SuperAPI AI Directory
Partner with us中文AI newsFavorites

All AI tools

35,721 results · Page 845 of 993

RivalSearchMCP

Other AI tools

RivalSearchMCP is a deterministic research MCP server based on FastMCP 3.x, enabling AI agents to use multi-source search tools including web, social, academic, news, code, and documentation through a single connection. It provides 9 specialized tools, covering 5 search engines, 9 social platforms, 5 news sources, 5 academic databases and 4 dataset hubs, as well as GitHub repository search and document analysis. All tools require no API keys, the server does not run an LLM internally, and it returns deterministic, auditable structured output, automatically accompanied by quality scores and conflict detection signals, making it convenient for agent chaining. The project is MIT-licensed, open source and free, and can connect to the official hosted server or be self-deployed.

Read more

Rizzqo

Other AI tools

Rizzqo is an asset-centric compliance and risk management platform that maps the requirements of frameworks such as ISO 27001, ISO 27002, NIS2, DORA, CRA, GDPR, and the EU AI Act onto a company's actual IT assets, and automatically turns them into tasks with clearly defined owners, preserving evidence in real time. The platform quantifies security risk in euros and calculates return on security investment. Data is hosted in the user's country, made in Germany. If evidence is scattered across screenshots, spreadsheets, and isolated tools, there is no need to search around before an audit; compliance work becomes work that the team truly owns and can prove. It supports custom frameworks, provides workflows for asset, compliance, risk, task, and requirement management as well as monetized risk assessment, and supports Excel import and export.

Read more

rlhf-book

AI learning

"Reinforcement Learning from Human Feedback" is a textbook about reinforcement learning from human feedback (RLHF) that also broadly introduces post-training language models. The project was created by Nathan Lambert and is open source, offering online reading, purchase of the print book (Manning or Amazon), and a Discord community. The repository contains not only the book source files, but also reference implementations of RLHF algorithms and diagram source files for learners and researchers to use. The content covers instruction fine-tuning, policy gradient methods, reward model training, direct alignment methods, rejection sampling, on-policy distillation, and more. The book chapters use the CC-BY-NC-SA-4.0 license, while the rest, including code, uses the MIT license. The community also maintains an unofficial Simplified Chinese translation.

Read more

Roark

Other AI tools

Roark is a QA and evaluation platform for voice AI, focused on testing, monitoring, and improving existing voice AI agents, rather than being an agent-building tool. It can simulate-test agents before launch with hundreds of simulated callers (including different personas, accents, background noise, and edge cases), and score every production call against 500+ audio-native metrics, helping teams find failures, verify fixes, and launch with evidence. The platform supports 45 languages and accents, load and health checks, regression testing, CI/CD quality gates, and also provides red teaming, pronunciation testing, custom metrics, human review, issue tracking, alert monitoring, and OTEL tracing. It works with Vapi, Retell, LiveKit, Pipecat, and users' own tech stacks, and sign-up includes $50 in credits with no credit card required.

Read more