Deep Research

Overview

Deep Research is an AI research capability that packages multi-step information gathering, synthesis, and output generation into a structured workflow rather than a single prompt-response interaction. In the newsletter coverage, it appears both as a product feature inside Perplexity Computer and as an API-enabled workflow pattern in the Gemini API and Gemini Interactions API, supporting multimodal inputs, planning, web research, and rich output generation.

For AI Product Managers, Deep Research matters because it represents a shift from standalone model access to embedded agent skills. Instead of asking users to manually orchestrate searches, uploads, synthesis, and revisions across tools, Deep Research wraps those steps into a reusable capability inside an agent or application. This has implications for product packaging, UX design, workflow reliability, and differentiation: the value is not just the base model, but how research is operationalized into repeatable user outcomes.

Key Developments

• 2026-01-07 — Phil Schmid shared that the Gemini Interactions API (beta) added Deep Research support for multimodal inputs, including images, PDFs, CSVs, and custom data.
• 2026-04-22 — Sundar Pichai announced upgrades to Deep Research in the Gemini API, including improved quality, MCP support, and native chart/infographic generation.
• 2026-04-25 — Philipp Schmid launched collaborative planning for Gemini API Deep Research, adding a `collaborative_planning` flag so users can request and refine draft research outlines.
• 2026-04-30 — Philipp Schmid published a developer guide for building and running Deep Research workflows with the Gemini API, covering setup, workflow construction, and execution of deep research queries.
• 2026-06-12 — Aravind Srinivas announced that Perplexity Computer now embeds Deep Research natively as a built-in skill in its agent harness, enabling seamless advanced research without mode switching.

Relevance to AI PMs

• Design research as a workflow, not a prompt. Deep Research shows that users want planning, search, ingestion, synthesis, and output formatting bundled into one experience. PMs should define the full research job-to-be-done and identify where automation meaningfully reduces user effort.
• Package advanced capabilities into agent skills. The Perplexity Computer mention is important because it frames research as an embedded capability inside a broader agent system. PMs can apply this pattern by turning complex tasks into modular skills that can be invoked contextually rather than exposed as isolated features.
• Prioritize controllability and artifact quality. Features like collaborative planning, multimodal input support, and chart generation indicate that users need more than raw answers—they need steerable workflows and presentation-ready outputs. PMs should evaluate products on plan editing, source handling, and deliverable quality, not just model intelligence.

Related

• perplexity-computer — The clearest example in these mentions of Deep Research becoming a built-in agent skill rather than a separate mode.
• gemini-api — A major platform where Deep Research evolved through quality improvements, MCP support, collaborative planning, and workflow documentation.
• gemini-interactions-api — Early signal that Deep Research would support multimodal inputs and custom data in API-driven experiences.
• phil-schmid / philipp-schmid — Key developer-facing evangelist and educator for how to build with Deep Research in Gemini tooling.
• sundar-pichai — Announced notable product upgrades that positioned Deep Research as a more capable platform feature.
• aravind-srinivas — Connected Deep Research to Perplexity Computer’s agent harness and embedded-skill packaging.
• claude, cgp, genai — Referenced as AI tools or categories where “deep research mode” helps users search, summarize, and generate outputs across sources.
• andrew-ng — Helped frame Deep Research as a practical power-user workflow pattern through educational content.