Columns AI (formerly Columns) vs PromptLoop

Columns AI takes messy files and live data sources through ingestion, automation, and chart publishing without making you stitch together separate tools. Drive ingests CSV, Excel, PDF, and images with OCR, Flow connects warehouses and apps with plain-language transformations, and Graph publishes brand-matched visuals you can embed or export.

Most BI tools assume your data is already clean and your team knows SQL. Columns splits the problem into three steps on purpose: structure files in Drive, automate sync and transforms in Flow, then ship client-ready charts from Graph without opening a separate design tool. You pay per row processed, and the Basic plan has no monthly base fee.

The site lists 12,000+ users across business operators, consultants, marketing and RevOps teams, analysts, and freelancers. Founder Shawn Cao previously built data infrastructure at Microsoft, Facebook, and Pinterest.

Upload a CSV or Excel file with company names or URLs, tell PromptLoop what data you need, and AI agents crawl the web to fill new columns in your dataset. The platform targets go-to-market teams who want prospect lists enriched with live web signals rather than hand-researched one row at a time. Results land back in a filterable workspace you can export or sync to HubSpot and Salesforce.

Most Excel AI tools generate formulas or analyze cells you already have. PromptLoop treats each row as a research job instead. Its agents browse live sites with configurable crawl depth, return structured fields like phone numbers or pricing signals, and scale from a single test row to 50,000 inputs per job. That workflow sits closer to Clay-style list enrichment than a copilot inside one spreadsheet cell.

Growth, sales, and revenue teams use PromptLoop to qualify outbound targets, scrape competitor pages at scale, and push enriched records into HubSpot or Salesforce. Customers listed on the site include BCG, Arketa, and Kinside. Excel and Google Sheets integrations remain available for teams that want formula-style access alongside the main dataset builder.