Petals lets users run large AI models collaboratively via distributed network

·1 views

Petals is a new platform that enables users to run large language models at home using a distributed, BitTorrent-style approach. Rather than requiring a single powerful machine, the system splits model layers across multiple volunteer computers on a network. This makes running state-of-the-art AI models accessible without expensive hardware. The project was shared on Hacker News, where it attracted community attention and discussion. More details are available at petals.dev.

Read the full story at Hacker News

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

Discussion (0)

How to Build Idempotent Stripe Webhook Handlers for File Delivery

A software team building a document-generation SaaS product shared lessons from hardening their post-payment fulfillment pipeline, which spans databases, PDF renderers, cloud storage, and email providers. The core insight was to treat Stripe webhooks not as a trigger to execute tasks, but as events that advance a persisted state machine. Before redirecting users to Stripe Checkout, the team persists all order data — including document content and a deterministic hash — so fulfillment can proceed even if the customer closes their browser after paying. To prevent duplicate processing, they use the Stripe Checkout Session ID as a unique database key and enforce a single generation job per purchase type, resetting only explicitly retryable failure states. Granular job statuses such as 'pdf_generation_failed' and 'email_failed' are stored separately from purchase status, allowing engineers to diagnose and recover from failures at each stage independently.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Why PDF Export Tools Need Three Validation States, Not Two

A development team building a print-oriented web app found that standard two-state form validation — valid or invalid — was insufficient for safely generating exportable PDFs. They designed a three-state model using statuses of 'pass', 'needs_review', and 'cannot_export' to better reflect real-world document risks. Errors such as missing required fields or overflowing text blocks block export entirely, while warnings like a low-resolution photo allow the user to proceed after acknowledgment. Each flagged issue is tied to a specific section and includes a direct repair link, reducing the burden on users to locate problems manually. The model also validates renderer output and layout geometry, catching structural failures that field-level validators cannot detect.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Gitlord turns AI agent conversations into trackable git commits

A new open-source tool called Gitlord converts every turn in an AI agent conversation into a git commit, making session history navigable, forkable, and diffable like code. The tool supports branching for subagents, allowing parallel agent workflows without disrupting the main session. Gitlord works with over 170 AI providers and nearly 3,000 models, including local models, and connects external tools via the Model Context Protocol. It also includes built-in RAG support using ChromaDB for semantic search across agent history. Released under the MIT license, the project is available on GitHub and aims to address context management and reproducibility challenges common in existing agent frameworks.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How Middleware Bridges POS, PMS, and ERP Systems in Hospitality Operations

Hotels and restaurants typically use separate systems — POS, PMS, and ERP — to manage payments, reservations, and back-office functions respectively. When these platforms operate in isolation, staff must manually transfer data, which slows workflows and leads to inconsistent records. Middleware addresses this by acting as a connection layer that routes information between systems automatically. Tools like Odoo Integra can link operational platforms such as Resto Plus and KloudHotels directly with Odoo ERP. Effective integration goes beyond simple connectivity, requiring data mapping, validation rules, security controls, and ongoing monitoring to ensure accuracy.

0 comments Read more at DEV Community