What this covers
A repeatable 6-stage AI research workflow — from question framing to cite-ready output — using Perplexity, NotebookLM and Deep Research together.
Who this is for
Analysts, PMs, consultants, founders and grad students who run research weekly and want a system, not a one-off prompt.
When to reach for it
When you keep redoing the same research from scratch and the output quality drifts depending on which tool you grabbed first. If you specifically want to lean on Gemini’s fresh web data and Workspace integration in stage 3, the Gemini Deep Research workflow gives a ~45-minute path to a defensible brief inside that single tool.
Step by step
Every stage ships a copy-paste prompt. Replace <...> with your real topic, then send to the named tool.
-
Write the research question on one line. If you cannot, split it into 1 primary + 2-3 sub-questions. Use this template:
Primary question: <verb + object + scope + time> Example: "How do B2B SaaS content sites convert SEO traffic to Newsletter signups in 2026" Sub-questions (≤3): 1. <sub-question 1> 2. <sub-question 2> 3. <sub-question 3> Deliverable: <e.g. "1500-word brief + 3 charts + 5 actionable recommendations"> Deadline: <e.g. "by Thursday afternoon"> Audience: <e.g. "my co-founder + 2 investors, all SEO-literate">If the primary question cannot be stated in one sentence, you cannot open AI yet — keep splitting.
-
Stage 1 — Landscape scan (Perplexity). Open perplexity.ai, new thread:
I am researching: <primary question>. Scan how experts / leading practitioners / academia map this: 1. List 5-7 main points of disagreement (not consensus points) 2. For each: position A (≤1 sentence) + position B (≤1 sentence) + 1-2 representative sources each (with links) 3. What has materially changed in the last 12 months (new data, events, tools) 4. Three "pseudo-consensus" points that are repeated often but rest on weak evidence Only use sources from 2024 or later. End with a Markdown list of all cited links.Copy every citation link to a local file
research_<topic>/sources.mdwith a date + one-line annotation. -
Stage 2 — Source synthesis (NotebookLM). Pick the 8-15 most relevant sources from stage 1, download as PDF or save as web archive. Open notebooklm.google.com → “New notebook” → upload sources. In chat:
Based only on the uploaded 8-15 sources, answer: 1. Consensus: 3-5 facts / judgments every source agrees on 2. Disagreement: explicit conflicts between sources — for each, name which sources hold which side (use file names) 3. Gaps: information none of the sources covers but is necessary to answer "<primary question>" 4. Data: every key number / growth rate / market size that appears, each with source citation Do not use information outside the uploaded sources. Every claim must cite source filename + section.Use NotebookLM’s ”⊕ Add note” to save the synthesis.
-
Stage 3 — Deep Research draft. ChatGPT (GPT-5 / o3 deep research) or Gemini Deep Research or Claude Research. Paste the stage-2 synthesis as context, then:
I have already done the landscape scan and source synthesis. Here are the findings: <paste stage-2 output> Now produce a brief based on the above + the latest material you can search for: - Length: 1500 words - Structure: 1. Executive Summary (200 words, 3 most important takeaways) 2. Current state (300 words) 3. Major disagreements & evidence (500 words, with sub-headings) 4. Data & trends (300 words + a key-numbers table) 5. 3-5 actionable recommendations for <audience> (200 words) - Every factual claim must carry an inline citation [n] - End with full citation list — link + publication date for each - No pre-2024 sources as primary evidenceIf you run stage 3 in ChatGPT specifically, pair with the ChatGPT research workflow for Project setup and the question-challenge / weak-section re-research loop.
-
Stage 4 — Citation spot-check. Open the Deep Research output. Pick 3-5 inline citations at random. For each:
- Click the link, confirm the page exists
- Cmd+F the quoted phrase in the original page, confirm it’s actually there (not invented)
- Flag “weak” if: original says the opposite / source is a blog with no underlying data / source is >2 years stale
Re-run Deep Research for the section containing any weak citation with the same prompt + “only use sources of type X”.
-
Stage 5 — Human synthesis. Deep Research gives you a fact mosaic; “what counts as news for the reader” only you can write. Open a new doc, do not paste the AI executive summary, write your own:
- “If the reader reads only 3 sentences, what should they take away?” — write those 3.
- Which 1 finding violates the reader’s current intuition? Put it first.
- 3-5 actionable recommendations: each = verb + object + how to measure done.
-
Stage 6 — Archive. Create a local folder
research_<topic>_<YYYYMMDD>/:research_saas_newsletter_2026_05_21/ ├── 00_question.md # primary + sub-questions (stage-1 template) ├── 01_perplexity_scan.md # stage-2 output + citations ├── 02_notebooklm_synth.md # stage-3 synthesis ├── 03_deep_research.md # raw + redone version ├── 04_citation_audit.md # weak citations + reruns ├── 05_my_brief.md # your hand-written brief └── prompts.md # every prompt used (next time, change placeholders)Next time you do a similar topic: duplicate the folder, change the name and
00_question.md, and you’re already 80% set up.
Recommended workflow
Question → landscape scan (Perplexity) → source synthesis (NotebookLM) → Deep Research draft → manual spot-check → human-written synthesis → archived bundle.
Common mistakes
- Skipping the one-line question step
- Letting one tool do all 6 stages
- Treating the Deep Research output as the final brief
- Never archiving — so you redo this workflow from scratch every time
FAQ
- How long does the full workflow take?: A first run takes 2-4 hours; once you have a template it drops to 60-90 minutes.
- Do I need paid tools?: No — free tiers of Perplexity + NotebookLM + ChatGPT cover most workflows. Pro pays off only when research is daily.