Building app search without running infrastructure

If you want app search without operating infrastructure, the main goal is usually: don’t run your own search cluster.

Here are the common approaches, from most “fully managed” to most lightweight:

1) Use a hosted search service

Best if you want full-text search, typo tolerance, filtering, ranking, analytics, etc.

Popular options:

• Algolia — very easy to integrate, great developer experience, fast relevance tuning
• Meilisearch Cloud — simpler/open-source-friendly search with hosted option
• Typesense Cloud — similar to Meilisearch, good for app search
• Elastic Cloud — powerful, more complex
• Azure AI Search / AWS OpenSearch Service — managed by cloud vendor

Pros

• No servers to manage
• Scaling, replication, backups handled for you
• Fast to ship

Cons

• Costs can grow with data/query volume
• Vendor lock-in
• Some services are expensive for large datasets

---

2) Use your database’s built-in search

If your app is smaller or search needs are modest, this is often the simplest approach.

Examples:

• Postgres full-text search
• Postgres trigram search
• SQLite FTS5
• MongoDB text search (less common for app search, but usable)

Pros

• No extra infrastructure
• One data source
• Easy consistency with your primary records

Cons

• Less powerful relevance tuning
• Faceting, typo tolerance, and ranking can be limited
• Search performance may degrade as complexity grows

---

3) Use an embedded/local search engine with a managed platform

If you want no infrastructure but still need search-specific capabilities:

• Lunr.js / MiniSearch for client-side or small datasets
• SQLite FTS5 if you already have SQLite
• Edge/local indexing patterns with serverless functions

Good for:

• Static sites
• Small catalogs
• Offline-capable apps

---

4) Serverless search pattern

You can build search on:

• Serverless functions
• Managed DB
• Managed vector/text search

Typical architecture:

• App writes to primary DB
• DB change feeds or background jobs update search index
• Queries go to managed search service
• No always-on app infrastructure

This reduces ops, but you still need some glue for indexing.

---

Recommendation by app type

If you want the fastest path:

Algolia or Typesense Cloud

If you want low cost and simplicity:

Postgres full-text search

If you want open-source feel with managed hosting:

Meilisearch Cloud or Typesense Cloud

If your search is mostly “find records”:

Database search is often enough

---

Practical choice guide

Ask:

1. How many documents?
2. Do you need typo tolerance?
3. Do you need filters/facets?
4. Do you need autocomplete?
5. How much can search cost?
6. Do you need relevance tuning?

Rule of thumb

• < 100k records and simple search: Postgres/SQLite FTS
• Need app-quality search UX: hosted search service
• Large, complex, enterprise use case: Algolia/Elastic Cloud/OpenSearch

---

A simple architecture

For a typical app:

• Primary DB: Postgres
• Search: Algolia / Meilisearch Cloud / Typesense Cloud
• Sync: background job or webhook on insert/update/delete
• Frontend: search box hits hosted search API directly or through your backend

This gives you search without running search servers yourself.

If you want, I can give you:

• a recommendation matrix,
• a cost comparison, or
• a reference architecture for your specific app stack.