries. Currently, the chatbot is in pilot testing at around 400 locations. Target is also leveraging AI to enhance product display pages, offering personalized search results and product review summaries. Guided search features have also been introduced on Target’s online store, enabling broader and more relevant product searches. The enhanced search experience will be available to all customers later this summer.