Community Discussions

Adding images to an existing text search with [Marqo](https://saaskart.co/software/marqo) sounds appealing. What did the migration and re-indexing involve?

Image embeddings are big. How is [Marqo](https://saaskart.co/software/marqo) handling storage and query speed for large image collections?

Using [Marqo](https://saaskart.co/software/marqo) as the retrieval layer in RAG is straightforward given built-in embeddings. How does it compare to a classic v

Out-of-the-box relevance is a Marqo selling point, but real data needs tuning. What levers in [Marqo](https://saaskart.co/software/marqo) improved your search q

For those on [Marqo](https://saaskart.co/software/marqo) Cloud, how does pricing scale with index size and query volume, and has it stayed predictable?

Running [Marqo](https://saaskart.co/software/marqo) yourself bundles inference and search in one container. What are the resource requirements and scaling consi

[Marqo](https://saaskart.co/software/marqo) supports fine-tuned models for domain-specific search. How much did a custom model improve relevance over a general

You can pair Qdrant with a separate embedding service instead of using [Marqo](https://saaskart.co/software/marqo). Which gave you more control and better relev

For e-commerce, combining images and text is powerful. How is [Marqo](https://saaskart.co/software/marqo) performing for multimodal product search versus a text

[Marqo](https://saaskart.co/software/marqo) removes the separate embedding step, but that couples model and store. Where has this integrated approach helped or

For large-scale RAG, [Vespa](https://saaskart.co/software/vespa) offers hybrid retrieval and ranking in one place. How does it compare to stitching a vector DB

Managed [Vespa](https://saaskart.co/software/vespa) removes ops but adds cost. Where did the total cost of ownership land for you versus running your own cluste

Beyond search, [Vespa](https://saaskart.co/software/vespa) powers recommendations. How are you combining user features and item embeddings for personalized rank

The ranking and schema language in [Vespa](https://saaskart.co/software/vespa) is powerful but unfamiliar. How long did it take your team to become productive?

Some vector stores struggle with frequent updates. How well does [Vespa](https://saaskart.co/software/vespa) handle high-rate real-time indexing without hurting

At very large scale, few engines keep up. How is [Vespa](https://saaskart.co/software/vespa) performing for people with hundreds of millions to billions of docu

Self-hosting [Vespa](https://saaskart.co/software/vespa) is powerful but complex. What has the operational burden been, and did teams eventually move to Vespa C

[Vespa](https://saaskart.co/software/vespa) can evaluate ONNX models at query time for ranking. How much latency does in-engine inference add, and is it worth i

Combining BM25 and vector signals is a Vespa strength. How are people tuning hybrid ranking in [Vespa](https://saaskart.co/software/vespa) to beat either signal

[Vespa](https://saaskart.co/software/vespa) does far more than vector search. For teams that only need embeddings similarity, is Vespa overkill versus a focused