Poly Revamps Cloud File Storage With AI Search

Now Poly, which has come out of Y Combinator, is back with a more focused pitch to that end user — and creators in general: cloud storage designed from the ground up for AI-native search.

The company is launching a web and Mac app to allow users to collate documents, media and links, before asking them questions in plain English and receiving an instant answer or summary. The hook is generous as well — 100GB on the free tier and an AI-first experience where folders and file names are optional.

What Poly Offers: AI-Native Search Across Your Files

Unlike federated search tools that spider across multiple silos, Poly invites you to upload directly into its cloud, where it will index text files, PDFs, Office documents, images, and audio and video, as well as web pages if you provide URLs. You can tag assets, ask questions of your whole library, pull out insights, generate summaries or translations, and let it auto-organize as you go by making folders or renaming files.

Shared drives are baked in: teams can pool files, then use AI to search the corpus — useful for research sprints, client handoffs, or onboarding packs. In tests, users were able to paste in a YouTube link and get a short summary — just one of Poly’s not-so-little demonstrations of its “bring context, get answers” modus operandi early on.

Pricing and Storage Limits for Free and Paid Plans

Poly’s free 100GB offering is among the most aggressive in a market that has narrowed its storage ceiling over time. By comparison, Google Drive’s free plan gives you 15GB of space across its various services, Dropbox Basic offers you 2GB, and OneDrive gives you 5GB; Box’s personal option generally comes with about 10GB. Poly has a paid plan at $10 per month for 2TB, aimed more at active project work than cold storage.

There’s no automatic photo sync so far, but if that rolls out in addition to AI-led curation, Poly could feasibly take on consumer media libraries. At least for now, most early adopters are treating the drive as a working drive for projects that involve heavy research and content creation.

How Poly Compares With Incumbent Cloud Storage Rivals

Poly is competing in a field where incumbents already brandish AI search, in the form of Google Drive and Dropbox’s AI features. The part that distinguishes it is one of intent: Poly is a storage offering built with retrieval and understanding, rather than just syncing and sharing, as core principles. It adapts the “gather and reason” workflow common to tools like Google’s NotebookLM — in which you grab contents, ask questions and develop insights — with stronger file-organization capabilities. But Poly does not currently offer live web knowledge or media generation over and above summaries.

In brief, it’s looking to be where you put things so that you can find the answer to them down the line, not just where your stuff gets synced because you need it elsewhere.

Roadmap and Integrations Planned Across Apps and APIs

Poly is currently available on the web and for Mac, with a Windows app in development. The company is currently testing features like web search, report generation with stylish layouts, a built-in text and Markdown editor, custom metadata items and the option to paste in links from Google Docs that become first-class, searchable objects. Poly says spreadsheet analysis is coming with the help of AI agents that will be able to do calculations and explain results.

For developer and power-user workflows, Poly already has a Model Context Protocol server API that is accessible from tools such as ChatGPT or Cursor. Although it doesn’t connect directly with third-party drives at this stage, the team says its adoption of virtual file references means importing from other services should be possible. Direct sharing of individual files and folders outside of shared drives is also on the roadmap.

Who Poly Targets: Generative AI Creators and Teams

Poly is focusing on generative AI–native creators and knowledge workers who spend an outsized amount of time wrangling assets — think researchers building briefs, podcasters and video teams working with show transcripts or source media, or service leads pulling themes from customer calls. It’s not necessarily about deep archival data, according to the company, which means this is a solution best suited for active projects that are more likely to benefit from conversational retrieval.

Backstory and Funding Behind Poly’s Strategic Pivot

Poly’s relaunch comes amid a strategic pivot. The startup was founded in 2022 by Abhay Agarwal and Sam Young, and started out generating 3D assets based on prompts. With the generative imagery market exploding with well-funded competitors, Poly wound down that product in 2023 and rebuilt in stealth around AI-first file organization. Young has since left the company. The reset was led by Agarwal, who has a research background in vision AI for accessibility.

The company has raised $8 million in seed funding led by Felicis, with Bloomberg Beta, NextView, Figma Ventures, AI Grant, Wing Ventures and MVP Ventures also participating. That amount includes a previous $3.9 million round in 2022.

Why It Matters: Closing the File Search and Findability Gap

Information workers routinely squander time hunting for information. McKinsey estimates that workers spend on average about 19% of their week searching for and gathering information. By bringing a level of semantic understanding to jumbled, mixed media and messy file systems, AI turns the “findability gap” into something that can actually be solved — and used as a competitive edge by leveraging products that solve it without carry-along overhead.

Poly is betting that search quality, not sync coverage, will determine the next generation of storage tools. It wants to put pressure on incumbent clouds on value, with a precedent-setting level of included free storage and a roadmap aimed at the analyst and collaboration crowd while etching out its retrieval-first niche. If it can ship those promised integrations, and maintain that accuracy in answers across formats, Poly could be a default for teams who measure output in insights, not megabytes.