You decide what goes in.
TinySet shows you the columns before it builds. Change them, add your own or drop what you don’t need.
/datasets
The dataset builder
for any kind of data.
TinySet builds the dataset that doesn’t exist yet, on any topic, from one simple prompt.
Data centers under construction in the US, with operator, location and planned capacity
Build
Your agent
7 columns approvedprimary key · Project URL
Project URLprimary keyurl
Project namestring
Operatorstring
Locationstring
Planned capacity (MW)number
Expected completion datedate
Under constructionfilterboolean
Build dataset
TinySet· Follow the AI data center buildout
| Project name | Operator | Planned capacity (MW) | |
|---|---|---|---|
| 1 | T5@Atlanta III Bldg 1 | T5 Data Centers | – |
| 2 | IREN: Horizon 2 Data Center | IREN | 50 |
| 3 | STACK Phoenix 01 Campus | STACK Infrastructure | 225 |
| 4 | Skybox: Hutto 3 Data Center | Skybox Datacenters (with Prologis) | – |
| 5 | PowerHouse: Irving Campus | PowerHouse Data Centers | 201 |
| 6 | Corscale Gainesville 3 | Corscale | 54 |
148 rowsempty cells had no source
148 projects, each with operator, location and planned capacity.
Follow the AI data center buildoutCSV dataset · 148 rowsDownload
Trusted by
Features
Custom datasets, for data that is scattered across many sites.
Goes wide across the public web.
TinySet searches across many public sites and sources, so you get far more than the first dozen rows.
Your topic, your columns.
This is a custom dataset. You choose the subject and the fields, so it is built around your question instead of a fixed set of company or people fields.
Evidence and gaps stay visible.
A cell is filled only from a page TinySet read during the build, and left empty otherwise. Add source URLs and dates as columns so reviewers can check any value.
Use cases
What you can build with TinySet.
Use TinySet when you need the same fields across many companies, projects, places or products, and the data is spread across sites with no API and no download button.
Prompt
Humanoid robotics companies, with funding, robot name and stage
What came back| Company name | Robot name | Funding amount (USD) |
|---|---|---|
| 1X Technologies | NEO | – |
| Addverb | Elixis | – |
| AEI Robot | Alice | 9700000 |
| AgiBot | A2 / X2 | 86000000 |
Size up the humanoid robotics raceCSV dataset · 115 rowsDownload
Use it anywhere
A dataset builder your agents can call.
Use TinySet from 
Claude Code, 
Cursor, 
Codex, 
Grok and the assistants your team already runs.
One command. Connect your agent to TinySet through MCP, the CLI or the SDK.
Add the TinyFish MCP server at https://agent.tinyfish.chat/mcp and help me sign in. Then use TinySet to build this dataset: Small modular reactor projects worldwide, with developer, country, capacity and status.
Copy and paste this into your AI agent.
No code. Run it where you already work.
Add TinyFish to ChatGPTInstall the plugin, then ask for a dataset in any chat.- Add TinyFish to ClaudeConnect in one tap, then ask for a dataset in any chat.
Open the PlaygroundDescribe the rows you need and watch the table fill.
Signs in with your TinyFish account. No API key needed.
Also works with
Python SDK
n8n
Skills
Hermes Agent
Dify
OpenClaw
LangChain
TypeScript SDK
Codex
REST API
CLICursorClaude Code
MCPPython SDKn8nSkillsHermes AgentDifyOpenClawLangChainTypeScript SDKCodexREST APICLICursorClaude CodeMCPSkillsOpenClawTypeScript SDKDifyLangChainn8nPython SDKHermes AgentClaude CodeCursorMCPCLIREST APICodexSkillsOpenClawTypeScript SDKDifyLangChainn8nPython SDKHermes AgentClaude CodeCursorMCPCLIREST APICodexDifySkillsn8nPython SDKOpenClawTypeScript SDKHermes AgentLangChainCLICodexCursorREST APIMCPClaude CodeDifySkillsn8nPython SDKOpenClawTypeScript SDKHermes AgentLangChainCLICodexCursorREST APIMCPClaude Code
Pricing
Simple pricing for datasets.
$0.10
Per row
Paid from your Wallet. One build collects up to 250 rows across up to 64 columns. Need more than 250 rows? Talk to Sales.
FAQ
Questions, answered.
What is TinySet, and is it an AI dataset builder?
TinySet is an AI dataset builder that creates live web datasets. Describe what each row is, the scope and the fields you need, and it collects a custom dataset from public web pages.
When should I use TinySet instead of TinyResearch?
Use TinySet when you know the columns you want and need many rows with the same fields. Use TinyResearch when you have a question that needs a written, cited report.
How is TinySet different from web scraping?
A scraper pulls a page you point it at. TinySet starts from a description, finds the matching entities across many sites, and fills the same fields for each one.
How is it different from asking a chatbot for a table?
A chat assistant usually stops at a dozen rows and needs several prompts to get columns and keys right. TinySet is built to collect up to 250 rows in one build, with columns you review first.
Can TinySet enrich a list I already have?
No. TinySet builds a new dataset from your description.
Where does the data come from?
Public web pages that TinySet's agents open during the build. It does not collect from behind logins or paywalls.
Does the dataset stay current?
A dataset is a snapshot of the web at the time you build it. You can add more rows to it or reset it and rebuild.
How many rows can I build?
A build targets 1 to 250 rows and 1 to 64 columns, and typically takes 5 to 20 minutes. It can return fewer rows than you asked for, depending on what is on the web. Need more than 250 rows? Talk to Sales and we can set it up.
How is TinySet priced?
10 cents per row, drawn from your Wallet.
Build any dataset from the web.
Use TinySet when the data you need is public, spread across many sites, and missing in one place.