Let's take a look at some of the most significant changes in Meilisearch's latest update. We’ll go over the main changes in this article, but you can also view the [full changelog](https://github.com/meilisearch/meilisearch/releases/tag/v1.3.0) on GitHub.

v1.3 is now available on [Meilisearch Cloud](https://www.meilisearch.com/cloud), including all experimental features. Upgrade your Meilisearch instance in one click without downtime.

Experimental feature: vector search
-----------------------------------

<Callout>
  Using LangChain? You can now use the [Meilisearch vector store](https://python.langchain.com/docs/integrations/vectorstores/meilisearch) to benefit from powerful search features!
</Callout>

We are excited to introduce [vector storage](https://www.meilisearch.com/docs/learn/experimental/vector_search)! Now, you can index and search documents using vector embeddings—numerical representations of data. For the first iteration of this feature, you need to use an external tool such as [Hugging Face](https://huggingface.co/), [Cohere](https://cohere.com/), or [OpenAI](https://openai.com/), to create such embeddings. Check out the [announcement post](/blog/vector-search-announcement/ to discover the extensive possibilities and potential applications of this feature.

<Callout>
  This feature is experimental, you can enable it using the new [experimental-features endpoint](https://www.meilisearch.com/docs/reference/api/experimental_features).
</Callout>

To search documents based on their vector proximity, make sure they contain a `_vectors` field:

```bash
curl -X POST -H 'content-type: application/json' \
  'localhost:7700/indexes/songs/documents' \
  --data-binary '[
      { "id": 0, "_vectors": [0, 0.8, -0.2], "title": "Across The Universe" },
      { "id": 1, "_vectors": [1, -0.2, 0], "title": "All Things Must Pass" },
      { "id": 2, "_vectors": [[0.5, 3, 1], [-0.2, 4, 6]], "title": "And Your Bird Can Sing" }
  ]'
```

Once stored, you can search your vectorized documents by using the new `vector` search parameter in combination with the [search](https://www.meilisearch.com/docs/reference/api/search) and [multi-search](https://www.meilisearch.com/docs/reference/api/multi_search) routes.

```bash
curl -X POST -H 'content-type: application/json' \
'localhost:7700/indexes/songs/search' \
--data-binary '{ "vector": [0, 1, 2] }'
```

👉 Note that you need to generate the vector in your query using a third-party tool.

The returned documents include a `_semanticScore` field indicating the semantic similarity—or relevancy—of each document with the query.

```JSON
{
  "hits": [
    { "id": 0, "_vectors": [0, 0.8, -0.2], "title": "Across The Universe", "_semanticScore": 0.6754 },
    { "id": 1, "_vectors": [1, -0.2, 0], "title": "All Things Must Pass", "_semanticScore": 0.7546 },
    { "id": 2, "_vectors": [[0.5, 3, 1], [-0.2, 4, 6]], "title": "And Your Bird Can Sing", "_semanticScore": 0.78 }
  ],
  "query": "",
  "vector": [0, 1, 2],
  "processingTimeMs": 0,
  "limit": 20,
  "offset": 0,
  "estimatedTotalHits": 2
}
```
<Callout>
  This feature is experimental and we need your help to improve it! Share feedback on [this GitHub discussion](https://github.com/meilisearch/product/discussions/677).
</Callout>

New feature: display ranking scores at search
---------------------------------------------

Use the new `showRankingScore` search parameter to see the [ranking score](https://www.meilisearch.com/docs/reference/api/search#ranking-score) of relevancy of each document:

```bash
curl \
  -X POST 'http://localhost:7700/indexes/movies/search' \
  -H 'Content-Type: application/json' \
  --data-binary '{ "q": "Batman Returns", "showRankingScore": true }'
```

Each document in the response will include a `_rankingScore` property, which represents a score between 0 and 1:

```JSON
"_rankingScore": 0.8575757575757575,
```

The higher the `_rankingScore`, the more relevant the document is.

Experimental feature: ranking score details
-------------------------------------------

For every document returned, Meilisearch provides a [detail of the ranking score](https://www.meilisearch.com/docs/learn/experimental/ranking_rule_score_details) of each ranking rule. This feature is a perfect example of [how community engagement can enrich the design process](/blog/from-ranking-to-scoring/ at Meilisearch.

<Callout>
  This feature is experimental, you can enable it using the new [experimental-features endpoint](https://www.meilisearch.com/docs/reference/api/experimental_features).
</Callout>

Set the `showRankingScoreDetails` search parameter to `true` to view the ranking score details:

```bash
curl \
  -X POST 'http://localhost:7700/indexes/movies/search' \
  -H 'Content-Type: application/json' \
  --data-binary '{ "q": "Batman Returns", "showRankingScoreDetails": true }'
```

  
The response should include the relevancy score of each rule:

```JSON
"_rankingScoreDetails": {
  "words": {
    "order": 0,
    "matchingWords": 1,
    "maxMatchingWords": 1,
    "score": 1.0
  },
  "typo": {
    "order": 1,
    "typoCount": 0,
    "maxTypoCount": 1,
    "score": 1.0
  },
  "proximity": {
    "order": 2,
    "score": 1.0
  },
  "attribute": {
    "order": 3,
    "attributesRankingOrder": 0.8,
    "attributesQueryWordOrder": 0.6363636363636364,
    "score": 0.7272727272727273
  },
  "exactness": {
    "order": 4,
    "matchType": "noExactMatch",
    "score": 0.3333333333333333
  }
}
```
<Callout>
  This feature is experimental. You can help us improve it by sharing your feedback on [this GitHub discussion](https://github.com/meilisearch/product/discussions/674).
</Callout>

New feature: search for facet values
------------------------------------

Once you have defined your [filterable attributes](https://www.meilisearch.com/docs/reference/api/settings#filterable-attributes), you can use the new endpoint `POST /indexes/{index}/facet-search` to [search for facet values](https://www.meilisearch.com/docs/learn/fine_tuning_results/faceted_search#searching-facet-values).

```bash
 curl \
  -X POST 'http://localhost:7700/indexes/movies/facet-search' \
  -H 'Content-Type: application/json'  \
  --data-binary '{
    "facetName": "genres",
    "facetQuery": "a"
  }'
```

  
As facet search is typo tolerant and supports prefix search, the above query will return de following results:

```JSON
{
    "facetHits": [
        {
            "value": "Action",
            "count": 5403
        },
        {
            "value": "Adventure",
            "count": 3020
        },
        {
            "value": "Animation",
            "count": 1969
        }
    ],
    "facetQuery": "a",
    "processingTimeMs": 0
}
```

To test this feature, check out our updated [ecommerce demo](https://ecommerce.meilisearch.com/).

  

![](https://lh5.googleusercontent.com/AvwqKTkbB0RfTYrWL4xeBUZEdzRwrEjgL0yKllbM5AZu0aXkpmO-peqP7y-5SC-O30TsAdaiXEMdOck4hiS44UvF8UFV-MwsM-mvtdRmxEeKtPowTUXQThTlqlP5yqz2p_5WohLCIgOtydEvK33TdAY)

  

New feature: sort facet values by count
---------------------------------------

By default, facets are sorted in ascending alphanumeric order. Now, you can sort facet values in descending order based on the number of matched documents containing each facet value.

To order facets by count, you can use the new `sortFacetValuesBy` property in the  [faceting index settings](https://www.meilisearch.com/docs/reference/api/settings#faceting).

To modify the order for all facets, you can use the asterisk (\*) wildcard:

```bash
 curl \
  -X PATCH 'http://localhost:7700/indexes/movies/settings/faceting \
  -H 'Content-Type: application/json'  \
  --data-binary '{
    "sortFacetValuesBy": {"*": "count"}
  }'
```

Alternatively, you have the flexibility to order a single facet by count while maintaining alphanumeric ordering for other attributes:

```json
 curl \
  -X PATCH 'http://localhost:7700/indexes/movies/settings/faceting \
  -H 'Content-Type: application/json'  \
--data-binary '{
    "sortFacetValuesBy": {"*": "alpha", "genre": "count"}
  }'
```

You can also try it in the [ecommerce demo](https://ecommerce.meilisearch.com/).

![](https://lh4.googleusercontent.com/s7qTls-ayHQpaIASqAQjNX0AnGUfc5TkZu6r83XvZBJyyyxU54qpeAXvjgofCgEbiMs72wwmVQF5m-LnLNkiQVkQFJULEvqgSYdtVh_dxGufHr8FC7zxag9PMsSXb0YbsjysSqBQuRRiSSFWcqNqyKc)

New feature: searchable attributes at query time
------------------------------------------------

With the new `attributesToSearchOn` search parameter, you can now [restrict the search](https://www.meilisearch.com/docs/reference/api/search#customize-attributes-to-search-on-at-search-time) to a subset of the [searchable attributes](https://www.meilisearch.com/docs/learn/configuration/displayed_searchable_attributes#the-searchableattributes-list) at query time. It accepts an array of strings indicating one or more document attributes:

```JSON
{
  "q": "adventure",
  "attributesToSearchOn": ["genre"]
}
```

Given the following documents, the above query will only return document with id 1.

```JSON
{
    "id": 0,
    "name": "An Adventure in Space and Time",
    "genre": ["drama"]
},
{
    "id": 1,
    "name": "A Strange and Stubborn Endurance",
    "genre": ["adventure"]
}
```

Relevancy change: attribute ranking rule
----------------------------------------

With v1.3, the [attribute ranking rule](https://www.meilisearch.com/docs/learn/core_concepts/relevancy#4-attribute) assigns higher relevancy to documents where the query words appear closer to their position in the query, compared to documents where the query words are farther from their position in the query.

Before, this rule measured relevancy based on the position of the words in the attribute rather than on their position in the search query. Documents with attributes containing the query words at the beginning of the attribute were considered more relevant.

Consider the following query: "Batman the dark knight returns" and the corresponding documents:

```JSON
{ 
    "id": 0, 
    "title": "Batman the dark knight returns" 
},
{ 
    "id": 1, 
    "title": "Dark the Batman knight returns" 
},
{ 
    "id": 2, 
    "title": "Batman returns: a retrospective", 
    "description": "The Dark knight is analyzed in this new Batman documentary" 
}
```

Prior to version 1.3, the attribute ranking rule placed document 2 at the top of the list, followed by documents 0 and 1, ranked in a tie.

Since version 1.3, the order is strictly: document 0, document 1, document 2, which provides a more natural ranking.

Other improvements
------------------

*   The `/tasks` route now shows the total number of tasks in the queue using the `total` property. It also displays the total number of tasks based on specific filters
*   Enhanced Japanese support and segmentation
*   Improvements in the Prometheus `/metrics` experimental feature

Contributors
------------

We are really grateful for all the community members who participated in this release. We would like to thank: [@vvv](https://github.com/vvv), [@jirutka](https://github.com/jirutka), [@gentcys](https://github.com/gentcys), [@cuishuang](https://github.com/cuishuang), [@0xflotus](https://github.com/0xflotus), and [@roy9495](https://github.com/roy9495) for their help with [Meilisearch](https://github.com/meilisearch/meilisearch?ref=blog.meilisearch.com).

We also want to send a special shout-out to our Meilistar, [@mosuka](https://github.com/mosuka), for his continuous Japanese improvements on [Charabia](https://github.com/meilisearch/charabia?ref=meilisearch-blog).

Conclusion
----------

And that’s it for v1.3! Remember to check the [changelog](https://github.com/meilisearch/meilisearch/releases/tag/v1.3.0) for the full release notes, and see you next time!

You can stay in the loop by subscribing to our [newsletter](https://meilisearch.us2.list-manage.com/subscribe?u=27870f7b71c908a8b359599fb&id=79582d828e&ref=blog.meilisearch.com). To learn more about Meilisearch's future and help shape it, take a look at our [roadmap](https://roadmap.meilisearch.com/?ref=blog.meilisearch.com) and come participate in our [Product Discussions](https://github.com/meilisearch/product/discussions?ref=blog.meilisearch.com).

For anything else, join our developers community on [Discord](https://discord.gg/meilisearch?ref=blog.meilisearch.com).

What’s new in v1.3?

Today we're diving into a complex, yet essential, topic: utilizing Meilisearch in a multi-tenant application. By the end of this guide, advanced and intermediate developers will have a clear understanding of how to configure Meilisearch to manage multi-tenancy effectively.

Multi-tenancy: what is it?
--------------------------

In a multi-tenant architecture, a single instance of a software application serves multiple customers or "tenants". Each tenant's data is isolated and remains invisible to other tenants.

### Multi-tenancy use cases

A typical use case for [multi-tenancy](https://www.meilisearch.com/docs/learn/security/tenant_tokens#what-is-multitenancy) is in [Software as a Service (SaaS) applications](https://saas.meilisearch.com/), where multiple companies share the same database. Each company becomes a tenant with data that should not be accessible by others. In such scenarios, you're likely already implementing this specificity in your primary data store.

Multi-tenancy with Meilisearch: the ultimate approach
-----------------------------------------------------

While it's possible to [create one index](https://www.meilisearch.com/docs/learn/core_concepts/indexes#index-creation) per company with Meilisearch, it's not recommended due to [performance](https://www.meilisearch.com/docs/learn/advanced/indexing) reasons. Meilisearch processes [tasks](https://www.meilisearch.com/docs/learn/async/asynchronous_operations) one index after another and cannot maximize the indexation throughput, as cannot process tasks on different indexes in parallel.

Instead, the best way is to put all tenants in the same index and use [Meilisearch's powerful tenant tokens](https://www.meilisearch.com/docs/learn/security/tenant_tokens#tenant-token-payload-reference) to restrict access to a specific user's scope only. For this, stored documents should have not only an [ID](https://www.meilisearch.com/docs/learn/core_concepts/primary_key) but also a company identifier to separate users by their company.

<Callout>
  Experience multi-tenancy in action through our demos:  
\- [SaaS demo](https://saas.meilisearch.com/)  
\- [Tenant tokens demo](https://tenant-token.meilisearch.com/)
</Callout>

### Preparing Meilisearch for multi-tenancy

Let's illustrate this with an example akin to a Slack application. In this application, users belong to a company, and each user can create text messages in different threads. Our goal is to create an index containing all messages but allow only people from a specific company to access their company's messages. Subsequently, we might want to perform a search applying [filters](https://www.meilisearch.com/docs/learn/fine_tuning_results/filtering) on users and threads.

![Database diagram showing the relationships between companies, users, and messages.](https://unable-actionable-car.media.strapiapp.com/DB_diagram_8e2aa6bb68.png)

First, we need to [transform our data](https://www.meilisearch.com/docs/learn/core_concepts/documents#dataset-format) to make it optimal for indexing by Meilisearch. Here's what a sample document could look like:

```JSON
{
	"id": 48735873,
	"user":{
		"id": 5749,
		"name": "bob"
	},
	"company": {
		"id": 838,
		"name": "Acme Inc"
	},
	"thread": "meilisearch-is-awesome",
	"data": "Meilisearch is so easy to use! 🔥",
	"created_at": 1543230000
}
```
<Callout>
  Learn how to [update your settings](https://www.meilisearch.com/docs/reference/api/settings#update-settings) in the documentation.
</Callout>

Next, you'll need to [create a messages index](https://www.meilisearch.com/docs/learn/core_concepts/indexes#index-creation) and add the following settings:

```JSON
{
  "displayedAttributes": [
    "id",
		"user.name",
		"thread",
		"data",
		"created_at"
  ],
  "searchableAttributes": [
    "data"
  ],
  "filterableAttributes": [
		"user.name",
		"company.id",
		"thread"
	],
  "sortableAttributes": [
		"created_at"
	]
}
```

Here's what each setting does:

*   **[`displayedAttributes`](https://www.meilisearch.com/docs/reference/api/settings#displayed-attributes)**: These are the attributes that will appear in your application's search results. We're omitting **`user.id`** and **`company.id`** for security reasons, and **`company.name`** isn't necessary as the user is already aware of their company.
*   **[`searchableAttributes`](https://www.meilisearch.com/docs/reference/api/settings#searchable-attributes)**: These are the attributes that the search function will use. In our case, only the **`data`** (or message text) is searchable.
*   **[`filterableAttributes`](https://www.meilisearch.com/docs/reference/api/settings#filterable-attributes)**: These are the attributes that you can filter by. We've included **`company.id`** for the tenant tokens, **`user.name`**, and **`thread`** for on-the-fly filtering.
*   **[`sortableAttributes`](https://www.meilisearch.com/docs/reference/api/settings#sortable-attributes)**: These are the attributes you can sort by. We've only included **`created_at`** as you might want to sort messages chronologically.

### Creating tenant tokens

With your Meilisearch configuration in place, the next step is to create a new tenant token for each company. If you don't have particular security concerns, you can omit the expiration date to simplify your onboarding process.

<Callout>
  Learn to [generate tenant tokens with an SDK](https://www.meilisearch.com/docs/learn/security/tenant_tokens#generating-tenant-tokens-with-an-sdk).
</Callout>

The crucial part of the tenant tokens is the `searchRules`. In our example, you'll add the following rule:

```JSON
{
  "messages": {
    "filter": "company.id = 838"
  }
}
```

With the generated token, your search will restrict access based on `company.id`, so users will only see messages within their company.

Store this token in your primary data store (Postgres, MySQL, etc.). This mechanism is similar to those used in tools like Stripe.

Each time a user logs into your app, load this key into local storage and use it for the searches, ensuring that each user only has access to data relevant to their company.

Conclusion
----------

[Creating a multi-tenant application using Meilisearch](/blog/multi-tenancy/)s quite straightforward once you understand how to structure your data and configure your search attributes correctly. By following this guide, you can ensure that your application provides a secure, efficient, and user-friendly search experience. If you have any question, you can join us on [Discord](https://discord.gg/meilisearch/?ref=blog.meilisearch.com).

Boost productivity and streamline development with **[Meilisearch Cloud](https://www.meilisearch.com/cloud)**. No more server deployment or manual updates. Try it **free** for 14 days, no credit card needed.

For more things Meilisearch, subscribe to our [newsletter](https://meilisearch.us2.list-manage.com/subscribe?u=27870f7b71c908a8b359599fb&id=79582d828e&ref=blog.meilisearch.com). You can learn more about our product by checking out our [roadmap](https://roadmap.meilisearch.com/?ref=blog.meilisearch.com) and participating in our [product discussions](https://github.com/meilisearch/product/discussions?ref=blog.meilisearch.com).

_Happy coding!_

Multi-tenancy guide

Welcome everyone to our latest Meilistar interview! To our faithful readers, you already know the gist of it all, but if it’s the first time reading this type of blog post, the [Meilistars](/blog/2023-contributor-program-restrospective/)re Meilisearch’s very own contributors program, populated by some of our most talented or oldest contributors.

Who is Vishal?  

-----------------

Today, we’re sitting down with Vishal Sodani, Mumbai-based freelancer. Not only is he a self-taught polyglot programmer, he’s also been building web-based products, mostly in Django, since 2011. He’s the developer behind learned.vcreation.xyz to which he integrated Meilisearch. Aside from his dev experience, Vishal is an avid reader and a chess enthusiast!

But that’s not all:

> “I love exploring Rust now and then. I built learned.vcreation.xyz (a note taking and bookmarking site) in actix-web and published my only crate for a client [https://crates.io/crates/migamake-api-cloudflare](https://crates.io/crates/migamake-api-cloudflare)”

Not unlike our very own integration manager Bruno, [whose interview you can read here](/blog/hacktoberfest-stories-from-contributor-to-manager/)ishal discovered Meilisearch at the occasion of Hacktoberfest, the yearly celebration of open source!

An October-themed adventure  

------------------------------

It is positively thrilling to know Hacktoberfest still remains a source of discovery for open-source enthusiasts year after year! Indeed, [Hacktoberfest 2022](/blog/hacktoberfest-2022-recap/)as far from Vishal’s first participation, but rather his fourth. We consider ourselves lucky to have managed to catch Vishal’s attention last October, especially considering it’s the only time of the year he’s able to contribute to various projects due to his busy schedule, but the developer returned our love tenfold:

> “I love the easy setup and usage of Meilisearch.  
> While searching for open-source projects to participate in Hacktoberfest, I unexpectedly came across Meilisearch. Having already implemented full-text search using Postgres, my curiosity piqued when I learned about Meilisearch, and I became increasingly interested in exploring it.”

This is definitely a point we take great pride in: being able to easily get into the Meiliverse and use Meilisearch right away! We’re glad Vishal got that feeling too!  
  
Coincidentally, we’d also written a blog post about Postgres ([that you can read here if you wish](/blog/postgres-full-text-search-limitations/)o we were curious to know if Vishal agreed with the content we wrote or if he’d had a different experience on his side:  

> “When it came to utilizing Postgres full-text search, I had a rudimentary understanding. Getting started with it involved delving into documentation and articles. Subsequently, I developed code in Golang to execute queries to the back-end for conducting searches. It was a trial-and-error process, but eventually, I managed to make it work. However, in contrast, using Meilisearch for the same use case was a breeze.”

Seeing as Vishal discovered Meilisearch during Hacktoberfest, we were curious to know how his contribution journey went at such a frantic time. Thankfully, he only had good things to share about our maintainers:

> “The experience was smooth. I was able to set up the repository easily. And, during my contribution I was given timely excellent guidance. So, this made my journey enjoyable.”

The battle for constant improvement isn’t over!  

--------------------------------------------------

As a final point, we were eager to hear about which features were Vishal’s favourite, and which ones could still use some work according to him:  

> “Right now my favorite features are - ease of setup, client library in golang which makes it easy to add documents from my golang web app, documentation is good which makes it easy to know the features available and their usage.  
> I think it would be good to have more complete examples in repository; though I have not looked at repo since sometime; collection of advice/notes from real-world usage shared by other users. Sometimes when I browse issues I find interesting comments by users which helps one to learn.”

Especially valuable feedback as we’re constantly working on improving our [SDK](https://www.meilisearch.com/docs/learn/what_is_meilisearch/sdks#sdks) documentation!  
  
A big thank you to Vishal for his time, it was lovely hearing about his experience and we hope his story, as well as [Bruno](/blog/hacktoberfest-stories-from-contributor-to-manager/)’s will inspire many more to join us during future events!

Vishal Sodani: POV of a Hacktoberfest contributor

[Qogita](https://www.qogita.com/) is a leading global wholesale B2B platform that is trying to revolutionize how products are discovered and procured. On a mission to simplify B2B trade, Qogita helps its customers generate higher margins and turnover while saving time with the help of its innovative technology.

**By partnering with Meilisearch, Qogita is able to ensure high exposure for sellers and increased relevancy for buyers, resulting in an excellent buyer-seller interaction experience.**

> _"Meilisearch has empowered us to deliver lightning-fast and highly relevant search results tailored to our customers' needs. Its seamless integration with our systems has positively impacted developer experience, making local environment setup, testing, and CI/CD automation significantly more efficient. Additionally, Meilisearch has proven to be a cost-effective solution, resulting in substantial savings for Qogita."_  
> **_\- Ivo Silva, Engineering Director at Qogita_**

![](https://unable-actionable-car.media.strapiapp.com/ezgif_com_video_to_gif_87c9670025.gif)

### Challenge

The wholesale industry of today, due to its size and maturity, remains highly traditional in terms of business practices and procurement methods. This means suppliers are hard to find, hard to work with, and often resistant to even taking on new customers.

Qogita aims to address this challenge by making it incredibly easy for sellers and buyers to connect through its platform, all in one place. By offering innovative solutions such as AI-powered pricing and cost optimization, tailoring product search capabilities to users’ actual needs, and providing an excellent buyer-seller interaction via search.

### Why Qogita chose Meilisearch

Prior to Meilisearch, Qogita used Algolia as their search provider and was satisfied in terms of functionality and performance. However, Qogita's team strove for further optimization when it came to both technology and costs. An evaluation was launched to assess alternatives, and a thorough comparison between Meilisearch, Typesense, and Algolia was performed.

The evaluation primarily focused on relevancy, ensuring that the search results were not significantly disrupted with Algolia's results serving as the baseline for comparison. Key searches were conducted for popular products and categories to determine the sorting order and overall performance. Additionally, the evaluation considered the handling of typos and common misspellings, which frequently occur in user searches.

The other factors that contributed to the choice included:  

1.  ******Cost-effectiveness******  
    All the alternatives performed well in the key areas of evaluation, but Meilisearch emerged as the preferred choice due to its cost-effectiveness. Since Qogita has a relatively lean DevOps team, selecting Meilisearch Cloud for production offered a desirable combination of affordability and provided services, including confidence in terms of infrastructure reliability and the high availability assurance.  
    
2.  ******Developer experience******  
    Throughout the entire internal proof-of-concept (PoC) process, the developer experience was a key aspect for evaluation. Meilisearch proved to be advantageous for Qogita as it provided a convenient Python library which simplified the migration process. Setting up containers on Meilisearch Cloud was also a straightforward task. Additionally, Meilisearch's vibrant and engaged open source community served as a significant advantage, establishing trust within Qogita's team early on.  
    
3.  ******Product cadence and roadmap******  
    Qogita’s team was impressed by the speed with which Meilisearch was able to achieve feature parity with the alternatives, despite being one of the youngest search solutions on the market. Additionally, Meilisearch presented a solid [roadmap](https://roadmap.meilisearch.com/tabs/1-under-consideration) that included several features aligned with the value envisioned by Qogita's team for its search initiatives in the near future.

### Implementation and moving to Meilisearch Cloud

Qogita initially self-hosted to evaluate if Meilisearch fulfilled its requirements in terms of development experience. Following the evaluation, production was migrated to Meilisearch Cloud, providing Qogita's team with peace of mind regarding infrastructure outsourcing. The transition from Algolia and the rollout of Meilisearch production went seamlessly, maintaining a high level of internal developer satisfaction.

### Results and Vision

Qogita highlighted higher developer satisfaction as one of the main benefits of choosing Meilisearch Cloud as their search provider, with ease of integration and out-of-the-box relevancy being key contributing factors.

The engineering department's goal was to ensure accurate data and product delivery, as well as providing the best user experience in connecting buyers and sellers through the platform. Meilisearch empowered Qogita team to deliver lightning-fast and highly relevant search results, tailored to meet the specific needs of their customers.

In terms of their vision for search efforts, Qogita expressed that the upcoming facet value search feature (coming to Meilisearch in v1.3) would enhance the user search experience even further. Facets allow customers to refine search results by multiple categories simultaneously. For a company like Qogita, which carries tens of thousands of brands in its catalog, facet search will enable the display and exposure of a greater number of brands to users, with further improved relevancy.

Globally, AI innovation is gaining prominence, and Qogita's team envisions a future that involves contextual search, meaningful search suggestions, and personalization. Meilisearch's upcoming [vector search release](/blog/vector-search-announcement/ marks an exciting venture into the AI space, aiming to fulfill these requirements and provide fast and relevant search for AI-powered applications.

Qogita simplifies B2B trade with Meilisearch Cloud

Today we’ll learn how to promote results by pinning given documents at the top of search results. This can be used in ecommerce to give more visibility to featured products or in-app search to prioritize user-specific content. We’ll use Node.js and React to implement promoted documents.

<Callout>
  Need promoted documents inside the Meilisearch engine? [Give your feedback](https://portal.productboard.com/meilisearch/1-meilisearch/c/98-promoting-documents?utm_medium=social&utm_source=portal_share)!
</Callout>

This tutorial will use a sample movie dataset. Our objective will be to have “Finding Nemo” at the top of the results when typing “funny”, “children”, or “fish”.

Folder structure
----------------

In this guide, we’ll separate the frontend code from the backend code used to configure Meilisearch.

We’ll organize our project using the following folder structure:

```json
my-app/
├─ node_modules/
├─ data/ <-- Backend code goes there
│  ├─ setup.js
│  ├─ movies.json
│  ├─ promoted-movies.json
├─ src/ <-- React code will go there
│  ├─ App.jsx
│  ├─ main.js
│  ├─ ...
├─ package.json
```

First, create a directory for our application (eg. `my-app`) or use an existing one. In our application directory, let’s create a `data` folder for the backend code.

Download the JSON dataset and store them in our new `data` folder. Download the datasets here:

[Movies](/blog/content/files/2023/07/movies.json) ("Download")

[Promoted movies](/blog/content/files/2023/07/promoted_movies.json) ("Download")

Alright. Let’s code!

Meilisearch setup
-----------------

First, we need to launch a Meilisearch instance. You can do so by following instructions in the [local installation guide](https://www.meilisearch.com/docs/learn/getting_started/installation#local-installation) or creating an account on [Meilisearch Cloud](https://www.meilisearch.com/pricing). We’ll write a setup script to configure our Meilisearch instance and seed the data.

<Callout>
  This setup script uses the JavaScript SDK, but you can use any of [Meilisearch SDKs](https://www.meilisearch.com/docs/learn/what_is_meilisearch/sdks).
</Callout>

In this tutorial, we're using [Node 18](https://nodejs.dev/en/download/). Let's install the JavaScript SDK:

```bash
npm install meilisearch
```

Then, let's create a `setup.js` file:

```javascript
// data/setup.js

import { MeiliSearch } from 'meilisearch'

// Load the datasets
import movies from './movies.json' assert { type: 'json' }
import promotedMovies from './promoted_movies.json' assert { type: 'json' }

// Load credentials from environment
const credentials = {
  host: 'your Meilisearch host',
  apiKey: 'your Meilisearch Admin API key' 
}

// Configuration
const INDEX_NAME = 'movies'
const PROMOTED_INDEX_NAME = `promoted-${INDEX_NAME}`

const searchableAttributes = ['title', 'overview', 'genre', 'release_date']
const displayedAttributes = ['id', 'title', 'overview', 'genre', 'poster', 'release_date']

const setup = async () => {
	console.log('🚀 Seeding your Meilisearch instance')
	
	const client = new MeiliSearch(credentials)
	
	// Adding searchable attributes to movies index
client.index(INDEX_NAME).updateSearchableAttributes(searchableAttributes)
	// In the promoted movies index, only `keywords` is searchable
client.index(PROMOTED_INDEX_NAME).updateSearchableAttributes(['keywords'])
	
	// Both indexes have the same visible attributes
	// `keywords` is hidden in the promoted movies index
	client
		.index(INDEX_NAME)
		.updateDisplayedAttributes(displayedAttributes)
	client
		.index(PROMOTED_INDEX_NAME)
		.updateDisplayedAttributes(displayedAttributes)
	
	// Adding documents
	await client.index(INDEX_NAME).addDocuments(movies)
	await client.index(PROMOTED_INDEX_NAME).addDocuments(promotedMovies)
	
	// Wait for tasks completion
	await watchTasks(client, INDEX_NAME)
	await watchTasks(client, PROMOTED_INDEX_NAME)
}

// Helper to watch tasks
const watchTasks = (client, uid) {
	console.log(`Watching tasks on index ${uid}`)
	const tasks = await client.index(uid).getTasks()
	console.log(`${uid} index: waiting for tasks`)
	await client.index(uid).waitForTasks(tasks)
	console.log(`${uid} index: tasks finished`)
}

await setup()
```

This script creates our two indexes. Let’s recap what it does:

*   It sets the same [displayed attributes](https://www.meilisearch.com/docs/learn/configuration/displayed_searchable_attributes#displayed-fields) for both `movies` and `promoted-movies`.
*   It sets the [searchable attributes](https://www.meilisearch.com/docs/learn/configuration/displayed_searchable_attributes#searchable-fields); only `keywords` is searchable in the `promoted-movies` index.
*   It adds documents to our Meilisearch indexes.

<Callout>
  To [optimize indexing speed](/blog/best-practices-for-faster-indexing/ always add documents _after_ configuring settings.
</Callout>

We can now run the script with Node.js:

```bash
node setup.js
```

Our Meilisearch instance is now configured and seeded. 🥳 It’s time to implement promoted results using React!

Displaying promoted results
---------------------------

First, we’ll navigate back to the root directory of our application (eg. my-app). We’ll use Vite to create the React app in this folder.

```bash
npm create vite@latest . --template react
```

If the CLI asks to remove existing files from the current directory, answer 'No'.

Then, let’s install the additional dependencies for [integrating InstantSearch with Meilisearch](/blog/instant-meilisearch/):

```bash
npm install @meilisearch/instant-meilisearch react-instantsearch-hooks-web
```

Let’s edit our App component to display search results. We want the promoted results to show up before the rest. Let’s replace the `App.jsx` boilerplate code with our own:

```javascript
// src/App.jsx

import { instantMeiliSearch } from '@meilisearch/instant-meilisearch'
import { InstantSearch, SearchBox, Hits, Index } from 'react-instantsearch-hooks-web';
   
const MEILI_HOST = 'your Meilisearch host'
const MEILI_SEARCH_KEY = 'your Meilisearch Search API key'

const meilisearchClient = instantMeiliSearch(MEILI_HOST, MEILI_SEARCH_KEY)

const App = () => (
	<div>
		<h1>Hello React</h1>
		<InstantSearch searchClient={meilisearchClient}>
			<SearchBox />
			<h2>Results</h2>
			<Index indexName='promoted-movies'>
				<Hits />
			</Index>
			<h2>More results</h2>
			<Index indexName='movies'>
				<Hits />
			</Index>
		</InstantSearch>
	</div>
);

export default App;

```

We can now run the development server using `npm run dev`. In the browser, our app should display something like this:

![HTML output of the code above](https://unable-actionable-car.media.strapiapp.com/Screenshot_2023_07_20_at_12_50_49_f0c190da8d.png)

Displaying promoted results before other results

Congrats. We’ve successfully displayed our promoted results before the rest of the results. 🎉

<Callout>
  Got stuck? Don’t hesitate to ask for help in our [Discord community](https://discord.meilisearch.com/).
</Callout>

Going further
-------------

This tutorial explored one approach for implementing promoted results. An alternative technique would be implementing document pinning in the backend—using one of [Meilisearch SDKs](https://www.meilisearch.com/docs/learn/what_is_meilisearch/sdks). This approach has the benefit of allowing to merge results in a single response (as if they were coming from a single index).

Both techniques can achieve similar results. We also plan to integrate [promoted documents](https://portal.productboard.com/meilisearch/1-meilisearch/c/98-promoting-documents?utm_medium=social&utm_source=portal_share) in the Meilisearch engine. Give your feedback on the previous link to help us prioritize it.

For more things Meilisearch, you can subscribe to our [newsletter](https://meilisearch.us2.list-manage.com/subscribe?u=27870f7b71c908a8b359599fb&id=79582d828e). You can learn more about our product by checking out the [roadmap](https://roadmap.meilisearch.com/) and participating in our [product discussions](https://github.com/meilisearch/product/discussions).

For anything else, join our developers' community on [Discord](https://discord.meilisearch.com/).

I’ll see you there.

Implementing promoted search results with React InstantSearch

Today, we’re very happy to be sitting down with another of our Japanese language experts whose help was absolutely invaluable to reach the point that we have in support of the language: Miiton!

Getting to know Miiton
----------------------

As usual, we wanted to start the interview by getting to know a bit more about Miiton:

> “Hello, I'm @miiton, and I live in Japan. While I don't have a specific area of expertise, I've been involved in a variety of fields from networking to application development. Before I got into the kind of development work I'm doing now, I spent a lot of time setting up and configuring server racks. Recently, my work has been centered around developing and maintaining an EC site, primarily for B2B transactions, using React and Go mainly.”

We were impressed with the variety of his background, which prompted us to ask how he got into the field in the first place:

> **“**I used to work setting up and configuring server racks, configuring network, server machines, storage, etc. However, with the advent of VMware and the growth of AWS and GCP in the background, I shifted to software engineering while learning programming on my own. I originally liked things like automating deployments with PowerShell, so the transition was smooth.”

Milton also shares that he initially found Meilisearch on GitHub while trying to add an Algolia-like search feature to his support documents.  
  
Is his experience similar to any of our other contributors or users? Do let us know if it’s the case!

A constant work for the improvement of Japanese support
-------------------------------------------------------

  
We then asked what, according to him, could be improved about Meilisearch at this point in time:  

> “Japanese language support and analysis functions. Japanese language support has [progressed quickly starting with PR](https://github.com/meilisearch/charabia/pull/70), but it is still not enough to handle Japanese. [](https://github.com/meilisearch/charabia/pull/70)I am impressed that @ManyTheFish actively listens to me in Discussions and on Twitter. I respect him because he understands Japanese even though he doesn't speak Japanese. [](https://github.com/meilisearch/product/discussions/532)The [analysis function](https://github.com/meilisearch/product/discussions/532) is an important indicator for us to know what users are looking for and what content we need to provide.”

  
Indeed, Many, search engine engineer at Meilisearch, has been doing his best work to tame verious languages! The least we can say about it is that it’s far from a walk in the park. If you’re curious and wish to read more on that topic, you can [find Many’s interview here](/blog/improving-meilisearchs-language-support/  
  
It is not the first time we heard that Japanese is a particularly tricky language to tackle, and we wondered if Miiton would be able to explain to us why it is the case:

> “This is because it is difficult to determine whether a "Kanji-only strings" is Japanese or Chinese. The current implementation of Meilisearch recognizes "Kanji-only strings" as Chinese, so 東京 is normalized to 东京, making it impossible for Japanese users to search for Tokyo. This has been worked around because @ManyTheFish private-release a prototype that forces Japanese, but I hope it will somehow be resolved and incorporated into the main branch.”

  
We can only imagine how bothersome the current situation is for our Japanese users, which is also why we’re so grateful for contributors such as Miiton or [Mosuka](/blog/meilistar-interview-minoru-osuka/ for their continued support in improving the language.

Milton then goes on to explain that Meilisearch is the only open-source project to which he contributed in recent years! Naturally, we were intrigued to learn what motivated him to take that first step and start contributing:

> “It made me think, "I want to use this." Because I was shocked that I could start it up without thinking, just throw in some JSON, and use it right away.”

We were genuinely curious to hear about his firsthand experience contributing to Meilisearch, especially considering it was his first time contributing to an open-source project in a very long time:

> “Contributing to Meilisearch has allowed me to reconsider the multitude of factors involved in implementing a search system. It also gave me the option of using Meilisearch for now. I am convinced that this experience will definitely be useful in my future work. It was also a good opportunity to get in touch with Rust ”

And what a surprise to learn Miiton wasn’t familiar with Rust before contributing to Meilisearch!

A further insight into Miiton’s projects  

-------------------------------------------

Finally, we asked Miiton to share more about the projects he’s been using Meilisearch with:

![Image](https://lh4.googleusercontent.com/qEIsc8BYgC7K5i5TUuLlqU3oGP4tHgWNDOuw8iGtKE4oTJeApun7gutJBUaJ63q0lR0-i1H0hOR3LxzTT2ee_-O0RVYN5RcQZ_eaKLLqjNcYk-8ULyMGRCwpxO-gDpot79Igrz5pH0bHDHz_Nz6x-y4)

> “(I'll refrain from disclosing the specific site, as I can't.) An interesting recent effort is the integration into Shopify. Shopify has a standard & easy-to-use search function, but it only supports up to 1,000 products, and beyond that, you need to consider signing up for a third-party Shopify App. Furthermore, few of these apps are optimized for Japan, and even if they were, they were expensive. Using self-hosted Meilisearch has been a good fit so far. (I'm using the Prototype version with force Japanese language support, so this is not a implementation we can recommend to everyone at this time.) I'm operating a site with roughly 10,000 products on a 2GB memory instance, and I have plenty of resources to spare. I'm also using Prometheus' Metrics (experimental implementation) to create and monitor Grafana screens like this one.”

But that’s not all! As a hobby, Milton also created a public site where he showcases the search experience of Meilisearch in Japanese. You can find the website here, so don’t hesitate to give it a look and lots of love: [https://meilisearch-example-jp.miiton.dev/](https://meilisearch-example-jp.miiton.dev/).

And aside from language support, did any feature catch his attention in particular?

> “Of course response speed and UI development experience. I like that it is InstantSearch.js compatible because of the good development experience. It was very easy to create the above demo site. I don't have to tell you how fast the response time is.”

We couldn’t be more appreciative of Miiton’s time for this interview, and for all his continued support for the improvement of the Japanese language in Meilisearch.  
  
If like Miiton you’re also fluent in a non-european language and are interested in seeing your language improved in Meilisearch, our [Charabia repo is always open to you](https://github.com/meilisearch/charabia)!

Miiton: a master of Japanese has entered the fight!

Welcome to your bi-monthly dose of Meilisearch updates.

🤩 Meilisearch v1.2 is live
---------------------------

[![](https://unable-actionable-car.media.strapiapp.com/81602999_c47b_e299_27e9_41c0dee01e68_6a600f9d34.png)](/blog/v1-2-release/)

Meilisearch v1.2 is here everybody! Don’t forget to update your current version of Meilisearch to the latest one available to enjoy all the news features such as:

*   **Enhanced document management** via filters ⚒️
*   New IS NULL and IS EMPTY **filter operators** ✨
*   **Engine improvements 🚀**
*   And more!

Check out Laurent’s blog post for an in-depth look at [what is new in v1.2](/blog/v1-2-release/?utm_source=release-post&utm_id=v1-announcement&utm_content=whats-new-in-v1.2#new-update

* * *

🐾 What to expect from v1.3
---------------------------

With all that being said, we want to pump you up for the start of a new week! Here's the list of things we're working on for v1.3:

*   Search for **facet values**
*   Displaying **ranking details** at search
*   Sort **facet values** by alphanumerical or value count order
*   Customizing **text separators**
*   Define **fields to search on** at search-time

<Callout>
  The release candidate for v1.3 is out—[try it now](https://github.com/meilisearch/meilisearch/releases/tag/v1.3.0-rc.0)!
</Callout>

We hope to have caught your attention now…

* * *

🤖 Vector search for Meilisearch
--------------------------------

Last but not least, v1.3 will also bring **Vector Search**! We’ve started experimenting with it for the first time this year and are excited to take our first steps into AI. [Read the announcement](/blog/vector-search-announcement/  
  
If you’re as curious as us about this topic, we’re delighted to announce [we’ve released a prototype](https://github.com/meilisearch/product/discussions/621#discussioncomment-6183647) that allows Meilisearch to serve as a vector store. With this prototype, Meilisearch accepts vectors in the documents, stores them, and lets users retrieve the nearest documents using the search endpoint.  
  
We’d be grateful if you wanted to try our prototype and [share your feedback with us in this Discord conversation](https://discord.com/channels/1006923006964154428/1118827065890844733/1118827065890844733).

* * *

📰 In the news
--------------

  
**Community**

🎁 [June giveaway!](https://twitter.com/meilisearch/status/1666032750187429889)

This June, we’re hosting a giveaway! To join it, submit a PR in any of our repos, have it merged by a member of our team, and share the link to it in the ⁠[giveaway channel](https://discord.com/channels/1006923006964154428/1111273670657200198) of [our Discord](https://discord.gg/meilisearch). You’ll automatically be participating in the lottery for the chance to win some exclusive Meiliswag, similar to the one you can see below!

[![](https://unable-actionable-car.media.strapiapp.com/0f8251b0_f453_1509_8ed4_4f986132fc44_6510326af9.png)](https://twitter.com/meilisearch/status/1666032750187429889)

**Product**

☁️ [Search-as-a-service Meilisearch Cloud is open to the public](/blog/meilisearch-cloud-general-availability/?utm_source=release-post&utm_id=cloud-announcement&utm_content=earch-as-a-service#meilisearch-cloud)

Now any business can benefit from Meilisearch’s exceptional developer and search experience without dealing with server management!

🧠 [Squeezing millions of documents in 128 TB of virtual memory](/blog/dynamic-virtual-address-management/?utm_source=tech-post&utm_id=virtual-memory&utm_content=Squeezing-millions-of-documents-in-128-TB-of-virtual-memory)

Curious to know more about how dynamic management of virtual memory enabled us to remove limitations in Meilisearch indexing policy? Check out this blog post!

**Success stories**

👶 [Minipouce.fr provides a top-grade customer experience with Meilisearch's open-source offering](/blog/minipouce-top-grade-customer-experience/?utm_source=customer-experience&utm_id=minipouce-customer-experience&utm_content=Minipouce-fr-provides-a-top-grade-customer-experience-with-Meilisearchs-open-source-offering)

With Meilisearch, Minipouce.fr can maintain its commitment to delivering the best possible user experience. Read about Minipouce’s efforts to make parents’ lives easier.

📚 [Bookshop.org increases search-based purchases by 43% with Meilisearch](/blog/bookshop-increases-search-based-purchases/??utm_source=customer-experience&utm_id=bookshop-customer-experience&utm_content=Bookshop-org-increases-search-based-purchases-by-43%-with-Meilisearch)

How Meilisearch helped Bookshop.org boost their sales and how we made it happen.

* * *

🌾 Made with Meilisearch
------------------------

Our featured project of the month is [1jeune1solution.gouv.fr](https://www.1jeune1solution.gouv.fr/) submitted by our Meilistar **[LiquidIT](https://github.com/LiquidITGuy)**. This platform is an French government’s initiative to help young people get into the job market. Meilisearch allows users to search through a comprehensive list of jobs to quickly find their ideal offer.

[![](https://unable-actionable-car.media.strapiapp.com/0b6c7c06_926e_4a0f_bc3d_55cabc6dfc28_0a959bbdcf.gif)](https://www.1jeune1solution.gouv.fr/stages)

The website is built with **Next.js** and the repository is [available on Github](https://github.com/DNUM-SocialGouv/1j1s-front/). The codebase gives numerous examples of how to implement Meilisearch with **React InstantSearch Hooks** using facets, custom components, and URL synchronization.

We’re always looking to promote new projects from the community. Post yours in our [#built-with-meilisearch](https://discord.gg/meilisearch) Discord channel!

* * *

👯‍♂️ Meilisearch life
----------------------

The HR team presents a rundown of our best practices for a successful offsite with a remote-first team.

[![](https://unable-actionable-car.media.strapiapp.com/9347396d_ae70_588d_0d88_567674b7bec1_85da0966b1.jpg)](/blog/organizing-a-people-first-offsite/?utm_source=hr-post&utm_id=hr-organization&utm_content=organizing-a-people-first-offsite)

We are recruiting!

*   [Account Executive](https://jobs.lever.co/meili/92e2ebdc-3023-4fbe-8dd2-c3cfc9ecd7bf)

Wishing that you pass all your CI pipelines,

_The Meilisearch team_

Meilisearch July updates

Adding documents to Meilisearch is taking longer than expected? The following tips will help you speed up the indexing process.

<Callout>
  This is a very short guide. For a more comprehensive understanding, you can read our article on [indexing best practices](/blog/best-practices-for-faster-indexing/)
</Callout>

Define searchable attributes
----------------------------

To speed up indexing, review the [searchable attributes](https://www.meilisearch.com/docs/learn/configuration/displayed_searchable_attributes#searchable-fields) list and ensure it includes only the fields you want to be checked for query word matches. This improves relevance and search speed by eliminating irrelevant data and preventing unnecessary database growth.

Utilize filterable and sortable attributes
------------------------------------------

Some fields may not contain text but can still be vital for [filtering](https://www.meilisearch.com/docs/learn/fine_tuning_results/filtering) and [sorting](https://www.meilisearch.com/docs/learn/fine_tuning_results/sorting) results. Evaluate if numeric and boolean field values can be part of the filterable or sortable attributes lists.

Review ranking rules
--------------------

Modifying [ranking rules](https://www.meilisearch.com/docs/learn/core_concepts/relevancy#ranking-rules) may trigger a reindexing process. Consider the impact and plan accordingly.

Optimize document size
----------------------

Smaller documents are processed faster. Compress your data using methods such as `br`, `deflate`, or `gzip` supported by Meilisearch. Refer to the [documentation](https://www.meilisearch.com/docs/reference/api/overview#content-encoding) for more details.

Keep up with the latest stable version
--------------------------------------

Regularly update to the [latest Meilisearch release](https://github.com/meilisearch/meilisearch) to take advantage of performance improvements that can significantly enhance indexing speed.

Conclusion
----------

By following these recommendations, you can optimize the performance of Meilisearch and enhance the overall search experience for your users. If you have any question, you can join us on [Discord](https://discord.gg/meilisearch/?ref=blog.meilisearch.com).

Boost productivity and streamline development with [**Meilisearch Cloud**](https://www.meilisearch.com/pricing). No more server deployment or manual updates. Try it **free** for 14 days, no credit card needed.

_Happy coding!_

For more things Meilisearch, subscribe to our [newsletter](https://meilisearch.us2.list-manage.com/subscribe?u=27870f7b71c908a8b359599fb&id=79582d828e&ref=blog.meilisearch.com). You can learn more about our product by checking out our [roadmap](https://roadmap.meilisearch.com/?ref=blog.meilisearch.com) and participating in our [product discussions](https://github.com/meilisearch/product/discussions?ref=blog.meilisearch.com).

Indexing optimization guide

Vector search enables efficient retrieval of objects sharing similar characteristics. This AI-powered search technique uses embedding vectors. These vectors are mathematical representations of objects generated by machine learning models (like LLMs). Starting with the 1.3 release, Meilisearch supports storing and searching vectors.

<Callout>
  Meilisearch v1.3 is out! Read the [release notes](/blog/v1-3-release/?utm_campaign=vector-search&utm_source=blog&utm_medium=announcement)
</Callout>

Powering the Future of Search
-----------------------------

Vector search might be the dawn of a new era for search. Its use cases are numerous. In [ecommerce](https://ecommerce.meilisearch.com/?utm_campaign=vector-search&utm_source=blog&utm_medium=announcement), it enables offering recommendations for similar products. It also allows building multi-modal search, like image or audio search. Building upon conversational AI technologies enables the creation of Q&A applications. Combining vector search with user-provided information like geolocation and search history can power even more contextual search experiences.

Vector search is also the foundation for [semantic search](https://en.wikipedia.org/wiki/Semantic_search), which aims to understand the query’s meaning. Conversely, traditional lexical search only matches keywords. With semantic search, a `warm clothes` query could give results like `gloves`, `coat`, and more results related to winter clothing.

Vector search unlocks a world of new capabilities for search. Take a look at what some users have already implemented:

*   [similar videos recommendations](https://github.com/adrienpoly/rubyvideo/pull/19) on a site listing Ruby conferences
*   a [documentation chatbot](https://blazy-chat.vercel.app/) proof of concept using [GPT3.5](https://platform.openai.com/docs/models/gpt-3-5) and [LangChain](https://github.com/hwchase17/langchain)

<Callout>
  Meilisearch is available as a [LangChain vector store](https://python.langchain.com/docs/integrations/vectorstores/meilisearch).
</Callout>

Getting Started with Meilisearch Vector Store
---------------------------------------------

  
Starting with v1.3, you can use Meilisearch as a vector store. Meilisearch allows you to store vector embeddings alongside your documents conveniently. You will need to create the vector embeddings using your third-party tool of choice ([Hugging Face](https://huggingface.co/), [OpenAI](https://openai.com/)).

First, spin up a Meilisearch instance. You can [install Meilisearch locally](https://www.meilisearch.com/docs/learn/getting_started/installation?utm_campaign=vector-search&utm_source=blog&utm_medium=announcement#local-installation) or create a [Meilisearch Cloud](https://www.meilisearch.com/cloud?utm_campaign=vector-search&utm_source=blog&utm_medium=announcement) account.

Then, enable the _vector store_ experimental feature:

```Bash
curl \
  -X PATCH 'http://localhost:7700/experimental-features/' \
  -H 'Content-Type: application/json'  \
  --data-binary '{
    "vectorStore": true
  }'
```
<Callout>
  This guide uses \`curl\` to make HTTP requests to communicate with Meilisearch. In practice, we recommend using our [SDKs](https://github.com/meilisearch/integration-guides) instead.
</Callout>

Meilisearch now accepts a `_vector` field in your documents. Use it to store the vector embeddings corresponding to your document in this field.

```Bash
curl -X POST -H 'content-type: application/json' \
     'localhost:7700/indexes/songs/documents' \
     --data-binary '[
         { "id": 0, "_vectors": [0, 0.8, -0.2], "title": "Across The Universe" },
         { "id": 1, "_vectors": [1, -0.2, 0], "title": "All Things Must Pass" },
         { "id": 2, "_vectors": [0.5, 3, 1], "title": "And Your Bird Can Sing" }
     ]'
```

After storing your vectorized documents in Meilisearch, you can query them using the `search` or `multi-search` route. To do this, you need to compute the vector of your query (using a third-party tool) and send it in the `vector` field.

```Bash
curl -X POST -H 'content-type: application/json' \
   'localhost:7700/indexes/songs/search' \
   --data-binary '{ "vector": [0, 1, 2] }'
```

When using vector search, returned documents include a `semanticScore` field:

```Bash
{
  "hits": [
    { "id": 0, "_vectors": [0, 0.8, -0.2], "title": "Across The Universe", "_semanticScore": 0.6754 },
    { "id": 1, "_vectors": [1, -0.2, 0], "title": "All Things Must Pass", "_semanticScore": 0.7546 },
    { "id": 2, "_vectors": [0.5, 3, 1], "title": "And Your Bird Can Sing", "_semanticScore": 0.78 }
  ],
  "query": "",
  "vector": [0, 1, 2],
  "processingTimeMs": 0,
  "limit": 20,
  "offset": 0,
  "estimatedTotalHits": 2
}

```
<Callout>
  This API is experimental. You can help us improve it by sharing your feedback in [this Github discussion](https://github.com/meilisearch/product/discussions/677?utm_campaign=vector-search&utm_source=blog&utm_medium=textlink).
</Callout>

A First Step Toward Semantic Search
-----------------------------------

Vector search is our first step towards semantic search. But our long-term goal is to provide hybrid search—combining the benefits of full-text and semantic search to provide the most relevant search experience. Clément Renault, founder and CTO of Meilisearch, shared his thoughts on Github about [exploring semantic search](https://github.com/meilisearch/meilisearch/issues/3838?utm_campaign=vector-search&utm_source=blog&utm_medium=textlink) — read for a founder’s perspective. We can’t wait to share more with you!

Drop your email below to learn more about our progress with AI-powered search. We’ll keep you posted on all updates regarding vector search & semantic search.

Get Meilisearch AI updates straight to your inbox 💌
----------------------------------------------------

\* indicates required

Email Address \*

Marketing Permissions

Please select all the ways you would like to hear from Meilisearch:

Email

You can unsubscribe at any time.

We use [Mailchimp](https://mailchimp.com/legal/terms) to send emails.

We’re excited to walk our first steps toward semantic search. We can’t wait to hear your thoughts on integrating Meilisearch as a vector store. You can give your feedback in [this Github discussion](https://github.com/meilisearch/product/discussions/677?utm_campaign=vector-search&utm_source=blog&utm_medium=textlink).

You can stay in the loop by subscribing to our [newsletter](https://meilisearch.us2.list-manage.com/subscribe?u=27870f7b71c908a8b359599fb&id=79582d828e&utm_campaign=vector-search&utm_source=blog&utm_medium=textlink). To learn more about Meilisearch's future and help shape it, look at our [roadmap](https://roadmap.meilisearch.com/?utm_campaign=vector-search&utm_source=blog&utm_medium=textlink) and participate in our [Product Discussions](https://github.com/meilisearch/product/discussions/?utm_campaign=vector-search&utm_source=blog&utm_medium=textlink).

For anything else, join our developers community on [Discord](https://discord.meilisearch.com/?utm_campaign=vector-search&utm_source=blog&utm_medium=announcement).

I’ll see you there.

Meilisearch latest news and company updates