AI programming tools

runCLAUDErun is a native scheduler tool for macOS developers and automation users who use Claude Code to schedule Claude Code commands to run in the background at any time, interval, or condition. Its focus is on allowing local development tasks to be scheduled and retain local control on macOS, with key capabilities including native macOS apps, support for Apple Silicon and Intel Macs, and no registration or sign-in required. It's free to use, so it's good to start with a personal task. Note before use: Restrict directories, permissions, and modifiable file scopes before automatically executing code commands. If you plan to adopt it for a long time, it is recommended to test input lead time, output availability, manual review costs, and permission boundaries with real samples before deciding whether to put it into a fixed process.

Runcell is a Jupyter Notebook AI Agent for data analysts, researchers, and Python developers using Jupyter to understand notebooks, write Python code, execute cells, debug, and interpret analysis results. Its focus is on turning Jupyter into an interactive AI data analysis environment, with key capabilities such as being used as a Jupyter extension, writing code and executing cells, and supporting debugging and explaining data analysis. It offers free entry or trial credits, which are suitable for verifying results with small tasks first. Note before use: Data analysis conclusions still need to check samples, statistical methods, code results, and business context. If you plan to adopt it for a long time, it is recommended to test input lead time, output availability, manual review costs, and permission boundaries with real samples before deciding whether to put it into a fixed process.

Rosebud AI is an AI game and interactive world generation platform for game creators, educators, prototyping developers, and no-code creative users when creating games, 3D worlds, storybooks, and interactive apps with prompts. It focuses on moving game ideas from text descriptions to actionable prototypes, with key capabilities including support for Create Games with AI, including Game Creator, PixelVibe, and AI Storybook, and the ability to upload images as creative input. It offers free entry or trial credits, which are suitable for verifying results with small tasks first. Note before use: Before publishing, you need to check the gameplay copyright, material authorization, performance, and minor content boundaries. If you plan to adopt it for a long time, it is recommended to test input lead time, output availability, manual review costs, and permission boundaries with real samples before deciding whether to put it into a fixed process.

Rock-n-Roll is an AI application build plan generator tool for product managers and developers who are prototyping using tools like Lovable, Bolt, Claude, Cursor, and more to organize product ideas into structured implementation plans, prompts, and building blueprints. It focuses on clarifying scope, pages, data, and interaction logic before hands-on code generation, and its main capabilities include the ability to generate structured product plans, support for building tools like Lovable, Claude, and Cursor, and provide free project entry. It offers free entry or trial credits, which are suitable for verifying results with small tasks first. Note before use: After the plan is generated, the developer still checks the data structure, security permissions, and maintainability. If you plan to adopt it for a long time, it is recommended to test input lead time, output availability, manual review costs, and permission boundaries with real samples before deciding whether to put it into a fixed process.

Roboflow is a computer vision model development and deployment platform for developers, machine learning teams, and enterprises that need to build visual recognition applications to annotate images, train models, build vision workflows, and deploy to the edge, cloud, or API. It focuses on providing a complete computer vision toolchain from dataset to model deployment, with key capabilities including support for AI-assisted data annotation, Workflows, Train, Deploy, and Universe, and the ability to run models on device, edge, VPC, or API. It offers free entry or trial credits, which are suitable for verifying results with small tasks first. Note before use: Before the model goes live, verify data bias, misidentification risks, and accuracy in target scenarios. If you plan to adopt it for a long time, it is recommended to test input lead time, output availability, manual review costs, and permission boundaries with real samples before deciding whether to put it into a fixed process.

RepoClip is a GitHub repository demo video generator for developers, open-source project maintainers, and technical teams who need to demonstrate product prototypes when turning public GitHub repositories into demo videos with scripts, graphics, narration, and music. Its focus is on quickly forming shareable product explainer materials using repository links, and common capabilities include generating videos by entering public GitHub repositories, including scripts, visuals, narration, and music, and the first video can be generated for free. It offers free entry or trial credits, making it suitable for verifying the effect with small tasks first. Note before use: Private code, customer projects, and unpublished features should not be uploaded or publicly displayed. If the team is preparing for long-term adoption, it is recommended to test input materials, output quality, manual review costs, and permission boundaries with a set of real-world tasks before deciding whether to include a fixed process.

Reindeer is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

Refraction is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

Refact.ai is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

ReadmeChef is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

RapidNative is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

Raccoon AI is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

Qodex.ai is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

Qase is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

QA.tech is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

QA Sphere is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

ProtoBoost.ai is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

Prismic is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

pre.dev is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

playmix.ai is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

Pieces for Developers is an AI workflow tool for teams that need to create, organize, convert, or review task-specific material before final use. It should be used with clear source material, a defined output goal, and human review for accuracy, rights, privacy, and publishing quality.

Paperclip is an AI Agent management and automation platform that is mainly used to hire, organize and govern AI Agent teams, set goals, budgets, organizational structures and business tasks. It is suitable for developers, entrepreneurial teams, automation engineers and people who want to self-host AI Agent workflows. Common uses include building self-hosted AI Agent teams, setting goals and limits for business automation tasks, and studying AI labor management and governance models. When using it, be aware that Agent automation may perform real operations, and permissions, budgeting, auditing, and manual approval processes must be set. The open source project itself is free to use, and there are additional costs to run the model, database and infrastructure. It is recommended to use one or two low-risk tasks to test input materials, output quality, manual modification amount and final adoption ratio before deciding whether to put them into a fixed process.

Packfiles is a GitHub migration and configuration management tool mainly used to help enterprises plan and migrate code bases, teams and configurations such as Azure DevOps and Bitbucket Server, and accelerate migration tasks through Warp for Copilot. It is suitable for enterprise development platform teams, DevOps engineers, GitHub administrators and organizations that need to migrate code assets on a large scale. Common uses include unified migration of enterprises to GitHub, sorting out warehouses, teams and rights configurations, and reducing duplication during large-scale migrations. When using it, note that code migration involves permissions, history, CI/CD and security policies. Pilot warehouses and rollback plans should be made before formal migration. The page shows that starting from 100 free migrations, enterprise solutions can be billed per box or project. It is recommended to use one or two low-risk tasks to test input materials, output quality, manual modification amount and final adoption ratio before deciding whether to put them into a fixed process.

OwlityAI is an AI software quality testing platform that is mainly used to understand application interfaces through computer vision, automatically design tests, build automated processes, and discover defects. It is suitable for software teams, QA leaders, product teams and companies that need to reduce manual testing costs. Common uses include performing regression testing before going online, reducing duplication of manual QA work, and supplementing automated testing coverage for rapidly iterating products. When using it, note that autonomous testing cannot cover all business rules and boundary conditions. Complex authority, payment, compliance and core transaction processes still require manual QA to develop acceptance criteria. The page provides free trial and demonstration entrances, which is suitable for first verification with non-core applications. It is recommended to use one or two low-risk tasks to test input materials, output quality, manual modification amount and final adoption ratio before deciding whether to put them into a fixed process.

OnSpace.AI is an AI code-free application construction platform, which is mainly used to allow users to create applications, configure interfaces and generate business functions using natural language and visual processes, reducing the threshold for traditional zero-development. It is suitable for entrepreneurs, operations teams, product managers and non-technical users who want to quickly verify internal tools or MVPs. Common uses include verifying SaaS or tool-based MVPs, building internal management pages for teams, and turning forms, forms or processes into usable applications. When using it, note that code-less platforms are suitable for rapid verification and small and medium-sized businesses. When complex permissions, compliance data, deep integration, or high concurrency scenarios are involved, further architecture evaluation is needed. The page does not clearly display the full price. It is recommended to first confirm the release restrictions, data capacity and custom domain name capabilities from the trial or demonstration portal. It is recommended to use one or two low-risk tasks to test input materials, output quality, manual modification amount and final adoption ratio before deciding whether to put them into a fixed process.

Omnara is a coding agent command center mainly used to manage coding agents such as Codex and Claude Code from desktop, web and mobile terminals. It is suitable for developers, engineering teams, AI programming users, and people who need to manage agents remotely. It can manage coding agent sessions across devices, support parallel agents, work trees, and Git processes, and allow long-term tasks to continue running after the device is offline. When using it, note that the code generated by the agent still needs to be tested, reviewed and security checked, and cannot be directly merged into the production environment. It is recommended to use one or two low-risk tasks to test input materials, output quality, manual modification amount and final adoption ratio before deciding whether to put them into a fixed process.

Not Diamond is an AI model routing and proxy cost control platform, which is mainly used to automatically select appropriate models for different queries, improve accuracy and reduce costs. It is suitable for AI application developers, agent system teams, platform engineers and model operators. It can recommend more suitable models upon request, help agent systems control model invocation costs, and can also be used to compare model performance and routing strategies. When using it, note that routing results need to be continuously verified in conjunction with the evaluation set. Critical tasks cannot rely solely on automatic recommendations. It is recommended to use one or two low-risk tasks to test input materials, output quality, manual modification amount and final adoption ratio before deciding whether to put them into a fixed process.

Nora is a deep reasoning agent for software development. It is mainly used to provide deep reasoning and Web3 scenario support for software development tasks. It is suitable for developers, Web3 teams, technical leaders and people who need complex code assistance. It can conduct in-depth reasoning around software development tasks, assist in understanding code, requirements and technical contexts, and can also provide specialized support in Web3 development scenarios. Be aware when using it that it cannot replace code reviews, security audits and testing, especially contract or fund-related code. It is recommended to use one or two low-risk tasks to test the input materials, output quality, manual modification amount and final adoption ratio, before deciding whether to put them into a fixed process, and recording whether they are suitable for long-term use and team review.

Nitrode is an AI spatial inference data platform mainly used to provide dynamic environment understanding data for large models, intelligent agents and world models. It is suitable for AI research teams, robot teams, model training teams and data engineers. It can provide high-quality data related to spatial reasoning, help models understand dynamic environments and action relationships, and can also be suitable for preparation for LLM, agent and world model training. Pay attention when using it, it is biased towards data and model training scenarios, and the data authorization, format, evaluation method and training cost must be confirmed before adopting it. It is recommended to use one or two low-risk tasks to test input materials, output quality, manual modification amount and final adoption ratio before deciding whether to put them into a fixed process.

Nimbalyst is a visual workspace for coding agents, which is mainly used to manage agent sessions, tasks, files and documents such as Codex and Claude Code. It is suitable for developers, AI programming teams, open source maintainers, and people who need to manage agents in parallel. It can manage multiple coding agent sessions and tasks, edit Markdown, charts, mockups, and code materials, and use a visual workspace to organize files and development context. Be aware when using it, it is suitable for managing the agent collaboration process, and code merging, testing and online still need to be manually confirmed according to project specifications. It is recommended to use one or two low-risk tasks to test input materials, output quality, manual modification amount and final adoption ratio before deciding whether to put them into a fixed process.

Nadi is an application crash monitoring and AI care assistant that is mainly used to monitor application crashes and help development teams understand failure clues. It is suitable for mobile application teams, independent developers, SaaS technical teams and people who need to stably track exceptions. It can centrally view application crashes and exception signals, use AI to assist in understanding crash contexts and processing priorities, and help teams transform fault information into follow-up issues. Pay attention when using it. It is suitable for auxiliary troubleshooting and alarm sorting, and cannot replace real device reproduction, log analysis and code review; the SDK, privacy and abnormal data range must be confirmed before accessing. It is suitable to use one or two low-risk tasks to test the input materials, output quality, manual modification amount and final adoption ratio, and then decide whether to put them into a fixed process.

MyScale is an AI vector database combined with SQL. It is mainly used to process large-scale multimodal vector data using SQL analysis and vector search. It is suitable for AI application developers, data engineers, search teams and machine learning teams. It can combine vector search with SQL analysis and target large-scale multimodal vector data queries. It can also provide hosting capabilities to facilitate the construction of semantic search and RAG applications. Be aware when using it that it is more suitable for teams with data and engineering foundations; evaluate query performance, index costs, data size, and existing database compatibility before migrating. It is suitable to use one or two low-risk tasks to test input materials, output quality, modification costs and final adoption ratio before deciding whether to put them into a fixed process.

Mocha is an AI code-less application builder that is mainly used to turn application ideas into deployable websites or internal tools. It is suitable for entrepreneurs, product managers, independent developers and non-technical business teams. It can use AI to generate applications from descriptions, support application deployment and web building, and provide a rapid prototyping process without code. When using it, it should be noted that complex permissions, payments, data models and production operation and maintenance still require technical review to provide free points and payment plans. Before formal adoption, it is recommended to use low-risk samples to test once, and record the input materials, output results, and manual modifications. and the final adoption ratio before deciding whether to put them into a fixed process. At the same time, it is recommended to compare the trial results with existing processes to confirm whether the team can reuse them stably, rather than making long-term choices based on the one-time generation effect.

Mito is an AI Jupyter Notebook and data automation tool mainly used to handle database, Excel and Python automation in Notebooks. It is suitable for data analysts, Python users, researchers, and business analysis teams, and can provide AI Agents, connect SQL, database and spreadsheet data, generate analytical code, dashboards, and automate processes in the Jupyter workflow. Pay attention when using it. Automatic code generation requires running tests and checking data caliber. Production environment tasks must be included in version management. Free quotas and team plans are provided. Before formal adoption, it is recommended to test with low-risk samples first, record input materials, output results, The amount of manual modifications and the final adoption ratio, and then decide whether to put them into a fixed process.

Mintlify is an AI-native document and knowledge platform mainly used to create product documents and knowledge content for developers and AI. It is suitable for developer tool companies, API teams, technical writing teams, and engineering organizations. It can create beautiful technical document stations, support AI native knowledge management, and be structured for developers and AI agent reading. When using it, you should note that the quality of documents depends on the information architecture, version maintenance and accurate examples. You cannot rely solely on automatic generation to provide trials and team subscriptions. Before formal adoption, it is recommended to use low-risk samples to test once, and record the input materials, output results, and manual modifications. The amount and final adoption ratio are then decided whether to put them into a fixed process.

MindsDB is a natural language business intelligence and predictive analysis platform that is mainly used to analyze enterprise data, trigger actions and generate predictive insights in natural language. It is suitable for data teams, business analysts, developers and enterprise AI application teams. It can query business data through natural language, build predictive insights and automated actions, and connect enterprise data sources for real-time AI analysis. Attention should be paid to when using it. Before accessing production data, permissions, audits, indicator definitions and security policies are required. Automatic actions must be set up to provide a manual confirmation link to provide cloud and enterprise use paths. Before formal adoption, it is recommended to test once with low-risk samples and record the input materials, output results, manual modifications and final adoption ratio, and then decide whether to put them into a fixed process.

Middleware is an AI SRE full-stack observable platform that is mainly used to monitor applications, infrastructure, logs and real user sessions. It is suitable for DevOps, SRE, back-end teams and cloud-native engineering organizations. It can provide full-stack monitoring and diagnosis, use AI SRE Agents to assist in problem discovery, and support data ingestion control and synthetic inspections. Pay attention when using it. The observable platform needs to be correctly buried, alarm classification and accident process. Otherwise, it is easy to generate noise. Free data quotas and charge-by-volume. Before formal adoption, it is recommended to use low-risk samples to test once and record the input materials., output results, manual modifications and final adoption ratio, and then decide whether to put them into a fixed process.

Metaflow is a machine learning and AI project workflow framework mainly used to build, expand and deploy real-life machine learning, AI and data science projects. It is suitable for machine learning engineers, data scientists, platform teams and production environment AI projects. It can manage machine learning and data science workflows, support cloud expansion and production deployment, and make experimental, computing and deployment processes more traceable. When using it, it should be noted that it is an engineering framework that requires experience in code, infrastructure and model operation and maintenance. It is not suitable for open source use by completely untechnical users. Cloud resources are calculated separately for cost calculation. Before formal adoption, it is recommended to test it with low-risk samples and record the input materials, output results, manual modifications and final adoption ratio, and then decide whether to put it into a fixed process.

Metabob is an AI code analysis and debugging auxiliary tool. It is mainly used to assist parallel generative programming tools for defect analysis and refactoring. It is suitable for software engineers, code reviewers, AI programming users and R & D teams. It can provide real-time intelligent code analysis, discover potential defects and security implementation issues, and assist in debugging and refactoring legacy code. Pay attention when using it that static analysis and AI recommendations require developer verification and cannot replace testing, code review and security evaluation. Before individual developer plans and team payment plans are officially adopted, it is recommended to test with low-risk samples first and record the input materials., output results, manual modifications and final adoption ratio, and then decide whether to put them into a fixed process.

MaxKB is an open source enterprise-level Agent and knowledge base platform, mainly used to build RAG knowledge base, enterprise customer service, internal Q & A and intelligent workflow. It is suitable for development teams, enterprise IT, knowledge base administrators, customer service teams and educational and research institutions. It can integrate RAG processes for knowledge-enhanced Q & A, support Agent workflow and model-independent access, and provide enterprise-oriented capabilities such as MCP tool calls. When using it, note that document governance, permissions, model selection, and operation and maintenance resources need to be prepared before deployment; the quality of the knowledge base directly affects the reliability of answers. Open source platforms can be deployed on their own, and enterprises need to calculate server, model and maintenance costs for implementation. Before formal adoption, it is recommended to test once with low-risk materials or small samples, record the input quality, output results, manual modifications and final adoption ratio, and then decide whether to put them into the long-term workflow.

Macroscope is an AI code review tool for modern development teams, offering automated PR descriptions, code reviews, status updates, ticket context, and collaboration link support for Slack, GitHub, Jira, Libratic, and more. It's suitable for teams looking to merge PRs faster, catch bugs early, and synchronize development status. Make it clear that it is a code review aid rather than a final quality assurance, and key architecture, security, permissions, and data migration changes still need to be reviewed and tested by senior engineers. Before formal adoption, it is recommended to test with real but low-risk materials to check output quality, authorization boundaries, privacy handling, and manual review costs before deciding whether to put them into a long-term workflow.

Macaly is an AI website and web application building platform for founders, small businesses, and non-professional developers to generate launchable websites or apps using natural language prompts. It supports conversational description of requirements, sketch to build, data saving, direct editing, file management, access analysis, and SEO-related capabilities. When using it, it should be used as a rapid prototype and lightweight product entrance, and formal business still needs to check security, data structure, mobile adaptation, payment permissions and long-term maintenance methods. Before formal adoption, it is recommended to test with real but low-risk materials to check output quality, authorization boundaries, privacy handling, and manual review costs before deciding whether to put them into a long-term workflow. For individuals and teams, a safer approach is to retain the manual review node first, and then decide whether to expand the scope based on the results of several consecutive times.

Lumi.new is a website development tool that generates websites through AI chat, allowing users to describe desired pages or apps in natural language and quickly get iterative website results. It's suitable for solo developers, entrepreneurs, designers, product managers, and those who need to prototype quickly. AI-generated code needs to be manually checked for security, performance, responsive layout, copyright materials, and deployment configurations, and unreviewed results cannot be directly used for formal business. Before official adoption, it is recommended to make a sample around the "Generate Website with AI Dialogue" to check whether the output meets the requirements of real tasks, material authorization, data security, and manual review before deciding whether to enter the long-term process.

LM Studio is a desktop-oriented local large model running tool suitable for downloading, managing, and running open models such as gpt-oss, Llama, Gemma, Qwen, and DeepSeek on personal computers. It emphasizes local and private use, making it suitable for developers, researchers, content teams, and users who want to test the model's performance natively first. Before use, you need to pay attention to the computer memory, memory, model volume, and licensing terms, and make sure that the data is completely left in the local process when formally handling sensitive data. Before official adoption, it is recommended to make a sample around "downloading and managing a variety of open large models on the desktop" to check whether the output meets the requirements of real tasks, material authorization, data security and manual review, and then decide whether to enter the long-term process.

Locofy.ai is an AI design conversion tool that converts drafts like Figma, Penpot, and more into front-end code like React, React Native, HTML, Flutter, Vue, Angular, or Next.js. It's suitable for front-end teams, design systems teams, startup product teams, and those who need to quickly turn prototypes into code. Before use, it is recommended to conduct small-scale testing with real materials or real processes, focusing on observing output quality, review costs, payment boundaries, data permissions, and whether the team can establish a stable manual review process. Before handling formal business, it should also be judged based on material authorization, privacy requirements, and manual review standards, and avoid using automatic results directly for external release or key decisions. If used in team, client, or teaching scenarios, the source of information, the responsibility for reviewing the results, and the scope of external use should also be clearly entered first.

Local AI Playground is a local AI model experimentation tool that allows users to experiment with AI models locally, lowering the barrier to technical setting and providing model management, validation, and inference-related capabilities. It's suitable for AI enthusiasts, students, developers, product prototyping teams, and those looking to test models locally. Before use, it is recommended to conduct small-scale testing with real materials or real processes, focusing on observing output quality, review costs, payment boundaries, data permissions, and whether the team can establish a stable manual review process. Before handling formal business, it should also be judged based on material authorization, privacy requirements, and manual review standards, and avoid using automatic results directly for external release or key decisions. If used in team, client, or teaching scenarios, the source of information, the responsibility for reviewing the results, and the scope of external use should also be clearly entered first.

Liner.ai is a no-code machine learning tool that allows users to train and deploy machine learning applications without writing code, making it ideal for quickly validating image, text, or tabular model ideas. It's suitable for students, product teams, data beginners, educators, and non-engineering users who want to validate machine learning ideas. Before use, it is recommended to conduct small-scale testing with real materials or real processes, focusing on observing output quality, review costs, payment boundaries, data permissions, and whether the team can establish a stable manual review process. Before handling formal business, it should also be judged based on material authorization, privacy requirements, and manual review standards, and avoid using automatic results directly for external release or key decisions. If it is used for teams, customers, or teaching scenarios, it is also necessary to clarify the input source, result review responsibility, and scope of external use, and avoid putting the trial results directly into the formal process.

Lightning AI is an AI development platform that provides an all-in-one AI development environment from browser coding, prototyping, model training, to deployment services, from the PyTorch Lightning team. It is suitable for machine learning engineers, AI startup teams, researchers, students, and developers who need a cloud-based development environment. Before use, it is recommended to conduct small-scale testing with real materials or real processes, focusing on observing output quality, review costs, payment boundaries, data permissions, and whether the team can establish a stable manual review process. Before handling formal business, it should also be judged based on material authorization, privacy requirements, and manual review standards, and avoid using automatic results directly for external release or key decisions. If it is used for teams, customers, or teaching scenarios, it is also necessary to clarify the input source, result review responsibility, and scope of external use, and avoid putting the trial results directly into the formal process.