SShortSingh.
Back to feed

Sakana AI Launches Fugu, a Multi-Agent System Coordinating GPT, Claude and Gemini

0
·1 views

Sakana AI has introduced Fugu, a multi-agent AI framework that orchestrates multiple frontier models — including GPT-5, Claude Opus, and Gemini Pro — through a single API endpoint. Rather than relying on one large model, Fugu acts as a project manager, breaking complex tasks into steps and assigning specialized roles such as coder, researcher, and editor to different AI models. The system is powered by two core components, TRINITY and the Conductor, which handle role assignment and inter-model communication respectively, with findings presented at the ICLR 2026 conference. Fugu is designed to be future-proof, allowing new models to be added to its pool without changing the user-facing API. Developers on macOS or Ubuntu can install and run Fugu locally using a single terminal command.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Log in to join the discussion and vote.

Log in

Related stories

0
ProgrammingDEV Community ·

Only 23% of LATAM Firms Gain Real Value from AI; Open ERPs May Close the Gap

Despite rapid growth in AI investment across Latin America — with the regional market projected to reach USD 40.5 billion in 2026 — only 23% of organizations generate measurable economic value from generative AI, and six in ten SMEs report no returns at all, according to a World Economic Forum and McKinsey report. Analysts attribute this 'value gap' primarily to poor integration, with AI tools deployed in isolation rather than embedded in core business systems. The argument is that embedding AI directly into an ERP — which already centralises accounting, sales, inventory, and purchasing data — addresses the root causes of low returns. Odoo's version 19, launched in September 2025, took this approach by integrating AI agents into core modules such as CRM, Accounting, and HR, using a human-review model to maintain oversight. For Mexican SMEs, where open-source software adoption is notably high at around 65%, open ERP platforms are positioned as a practical path from AI experimentation to measurable business outcomes.

0
ProgrammingDEV Community ·

Why AI May Be Stuck at Rule-Following While Human Creativity Gains Value

A developer essay argues that skill mastery occurs in two phases: an early 'science' stage of following inherited rules, and a later 'art' stage of transcending them through earned judgment. This framework mirrors the Dreyfus model of skill acquisition, which tracks learners from rigid rule-following as novices to fluid improvisation as experts. The author contends that generative AI is fundamentally anchored in the 'science' tier, since models are trained to predict statistically likely outputs that trend toward smooth, averaged results. Human creative work, by contrast, carries irregular and idiosyncratic choices that reflect a specific perspective shaped by personal context and history. Rather than devaluing human creativity, the author suggests AI's limitations may actually increase the worth of work that demonstrates genuine rule-transcending judgment.

0
ProgrammingHacker News ·

DNSGlobe: Open-Source Rust Tool Tracks DNS Propagation Globally in Terminal

DNSGlobe is a newly released open-source tool built using the Rust programming language. It provides a terminal user interface (TUI) that allows users to monitor DNS propagation across different regions of the world in real time. The project is hosted on GitHub under the organization 514-labs. It is designed for developers and network administrators who need visibility into how DNS changes spread globally. The tool was shared on Hacker News, where it garnered initial community attention.

0
ProgrammingDEV Community ·

JS Proxy API Powers Fine-Grained Reactivity in Modernised AngularJS Apps

A developer series exploring the migration from legacy AngularJS to modern fine-grained reactivity has released its second installment, focusing on a JavaScript Proxy-based runtime. The approach builds on a custom Go compiler introduced in Part 1, which transforms AngularJS templates into optimised JavaScript modules with mount and update methods. Rather than using AngularJS's expensive digest cycle for dirty-checking, the new runtime wraps the legacy $scope object in a Proxy to intercept property assignments the moment they occur. This allows state mutations made inside a controller to be automatically captured and translated into structured change objects passed to the compiled template's update method. The result is a lightweight, non-invasive bridge between legacy AngularJS controller patterns and a performant, surgically precise DOM update mechanism.

Sakana AI Launches Fugu, a Multi-Agent System Coordinating GPT, Claude and Gemini · ShortSingh