Microsoft Merges Copilot and Axes Deep Research—What Happens to Two Years of AI Work?

Merging Is Good, Cutting Features Is Not

On August 18, 2026, Microsoft combined Copilot and Microsoft 365 Copilot into a single application, simultaneously removing the Deep Research and Podcasts functions. Deep Research was the tool that let the author feed a collection of documents, tweak parameters repeatedly, and generate a comprehensive research report that finally felt usable.

The author discovered the removal when a notification appeared in Copilot: two days earlier he had used Deep Research to produce a 12‑page competitive analysis of OpenAI versus Google, complete with a timeline comparison chart, and shared a screenshot praising the feature. Two days later, the feature vanished.

What Exactly Did I Lose

The author relied on Deep Research for three core tasks:

Competitive analysis : Uploading a set of competitor documents, the tool automatically organized them into a timeline and comparison tables. It had previously generated reports on OpenAI and Google model release schedules that were ready to distribute after minor formatting.

Technical research : For topics like Retrieval‑Augmented Generation (RAG), Deep Research would search papers, read documents, and produce a draft comparing implementation options—work that would otherwise take the author two days of manual literature review.

Weekly report material : By feeding the week’s project documents, the tool extracted structured summaries of what was done, upcoming plans, and identified risks, producing more organized output than the author could write himself.

All three capabilities disappeared. The author notes that Deep Research was a paid feature in GPT, and Claude’s analogous functionality changed after its Cowork integration, leaving no perfect replacement.

Attempting to replicate the workflow with Claude required manually guiding the model—specifying chapter order, table formatting, and headings—resulting in roughly three times the effort compared to the original Deep Research automation.

What This Means for Ordinary Users

The merger represents a “do‑the‑math” approach: consolidating scattered functions into a single entry point reduces cognitive load but sacrifices niche, power‑user workflows. For casual users the change may be unnoticed, but for heavy users it feels like a betrayal of established habits.

The author’s colleague, who used the Podcasts feature for AI‑generated news summaries during commutes, was left without a comparable alternative when Podcasts disappeared, highlighting the loss of customized AI‑driven content.

Claude’s earlier integration of Cowork into Chat produced a similar backlash; Cowork supported multi‑person collaborative coding sessions, which became less fluid after the merge.

Both AI giants are pursuing a “super‑app” strategy—consolidating functions to simplify interfaces—yet this can alienate users who depend on specialized tools.

Claude’s analysis: “This is essentially a ‘super‑app’ strategy in AI—converging dispersed features into one entry point lowers cognitive burden, but the cost is that niche, professional users lose their accustomed workflows. The disappearance of Deep Research is invisible to casual users but disastrous for power users.”

In summary, the feature removal illustrates how free or low‑cost AI services may eventually require users to shift to alternative solutions or pay for comparable capabilities.