Weaviate Alternatives

Key Takeaways

When choosing a vector database for your AI applications, understanding the landscape of Weaviate alternatives is crucial. Based on extensive research and community feedback, here are the essential insights to guide your decision:

• Qdrant stands out for its performance and user-friendly API, with users reporting 98-99% accuracy and significantly faster query response times compared to Weaviate.
• Pinecone offers a fully-managed cloud service that excels in user experience but comes at a higher cost ($70/month compared to Weaviate's $25/month).
• Milvus has the strongest community support with 32,742 GitHub stars (compared to Weaviate's 12,533) and excels at managing billions of vector embeddings.
• Chroma is optimized for AI-native applications and large language model development, though it may have limitations in horizontal scaling.
• PgVector provides a cost-effective solution by extending PostgreSQL, making it ideal for organizations already using PostgreSQL infrastructure.

Notable Alternatives to Weaviate

The vector database landscape offers several robust alternatives to Weaviate, each with distinct advantages. According to MyScale's comparison, the most notable options include:

1. Pinecone: Known for its focus on user experience and seamless interface, Pinecone provides a starter edition free for up to 100,000 records.
2. Qdrant: This cloud-native search engine emphasizes scalability and adaptability, making it particularly effective for high-volume workloads.
3. Milvus: Renowned for its lightning-fast searches, Milvus excels in managing high-volume queries, supporting up to 65,536 collections.
4. Chroma: Features advanced visualization tools and streamlines LLM application development with efficient embedding management.
5. Faiss: Developed by Facebook AI Research, this library specializes in efficient similarity searches and clustering for large vector sets. Community feedback consistently highlights Qdrant and Milvus as particularly strong alternatives. As noted in Reddit discussions, many users report improved stability and performance after switching from Weaviate to these alternatives.

Key Differences and Similarities

When comparing Weaviate with its alternatives, several key differences emerge:

Performance Metrics

According to a performance analysis, Weaviate has shown limited improvement over time in terms of requests per second (RPS) and latency, becoming one of the slower options. In direct comparison:

• Search Speed: Weaviate averages 0.12 seconds for queries versus Pinecone's 0.88 seconds, according to comparative benchmarks.
• Data Loading: Pinecone loads at 4.3 milliseconds per vector versus Weaviate's 5 milliseconds per vector.
• Scalability: Milvus can handle over 3 million records, while Weaviate faces challenges with multi-billion vector datasets.

Feature Comparison

Each alternative offers unique features:

• Weaviate: Combines object and vector storage, supports GraphQL API, and enables hybrid search with structured filtering.
• Qdrant: Features an efficient Rust-based architecture with vector quantization techniques that reduce RAM usage by up to 97%.
• Pinecone: Offers fully-managed service with separation of compute and storage, simplifying infrastructure management.
• Milvus: Provides advanced indexing capabilities and millisecond search results on datasets with trillions of vectors.
• Chroma: Specializes in AI-native applications with particular strength in audio data processing.

Deployment Options

According to Zilliz's comparison, deployment flexibility varies significantly:

• Weaviate: Supports on-premises, cloud, and embedded deployments.
• Pinecone: Exclusively cloud-based with no local deployment options.
• Milvus: Offers both on-premises and cloud deployments with flexible scaling.
• Qdrant: Supports both on-premises and cloud deployments.
• Chroma: Features an embedded-first architecture ideal for local environments.

User Experiences and Community Support

Community feedback provides valuable insights into real-world performance and usability:

GitHub Engagement

GitHub stars serve as a metric for community interest:

• Milvus: 32,742 stars
• Qdrant: 22,124 stars
• Chroma: 18,077 stars
• Weaviate: 12,533 stars

User Testimonials

According to G2's competitor analysis, users consistently report that alternatives like Qdrant and Milvus are:

• Easier to set up
• Provide better support
• Offer superior performance for large-scale applications One user noted that Qdrant delivers "perfect" performance with accuracy between 98-99%, while another highlighted Milvus's exceptional community support for troubleshooting.

Best Practices for Choosing the Right Vector Database

When selecting a Weaviate alternative, consider these essential factors:

1. Project Scale: For smaller projects, Chroma and PgVector offer simplicity. For enterprise-scale needs, Milvus and Pinecone provide robust scaling capabilities.
2. Deployment Preferences: If you need on-premises deployment, avoid Pinecone and consider Milvus, Qdrant, or Weaviate.
3. Budget Constraints: Self-hosted options like PgVector and Qdrant generally offer better cost efficiency than managed services like Pinecone.
4. Performance Requirements: For applications requiring millisecond response times, Milvus and Qdrant consistently outperform other options.
5. Integration Needs: Consider compatibility with existing infrastructure—PgVector excels if you're already using PostgreSQL.
6. Data Complexity: For multi-modal data (text, images, audio), Weaviate and Chroma offer superior handling capabilities. As vector databases continue to evolve, the best choice ultimately depends on your specific use case, technical requirements, and organizational constraints.

---

🚀 Take Action Now

• Find your next profitable AI app idea validated by real data
• Unlock access to 61,988+ (and growing) validated keywords with market demand
• Explore the fastest-growing AI tools and competition
• Search our database of 2,269+ (and growing) AI applications to inform your next project

---

Introduction

The AI landscape is evolving at breakneck speed, generating an unprecedented volume of unstructured data. According to Analytics Vidhya, we're now producing over 3.5 quintillion bytes of data daily. Traditional databases, designed for structured information, simply can't handle this explosion of complex, high-dimensional data efficiently.

Vector databases have emerged as the critical infrastructure for modern AI applications. These specialized systems transform unstructured data—text, images, audio, and video—into mathematical vectors, enabling machines to understand semantic relationships and similarities. Forrester forecasts vector database adoption will triple from 6% to 18% within the next year, underlining their growing importance.

Why this sudden surge in popularity? Vector databases excel where traditional databases falter:

• Enhanced Security and Privacy: They provide better protection for sensitive data.
• Improved Scalability: They efficiently handle the growing volume of AI-generated content.
• Greater Trust in Results: They deliver more accurate and contextually relevant search outcomes.
• Native Support for Vector Data: They're optimized for the high-dimensional data that powers AI applications. In this rapidly expanding ecosystem, Weaviate has established itself as a leading open-source vector search engine. Its versatility through contextualized embeddings and user-friendly attributes has made it a popular choice for AI product development. Weaviate combines object and vector storage, allowing for both semantic searches and structured filtering across multiple media types.

However, as AI applications become more diverse and demanding, developers increasingly need to evaluate whether Weaviate is the optimal solution for their specific requirements. With different projects requiring varying levels of performance, scalability, and specialized features, understanding the full spectrum of vector database options becomes crucial.

This article aims to provide a comprehensive comparison of the most viable Weaviate alternatives, including Qdrant, Pinecone, Milvus, and Chroma. We'll examine their distinct features, performance metrics, deployment options, and real-world applications to help you make an informed decision for your AI projects. Whether you're building a recommendation system, implementing semantic search, or developing the next generation of AI-powered applications, choosing the right vector database is a decision that will significantly impact your project's success.

Overview of Weaviate

Weaviate is an AI-native vector database designed to bridge the gap between structured and unstructured data. Launched in 2019 by Weaviate B.V., this open-source platform has quickly established itself in the vector database landscape with 12,533 GitHub stars as of recent counts, according to Zilliz's comparison.

What Weaviate Offers

At its core, Weaviate functions as a real-time vector search engine that integrates machine learning models for scalable applications. Built with Go, it provides a GraphQL-based interface that simplifies complex queries while maintaining powerful search capabilities. According to DB-Engines, Weaviate positions itself as an AI-native real-time vector database engine, focusing on generative AI and semantic search applications.

The platform's architecture stands out by combining both object and vector storage within a single system. This integration eliminates the need for third-party databases, creating a self-contained solution that stores both the original objects and their vector representations. As highlighted by Reddit discussions, this approach delivers quick response times for 10-Nearest-Neighbor queries—typically between 50 to 100 milliseconds—even when dealing with millions of objects.

Key Features and Strengths

Weaviate's distinctive capabilities make it suitable for various AI applications:

• Hybrid Search Capability: Weaviate excels in combining vector-based and traditional keyword searches, creating a versatile search experience. According to Medium, this feature makes it particularly effective for semantic searches that require understanding context and relationships.
• Multi-Modal Data Support: The platform handles diverse data types including text, images, audio, and video, making it adaptable for complex applications.
• Schema Flexibility: Rockset's comparison notes that Weaviate allows users to define classes for data associated with vectors, offering versatility similar to relational databases.
• Mixed Search Queries: Users can combine fuzzy vector searches with structured queries, enabling complex operations. For example, retrieving articles related to specific topics within particular date ranges.
• Integrated Vectorization: As documented by Weaviate's blog, users can directly import text for vectorization without preparing vectors beforehand, streamlining the workflow.
• Multiple Model Support: The system facilitates simultaneous use of different query/document embedding models, allowing for live experimentation without committing to complete transitions.

Use Cases and Target Users

Weaviate serves a diverse range of applications and user needs:

1. Semantic Search Implementation: Organizations looking to enhance search functionalities beyond keyword matching benefit from Weaviate's contextualized understanding.
2. Content Personalization: MyScale's analysis indicates that Weaviate's semantic capabilities make it ideal for delivering personalized content recommendations.
3. Data Classification: Enterprise resource planning (ERP) software can leverage Weaviate for efficient data categorization.
4. Natural Language Processing: Developers building NLP applications appreciate Weaviate's ability to understand linguistic nuances.
5. Image and Video Analysis: The multi-modal support enables effective processing of visual data alongside textual information. The target user base primarily consists of developers and organizations implementing AI-powered applications where semantic understanding is crucial. Weaviate particularly appeals to teams that prefer a self-contained database solution with integrated object and vector storage. Its BSD 3-Clause License model makes it accessible for various projects, while also offering enterprise versions for larger implementations.

Weaviate's pricing structure starts at $25 per month for hosting on a performance-optimized instance, according to comparative benchmarks. This makes it more cost-effective than some alternatives, though users must consider the technical expertise required for optimal deployment and scaling.

While Weaviate provides robust capabilities for many AI applications, understanding its particular strengths and limitations relative to alternatives is essential for selecting the most appropriate vector database for specific project requirements.

Comparative Analysis of Weaviate Alternatives

While Weaviate offers robust capabilities for AI applications, several alternatives have emerged with unique strengths and specialized features. Let's examine the most prominent options to help you determine which vector database best suits your specific requirements.

A. Qdrant

Introduction to Qdrant and Its Features

Qdrant is an open-source vector similarity search engine governed by the Apache License 2.0. With 22,124 GitHub stars, it has garnered significant community interest compared to Weaviate's 12,533, according to Zilliz's comparison. Built using Rust, this high-performance database specializes in neural network-based semantic matching.

Key features that distinguish Qdrant include:

• Vector Quantization: Techniques that can reduce RAM usage by up to 97%, enabling efficient resource management, as highlighted by Instaclustr.
• Hybrid Search Support: Handles both sparse and dense vectors, allowing for effective token weighting through transformer-based neural networks.
• Payload-Based Filtering: Enables filtering of search results based on vector payloads, enhancing search precision.
• Dynamic Sharding: Facilitates horizontal scaling, though Reddit discussions note this feature is still under development.

Performance Metrics Comparison with Weaviate

Performance benchmarks reveal significant advantages for Qdrant:

• Accuracy: Community feedback on Reddit reports Qdrant achieving 98-99% accuracy in production environments.
• Query Response: Qdrant demonstrates faster query responses, particularly for large datasets, with one user noting it's "26 times faster than Weaviate" for processing 1 million vectors, according to comparative testing.
• Resource Efficiency: Qdrant's Rust-based architecture provides superior performance with lower resource requirements, making it cost-effective for applications requiring up to 2TB of storage.

Advantages and Use Cases for Qdrant

Qdrant excels in several scenarios:

• Real-time Applications: Particularly effective for recommendation systems and personalized searches requiring immediate response.
• Large-Scale Deployments: Suitable for handling complex queries across massive datasets, with users like Johnson & Johnson and Discord leveraging its capabilities, as noted by CloudRaft.
• Hybrid Search Operations: Ideal for implementations requiring both vector similarity and traditional filtering.
• Cost-Sensitive Projects: The efficient resource utilization makes it economical for startups and resource-constrained environments.

B. Pinecone

Overview of Pinecone as a Managed Solution

Pinecone offers a fully managed vector database service optimized for similarity search. Unlike Weaviate, it operates exclusively in the cloud with no local deployment options. According to Zilliz's analysis, Pinecone focuses on high performance and scalability specifically for cloud environments.

Notable characteristics include:

• Proprietary Architecture: Operates under a closed-source model, limiting customization but potentially offering optimized performance.
• Real-time Indexing: Enables immediate updates, making it suitable for dynamic datasets.
• Separation of Compute and Storage: Provides flexibility in resource allocation and scaling.
• Starter Edition: Offers a free tier for up to 100,000 records, making it accessible for smaller projects.

Ease of Use and Performance Benchmarks

Pinecone prioritizes user experience at the expense of some performance metrics:

• Setup Simplicity: G2's competitor analysis indicates users find Pinecone easier to set up than Weaviate, though it's slower to achieve ROI.
• Loading Speed: Pinecone loads vectors at 4.3 milliseconds per vector compared to Weaviate's 5 milliseconds, according to Medium benchmarks.
• Query Latency: Despite its user-friendly interface, Pinecone shows higher query latency at 0.88 seconds versus Weaviate's 0.12 seconds in the same tests.
• Cost Structure: Pinecone's pricing starts at approximately $70 per month, significantly higher than Weaviate's $25 per month for comparable performance.

Scenarios Best Suited for Pinecone

Pinecone performs particularly well in these use cases:

• Enterprise Applications: Companies like Microsoft and Shopify leverage Pinecone for production workloads where managed infrastructure is preferred over self-maintenance.
• Rapid Prototyping: The straightforward API and managed nature make it ideal for quick development cycles.
• Scaling Concerns: Organizations anticipating rapid growth benefit from Pinecone's elastic infrastructure that scales without developer intervention.
• Threat Detection: Real-time analysis capabilities make it effective for cybersecurity applications requiring immediate pattern recognition.

C. Milvus

Features and Strengths of Milvus

Milvus stands as the most popular open-source vector database with 32,742 GitHub stars, significantly outpacing Weaviate's community engagement. Founded in 2019, this database is designed specifically for embedding similarity search with millisecond query capabilities across vast datasets.

Key strengths include:

• Massive Scale Support: Handles over 3 million records and up to 65,536 collections, according to Reddit discussions.
• Multiple Index Types: Supports various indexing methods to optimize for specific data characteristics.
• Cloud-Native Architecture: Improves elasticity and distributed operation for enterprise deployments.
• Hybrid Search: Combines scalar filtering with vector similarity for enhanced result relevance.
• Component-Level Scaling: Allows specific components to scale based on demand, optimizing resource usage.

Comparison of Functionalities with Weaviate

When compared directly to Weaviate, Milvus offers several distinctions:

• Search Performance: Achieves trillions of vector searches with millisecond results, though benchmarks from Medium show Milvus at 0.9 seconds for queries compared to Weaviate's 0.12 seconds in specific implementations.
• Memory Requirements: Requires higher memory allocation, leading to increased operational costs despite faster performance in many scenarios.
• Licensing: Uses the Apache 2.0 license versus Weaviate's BSD 3-Clause License, potentially offering greater flexibility for commercial applications.
• Integration Capabilities: Seamlessly connects with deep learning frameworks like PyTorch and TensorFlow, enhancing its appeal for machine learning workflows.

Use Cases That Align Best with Milvus

Milvus excels in these specific applications:

• E-commerce Recommendation Systems: Particularly effective for product recommendations based on customer behavior patterns.
• NLP and Image/Video Analysis: Trusted by enterprise businesses for these applications, according to KDnuggets.
• Large-Scale AI Projects: Organizations like Salesforce and Walmart leverage Milvus for handling extensive datasets requiring rapid similarity searches.
• Real-time Analytics: The high throughput capabilities make it suitable for applications demanding immediate insights from streaming data.

D. Chroma

Introduction and Main Features of Chroma

Chroma is an open-source embedding database designed specifically for Large Language Model (LLM) applications. With 18,077 GitHub stars, it has gained significant traction as a developer-friendly solution for AI workflows.

Distinctive features include:

• Embedded-First Architecture: Simplifies integration into local environments, ideal for rapid development cycles.
• LangChain Integration: Supports seamless connection with LangChain in both Python and JavaScript.
• Scalability Range: Functions effectively from Python notebooks to production clusters.
• Audio Data Optimization: Particularly adept at processing audio-related vector data.
• Advanced Visualization Tools: Enhances data exploration and analysis through intuitive visual interfaces.

Performance and Usability Insights Compared to Weaviate

Chroma presents different performance characteristics when compared to Weaviate:

• Development Speed: Amax's analysis indicates Chroma excels in rapid prototyping and iterative development, though it faces scalability limitations.
• Persistence Limitations: Community feedback on Reddit highlights concerns about data persistence when self-hosted.
• Horizontal Scaling: Unlike Weaviate, Chroma currently lacks robust horizontal scaling capabilities, limiting its applicability for very large datasets.
• User Experience: Generally considered more user-friendly for AI developers, with simplified setup processes compared to Weaviate's more complex configuration requirements.

Specific Applications Where Chroma Excels

Chroma performs exceptionally well in these scenarios:

• RAG (Retrieval-Augmented Generation): Optimized specifically for enhancing large language model outputs with relevant context.
• Audio Processing Applications: The specialized handling of audio data makes it ideal for voice-based AI systems.
• Rapid AI Prototyping: Perfect for teams focused on quick iteration and proof-of-concept development.
• Educational and Research Settings: The straightforward interface and notebook compatibility make it valuable for academic and experimental applications. Each of these alternatives offers distinct advantages in specific contexts. The optimal choice depends on your project's unique requirements for performance, scalability, ease of use, and specialized functionality.

Considerations for Migrating from Weaviate

After evaluating the alternatives to Weaviate, you may decide that another vector database better suits your specific requirements. Migrating between vector databases requires careful planning and execution. This section outlines key considerations to ensure a smooth transition while minimizing disruption to your AI applications.

Tips for Transitioning to an Alternative Vector Database

The migration process involves several critical steps:

1. Create Target Collections: According to Weaviate's migration guide, begin by creating collections in your target database that match the structure of your source collections. For example, if migrating a collection called "WineReview" from Weaviate, establish a corresponding collection in your new database.
2. Set Up Client Connections: Establish connections to both your Weaviate instance and the target database. Each alternative has its own client library and connection methods that you'll need to implement.
3. Define Migration Parameters: Specify the class names and tenants involved in the migration. This includes parameters like from_class_name, to_class_name, and tenant specifications if using multi-tenancy features.
4. Implement Batch Processing: Migrate data in manageable batches rather than attempting to transfer everything at once. Weaviate's developers recommend retrieving data in batches (default 500 items) and inserting in smaller batches (default 50 items) to balance performance and reliability.
5. Handle Cross-References: Pay special attention to cross-referenced objects to ensure they're migrated in the correct order, maintaining data integrity and relationships.
6. Verify Vector Integrity: Ensure that vector embeddings transfer correctly between systems. Some databases may handle vector dimensions or normalization differently.
7. Implement Error Management: Create robust error handling to catch issues during the migration process, allowing for retries or manual intervention when needed.

Key Aspects to Assess When Switching Platforms

Before and during migration, evaluate these critical factors:

Technical Compatibility

• API Differences: According to Weaviate's client migration guide, transitioning may involve significantly different APIs. For example, when migrating to a database like Pinecone or Milvus, you'll need to adapt to their specific query patterns and data structures.
• Schema Definitions: Assess your current schema definitions in Weaviate and align them with the requirements of your new database. Some alternatives may have different approaches to schema management.
• Query Structure Adaptation: Vector searches, filtering mechanisms, and result formatting often vary between platforms. Plan for these differences in your application code.

Operational Considerations

• Performance Benchmarking: Before fully committing, conduct benchmark tests to verify that the new database meets your performance requirements. Zilliz recommends using tools like VectorDBBench to compare databases with your specific datasets.
• Downtime Planning: Schedule migrations during low-traffic periods to minimize impact on users. Consider implementing a staged migration where both systems run in parallel temporarily.
• Data Validation: After migration, validate that search results in the new system match expectations from your Weaviate implementation. Small differences in vector processing can lead to changes in result relevance.

Cost and Resource Analysis

• Infrastructure Requirements: Different databases have varying hardware demands. For instance, Reddit discussions note that Milvus requires higher memory allocation than other alternatives.
• Pricing Model Differences: Calculate the total cost of ownership. While Weaviate costs approximately $25/month on a performance-optimized instance, Pinecone starts at around $70/month, according to comparative benchmarks.
• Scaling Expenses: Project how costs will change as your data volume and query load increase. Some databases offer more cost-effective scaling than others.

The Importance of User Support and Community Engagement

The strength of community and support resources can significantly impact your migration success:

Documentation Quality

Access to comprehensive documentation simplifies the migration process. Databases like Milvus and Qdrant offer extensive documentation with specific migration guides. According to G2's competitor analysis, users consistently rate alternatives like Qdrant higher for documentation quality and support responsiveness compared to Weaviate.

Community Resources

Active communities provide valuable troubleshooting assistance during migration challenges. GitHub stars serve as one metric for community engagement:

• Milvus: 32,742 stars
• Qdrant: 22,124 stars
• Chroma: 18,077 stars
• Weaviate: 12,533 stars Larger communities typically generate more tutorials, example code, and discussion threads that can help resolve migration issues.

Support Options

Consider the support channels available for your chosen alternative:

• Pinecone offers dedicated support through enterprise plans.
• Milvus provides community forums and enterprise support through Zilliz.
• Qdrant maintains active Discord channels and GitHub discussions.
• Chroma focuses on developer-friendly documentation and community engagement. User feedback on Reddit suggests that support quality significantly impacts migration success, with many users highlighting Qdrant's responsive community as a key factor in their successful transitions.

Migrating from Weaviate to another vector database represents a significant technical undertaking. However, with proper planning, thorough testing, and attention to the specific requirements of your application, you can successfully transition to a solution that better meets your needs for performance, scalability, or specialized functionality.

Conclusion

The vector database landscape continues to evolve rapidly as AI applications become increasingly sophisticated. While Weaviate offers robust capabilities for many use cases, alternatives like Qdrant, Pinecone, Milvus, and Chroma present compelling advantages for specific scenarios and requirements.

Selecting the optimal vector database is not merely a technical decision—it's a strategic choice that directly impacts your AI application's performance, scalability, and cost-effectiveness. According to Forrester's forecast, vector database adoption is expected to triple from 6% to 18% within the next year, underscoring the growing recognition of their importance in the AI infrastructure stack.

The data from our analysis reveals distinct patterns that should guide your decision-making process:

• Performance needs should dictate your choice when response time is critical. Qdrant has demonstrated superior performance with 98-99% accuracy according to user reports, while Milvus excels at handling billions of vectors with millisecond response times.
• Deployment preferences matter significantly. If you require on-premises deployment, Weaviate, Milvus, and Qdrant offer this flexibility. For cloud-only solutions with minimal management overhead, Pinecone provides a fully-managed service.
• Budget constraints will inevitably influence your decision. Self-hosted options like Qdrant and Milvus can be more cost-effective for large-scale deployments, while Pinecone's pricing starts higher at approximately $70/month compared to Weaviate's $25/month for comparable instances.
• Community support varies considerably across platforms. Milvus leads with 32,742 GitHub stars, followed by Qdrant (22,124), Chroma (18,077), and Weaviate (12,533), as reported by Zilliz. Larger communities typically provide more resources, examples, and troubleshooting assistance.
• Specialized features may be decisive for your specific application. Chroma excels with audio data and LLM integration, while Pinecone offers superior real-time indexing for dynamic datasets. The vector database you choose today will shape your AI application's capabilities tomorrow. As Dmitry Kan's analysis suggests, the field is still evolving rapidly, with significant investments flowing into these technologies—Pinecone raised $100 million, Weaviate secured $50 million, and Qdrant obtained $35.5 million in recent funding rounds.

Our recommendation is to approach this decision methodically. Begin by identifying your non-negotiable requirements in terms of performance, scalability, and specialized features. Then, conduct small-scale proof-of-concept implementations with your actual data before committing to full migration. The investment in proper evaluation will pay dividends in the form of optimized AI applications that deliver superior results to your users.

The future of AI development is inextricably linked to the capabilities of vector databases. By selecting the right foundation for your specific needs, you position your applications to leverage the full potential of semantic search, recommendation systems, and other AI-powered features that are rapidly becoming essential in modern software.

---

🚀 Take Action Now

• Find your next profitable AI app idea validated by real data
• Unlock access to 61,988+ (and growing) validated keywords with market demand
• Explore the fastest-growing AI tools and competition
• Search our database of 2,269+ (and growing) AI applications to inform your next project

---