Deep research as a product
Build a SaaS that produces long-form deep-research reports on any topic, monetized via subscription or per-report.
The idea, in plain language
Build a SaaS that produces long-form deep-research reports on any topic and monetise via subscription or per-report. Quoted numbers include Ryan Doser's Claude Code Skill Stack at $3,000 in 30-45 days, Matt Van Horn's 'Last 30 Days' Claude skill as a research product, Guillermo Rauch's V0 five-ideas demo (one of which is a 'deepest-research' tool) and Cloudflare's x402 payment rail pitched as enabling '1,000 AI millionaires'. Capital needs are near-zero: a domain, OpenAI/Anthropic/XAI API access and a Next.js site. The customer is a founder, executive or operator who would otherwise spend an afternoon on Perplexity Deep Research; the videos frame AI as compressing the 'last 30 days' or 'deepest research' workflow into a paid report or API call.
Where the sources agree
- AI tooling (Perplexity Deep Research, Claude Code skills, ChatGPT deep research) is now good enough to ship as a paid product rather than as a hack — every demo in this batch treats the report output as the deliverable.
- Niche framing wins: specialized SMB segments with recurring research questions are more likely to pay, while broad consumer research gets commoditized.
- Recency + source aggregation is the moat — Matt Van Horn's 'Last 30 Days' skill and Cloudflare's agent-readiness pitch both hinge on ingesting X, Reddit and the web into structured prompts.
- Packaging matters as much as the model — Claude skill stacks, custom GPTs, a V0 'deepest-research' tool, and a Cloudflare pay-per-crawl audit each repackage the same underlying APIs.
- Distribution (LinkedIn, X, Beehiiv) is the constraint — multiple videos argue a research SaaS without an audience never gets past $0 MRR.
Where they differ
- Koerner Office frames research as a Claude skill or PDF deliverable (Doser $3K/mo), MFM frames it as a $5B-founder evening brief or AI War Council, and Greg Isenberg frames it as an agent-readiness audit or LLM Vibes Radar — three distinct products.
- Some operators sell the research output per report, as Doser does, while others layer it on top of an existing product or demo workflow; pricing differs accordingly.
- Cloudflare's x402 payment rail envisions agent-to-agent monetisation at scale; the Koerner/MFM demos assume a human buyer; Isenberg demos a free-tier-for-traffic SaaS model.
- Some pitches sell to founders, such as Matt Van Horn's competitor teardowns and Isenberg's LLM Vibes Radar; others sell to enterprise, such as Cloudflare's data refinery for 100 businesses; Koerner's version sells to SMBs, creating three different buyers.
- Custom GPTs and Claude skills compete as the underlying format — the same research output ships as a custom GPT, a Claude Code skill, a Next.js site or a Slack bot depending on the operator.
What each source says
KoernerThe Beginner-Friendly Claude AI Side Hustle Nobody Talks About
Ryan Doser sells Claude skills (essentially SOPs for AI) as digital products — Claude Code Skill Stack hit $3,000 in 30-45 days, with demos of turning YouTube videos into SEO posts and building landing pages.
Watch on YouTubeIsenbergCloudflare will make 1000+ AI millionaires
Cloudflare's agent stack (pay-per-crawl, x402, agent-readiness audits, expert-archive agent tools) pitched as the platform that turns the web into resources-for-agents and creates 1,000+ AI millionaires.
Watch on YouTubeMFMYou’re One AI Idea Away From Changing Your Life
Zach's $0 to $20M in 10 months plus an AI journal, an iOS-to-Android project converter and remotely configurable onboarding flows — productised 'AI problem discovery' as the angle.
Watch on YouTubeMFMMy 6-Step Process To Find +$1,000,000 Business Ideas (#479)
Sam Parr's six-step process for finding million-dollar ideas and turning expertise into a productised service — research-as-service applied to niche selection.
Watch on YouTubeIsenbergVercel's CEO Shares 5 AI Startup Ideas So Good You’ll Quit Your Job
Guillermo Rauch demos a V0 workflow and pitches five ideas: viral AI camera app, conversational AI forms, Notion-style document tool, LLM Vibes Radar and a 'deepest-research' tool.
Watch on YouTubeIsenbergThe Claude Code Skill My Smartest Friends Use
Matt Van Horn's 'Last 30 Days' Claude Code skill aggregates X, Reddit and the web into Claude prompts — cold-email frameworks, competitor teardowns and recency-aware research as the core feature.
Watch on YouTubeMFMI put 80% of my money in the S&P after a billionaire investor told me not to
Sam and Shaan pitch a 'company brain' that ingests docs, tickets and emails into a searchable AI assistant — knowledge-base deep-research adjacent.
Watch on YouTubeIsenbergBe a 10x Vibe Coder (Claude Code + Cursor + MCP)
Chris Raroque on the Claude Code + Cursor combo for solo devs — plan mode, MCP servers, AI code review bots, dictation and deep research before implementation.
Watch on YouTubeKoernerAI Told Me to Start This Business. So I Did.
Chris validates a QuickBooks/Xero competitor idea by scraping GummySearch, Reddit, Capterra, G2 and TrustRadius; launches LazyBooks.com on Carrd + Beehiiv as the waitlist funnel.
Watch on YouTubeMFMMFM #160: How to Build a Paid Community Making $20M a Year
Tiger 21, a $20M+ revenue peer group, and Sole Savy, a sneaker community that started on Slack, illustrate paid-research communities; Shaan also walks through his own profitable community build.
Watch on YouTubeMFMWhy You Should Take a Think Week Like Bill Gates | My First Million #218
Sam and Shaan on Bill-Gates-style think weeks, niche content creators, crypto dev tools and NFT research tools — research framing as the output.
Watch on YouTubeKoernerAI Agents Are Overhyped. Use THIS Instead
HoldCo Bros argue AI agents are overhyped and chat-based deep research is the real unlock; ChatGPT vs Perplexity deep research compared, with custom GPT workflows.
Watch on YouTubeIsenbergClaude's Agent Mode was LEAKED (First Look)
Greg solo on Anthropic's agent-mode leak, Hyrox as a trend, NotebookLM's slide/infographic feature and a 'thousand people' ICP framework for pricing research output.
Watch on YouTubeMFMAre tariffs good or bad for founders?
Sam and Shaan debate tariffs and run a Medieval Times business case study — edge-finding research method more than productised deep research.
Watch on YouTubeMFMSuzy Batiz, founder @ Poo~Pourri and Supernatural
No take is available for this episode.
Watch on YouTubeTools people use for this
Personal fit
Estimate this idea against your skills
Tell us about your background and we'll estimate how each idea fits your goals, hours and constraints. It is a sorting aid, not a prediction.
Get your fit map Idea id: deep-research-as-productRelated ideas
- Competitor tracking SaaS
Build a SaaS that monitors competitors (pricing, feature changes, ads, content) and alerts subscribers weekly.
- Trend forecasting service / newsletter
Run a paid newsletter or research service that forecasts consumer, retail, and tech trends for brands and investors.
- AI readiness / implementation consulting for SMBs
Charge local businesses and SMBs a flat fee to audit their operations, propose AI tools, and oversee the rollout.
- B2B bookkeeping / accounting SaaS
Build a SaaS that automates bookkeeping, reconciliation, or month-end close for SMBs and accountants.
- ChatGPT-generated digital products
Use ChatGPT to mass-produce digital products (planners, prompts, courses) sold on Gumroad or Etsy.
- Cold email outreach agency
Run an agency that builds and sends cold email campaigns for B2B clients, monetized via per-campaign or retainer.
What members report
No member reports yet for Deep research as a product. Rates appear once 3 members have shared their numbers.
Share your numbers
Sign in to share what you made. Reports are anonymous on the page.