{
"$type": "site.standard.document",
"bskyPostRef": {
"cid": "bafyreiba5asuqdthvtlchz2gwdgeit7bhstiwllpjsfkhyjni7iems5z7q",
"uri": "at://did:plc:pgryn3ephfd2xgft23qokfzt/app.bsky.feed.post/3mk3mw4qj65w2"
},
"path": "/t/google-deep-research-went-api-accessible-with-web-off-mode-what-are-you-building-with-it/175462#post_1",
"publishedAt": "2026-04-22T12:48:53.000Z",
"site": "https://discuss.huggingface.co",
"textContent": "Google just dropped a significant Deep Research upgrade: collaborative planning, multi-tool orchestration (MCP servers, Code Execution, File Search), multimodal inputs (audio, video, PDFs, CSVs), web-off mode for private data grounding, and Gemini API access in public preview.\n\nThe combination I find most interesting for builders: web-off mode + MCP server integration + API access.\n\nThat means you can:\n\n * Build a research workflow grounded exclusively in internal company data\n * Connect custom data sources via MCP\n * Call it programmatically from your own application\n\n\n\nWhich opens up a category of use cases that previously required custom LLM orchestration: clinical literature review, competitive intelligence pipelines, legal due diligence tools, internal knowledge synthesis.\n\nCurious what use cases people here are thinking about. Are you looking at this as:\n\n * End-user tooling (just using Deep Research directly)\n * Developer primitive (building applications on top of the API)\n * Enterprise workflow integration\n\n\n\nAlso: the Code Execution tool being available within a research run is interesting — has anyone tested what that looks like in practice for data-heavy research tasks?",
"title": "Google Deep Research went API-accessible with web-off mode — what are you building with it?"
}