Bain & Company is hiring in eight US metros simultaneously for research roles that assume Perplexity Enterprise and Glean fluency. McKinsey has Lilli, BCG has Deckster, Bain has Sage. But there is no public operator manual for how a mid-level consultant chains Perplexity Deep Research, Glean, Claude Projects, and NotebookLM into a research-to-synthesis-to-deck pipeline that produces MBB-grade output in a fraction of the traditional time.
This hub is for the consultant, corporate strategy analyst, or independent advisor who bills their time and lives on research leverage. It curates benchmarks comparing Perplexity Deep Research, Claude Research, ChatGPT Deep Research, and Gemini Deep Research on a real due-diligence brief. It curates the Glean vs Perplexity Enterprise decision framework for corporate deployment. It curates the NotebookLM workflow that turns 40 PDFs into a coherent narrative. And it curates the prompt patterns that generate a defensible IC memo from a research corpus. Every skill assumes you are producing output a partner will sign.
Who this hub is for
Consultants (Consultant to Principal) at MBB, Big-4 Strategy, boutique advisors; corporate strategy analysts at Fortune 500; independent consultants billing $400-$1,000/hour.
A deep research agent workflow chains multiple AI tools into a research pipeline: gather sources with Perplexity or Deep Research, extract facts with Claude or ChatGPT, synthesize into a narrative with NotebookLM, structure into a memo or deck. Instead of one tool doing everything mediocrely, each stage uses the best tool for the job. Total time savings: 60-80% versus manual research on a typical due-diligence brief.
They optimize for different things. Perplexity Deep Research is best for breadth of citations and speed. Claude Research (via Projects with web search) is best for depth of analysis and reasoning. ChatGPT Deep Research is best for structured multi-source synthesis. On a real McKinsey-style brief we ran, Perplexity was 30% faster, Claude produced the most defensible reasoning, ChatGPT produced the cleanest final structure. Use all three, weighted by task.
They solve different problems. Glean indexes your internal systems (Google Drive, Slack, Confluence, Salesforce) and answers questions from your own data. Perplexity Enterprise indexes the public web and answers external market questions. A serious research stack needs both — Glean for "what did our team decide last quarter" and Perplexity for "who is winning in the vertical."
No — but it can absorb 40-60% of a first-year's workload, which changes the shape of the pyramid. First-years are increasingly evaluated on their AI leverage: can you generate a market map in an hour, can you build a competitor tear-sheet in twenty minutes, can you draft an IC memo from a corpus of PDFs. The role is not disappearing; the bar for it is rising.
Cite the underlying source, not the AI tool. If Perplexity found the market-size stat in a Grand View Research report, cite Grand View Research. Verify every citation manually — AI tools occasionally hallucinate URLs or misattribute sources. Some firms now include a footer disclosing "This research was AI-assisted; all sources verified by [Author Name]." Ask your firm what its policy is.
For synthesizing 20-100 documents into a coherent narrative, yes — it is the best tool in class as of 2026. Its audio-overview feature is a legitimate way to prep for a Monday meeting on a Sunday drive. Its slide deck feature is still rough. Use NotebookLM for synthesis, then move to Claude or ChatGPT for polished output, then to PowerPoint or Gamma for the final deck.