{AI Marketing Playbook}

← All examples

GPT Researcher

Research automationBreakout · Distribution-by-integration

Overview

An open-source autonomous "deep research" agent that, given a topic, plans sub-questions, fires parallel web searches, scrapes and summarizes 20+ sources, and writes a cited report. Created by Assaf Elovic; later became the open-source seed of the company Tavily. Current scale: 27,723 GitHub stars, 3,739 forks, 172 watchers as of June 15, 2026 (GitHub API). It is widely described as the first and most-starred open-source deep-research agent (their claim / commonly repeated).

First public appearance

The repo went public on GitHub as researchGPT (later renamed gpt-researcher) with the initial commit on May 12, 2023; the first tagged release, "Initial Alpha" (v1.0.0), shipped July 9, 2023, and the earliest Wayback capture (July 10, 2023) shows just 20 stars (Wayback, 2023-07-10). The format was an organic GitHub open-source drop, not a Show HN or Product Hunt launch. The README led with the KSP: "GPT Researcher is an autonomous agent designed for comprehensive online research on a variety of tasks," positioned explicitly against AutoGPT: "Inspired by AutoGPT and the recent Plan-and-Solve paper, GPT Researcher addresses issues of speed and determinism, offering a more stable performance and increased speed through parallelized agent work." The formal narrative launch came two months later via two same-day posts on July 25, 2023: a personal-site blog, "Introducing GPT Researcher, The Future of Online Research," and a Better Programming (Medium) article, "How I Built an Autonomous AI Agent for Online Research." Both opened with the same hook: "After AutoGPT was published, I immediately took it for a spin. The first use case that came to mind was autonomous online research," then sold the fix to AutoGPT's failures (infinite loops, constant human intervention, lost progress, non-completion) with hero numbers: a research task in roughly 3 minutes (their claim of ~85% faster than AutoGPT), 100% completion via deterministic planning, 1,200+ word reports (assafelovic.com blog).

Launch sequence

  • May 12, 2023
    Initial commit, repo created as researchGPT (Wayback shows "Initial commit, May 12, 2023"). Tavily.com already listed as the project homepage even at the 20-star stage, and Rotem Weiss (rotemweiss57) is a co-contributor from the start.
  • July 9, 2023
    v1.0.0 "Initial Alpha" release.
  • July 10, 2023
    20 stars, 0 forks; README badges already wired for Discord, GitHub stars, and Twitter (@assaf_elovic) (Wayback). Same day, the original X launch tweet went out: "Introducing 'GPT-Researcher'!" with a ~59-second video demo, crediting "Inspired by @Auto_GPT, @babyAGI_ and @langchain" and listing the KSPs (one prompt -> unbiased in-depth report, aggregates 20+ web sources, web UI, open source). Reach was modest: 18 likes, 6 retweets, 7 bookmarks, ~1,606 views (tweet, Jul 10 2023). Confirms the founder's own X reach was small and not the launch driver.
  • July 25, 2023
    Dual launch essays (personal blog + Better Programming/Medium), both leading with the AutoGPT-fix angle and the 3-minute / 100%-completion hero stats (blog, Better Programming). On Jul 27 Elovic tweeted thanks to LangChain for sharing the project in their blog ("AI agents are still in their infancy... more deterministic strategies are currently required"): 7 likes, ~381 views (tweet, Jul 27 2023).
  • Mid-Aug 2023
    Steep organic GitHub ramp. By the August 17, 2023 capture the repo was at 3.4k stars, 360 forks, 12 contributors, with v0.0.4 shipped Aug 14 (Wayback, 2023-08-17).
  • August 13, 2023
    LangChain published "GPT Researcher x LangChain," calling it "the leading open-source implementation of a research assistant" and announcing a LangChain OpenAI-adapter + LangSmith integration (LangChain blog). This is the key external endorsement and it lands exactly inside the star ramp.
  • August 14, 2023
    Elovic's tweet announcing the LangChain collaboration ("Excited to share our collaboration with @langchain to build the leading autonomous AI researcher... use any of 10+ supported LLMs," shouting out @hwchase17) became the standout of the entire launch window: 144 likes, 30 retweets, 3 quotes, 110 bookmarks, ~35,638 views (tweet, Aug 14 2023). That is ~22x the view count of the Jul 10 launch tweet, and it sits squarely inside the steep star ramp (3.4k stars by Aug 17), confirming the LangChain tie, not the founder's solo launch, carried the reach. A follow-on Aug 24 tweet on LangChain enabling open-source LLMs (LLaMA via Ollama, crediting @RLanceMartin) drew 32 likes / ~5,381 views (tweet, Aug 24 2023).
  • Sometime in 2023
    Product Hunt launch under the tagline "Autonomous agent designed for comprehensive online research," which reached only #13 product of the day with 117 upvotes (modest, not a breakout) (Product Hunt).
  • Nov 22, 2023
    5.6k stars, 635 forks, 20 contributors; docs had moved to docs.tavily.com, README now claimed "~3 minutes to complete, and costs ~$0.1" per research, and a Chinese README translation had been merged (internationalization) (Wayback, 2023-11-22).
  • Late 2023
    Tavily co-founded (Rotem Weiss CEO, Assaf Elovic) as the commercial search-API spinout, with GPT Researcher as its open-source flywheel (Crunchbase, Calcalist/Ctech).
  • 2024
    Project rebrands its own home from tavily.com to gptr.dev (own domain + docs.gptr.dev), relicenses MIT to Apache-2.0, ships a LangGraph-based multi-agent assistant (inspired by the STORM paper), PyPI package, Docker image, and Colab. By the Aug 25, 2024 capture: 13.7k stars, 1.8k forks, 66 contributors, cost reduced to ~$0.005/research with gpt-4o-mini/gpt-4o, and a Star History chart embedded in the README (Wayback, 2024-08-25). Notably, LangChain founder Harrison Chase (hwchase17) and core maintainer Bagatur (baskaryan) appear in the contributor list, confirming deep LangChain-ecosystem ties.
  • Aug 13, 2024
    A belated "Show HN: GPT Researcher" (by assafe) flopped at 4 points / 0 comments, despite the project already having 13k stars and 5k+ Discord members (HN item 41234914). Confirms HN was never the growth channel.
  • Aug 6, 2025
    Tavily (the spinout) announced $25M ($20M Series A led by Insight Partners, plus Alpha Wave Global and seed investors incl. Khosla Ventures), riding GPT Researcher's open-source brand into "search for AI agents" (Tavily blog, SiliconANGLE).
  • June 2026
    27.7k stars and still actively maintained (last push May 28, 2026) (GitHub API).

Channels & accounts

GitHub
assafelovic/gpt-researcher, 27,723 stars / 3,739 forks / 172 watchers (Jun 2026). The primary owned channel and the entire growth engine.
Own site / landing
gptr.dev (moved here in 2024 from tavily.com as homepage).
Docs-as-SEO
docs.gptr.dev (earlier docs.tavily.com); multilingual READMEs (English, Chinese, Japanese).
Discord
community server (5,000+ members per the Aug 2024 Show HN, their claim); invite linked from README.
PyPI
gpt-researcher pip package (the README pitches a 6-line code snippet to run research programmatically). Current install volume is substantial: ~105,138 downloads in the last month (~20,878/week, ~3,701/day) per pypistats.org (Jun 2026), confirming PyPI is a live, high-volume distribution surface and not a vanity listing.
Docker Hub
gptresearcher/gpt-researcher image; Google Colab notebook for one-click trial.
X/Twitter
founder account @assaf_elovic (created Apr 23, 2022; only 3,763 followers as of Jun 2026); project amplified through @tavilyai. README always carried a Twitter-follow badge.
Blog/Medium
Assaf Elovic on Medium (~2.3K followers); personal site assafelovic.com.
Product Hunt
GPT Researcher listing.
YouTube/Loom
install + live-demo Loom videos linked from the README; demo .mp4 embedded directly in the README.

Amplification & KOLs

LangChain (earned, peer-ecosystem)
the single biggest named amplifier. LangChain's official blog, "GPT Researcher x LangChain" (Aug 13, 2023), opened with "Here at LangChain we think that web research is a fantastic use case for LLMs" and named GPT Researcher "the leading open-source implementation of a research assistant" (LangChain blog). LangChain founder Harrison Chase (hwchase17) and core maintainer Bagatur (baskaryan) later show up as contributors to the repo, deepening the tie (Wayback, 2024-08-25).
AutoGPT wave (organic)
GPT Researcher rode the mid-2023 autonomous-agent / AutoGPT hype by positioning itself explicitly as the reliable, deterministic AutoGPT-for-research; it surfaced repeatedly in "build an autonomous researcher with LangChain/GPT" tutorial content (e.g. AI Jason, Toolify) as the reference implementation.
Tutorial / educator ecosystem (earned)
appears in third-party LangChain tutorials, boilerplate repos (lablab-ai), and "open-source deep research" roundups, which fed steady developer discovery.
No evidence of paid promotion, paid influencers, or rival-founder feuds. Amplification was organic plus the one big earned LangChain endorsement. There is no evidence of a famous-founder boost: Elovic's personal reach was tiny (~2.3K Medium, ~3.7K X) at launch.

Traction inflection

The breakout was organic GitHub virality in the mid-2023 AutoGPT autonomous-agent wave, with the LangChain "GPT Researcher x LangChain" blog (Aug 13, 2023) as the catalyst that crystallized it. Evidence: the star curve goes from 20 (Jul 10, 2023) to 3.4k (Aug 17, 2023), roughly 5 weeks, then 5.6k by Nov 2023 and 13.7k by Aug 2024. The LangChain endorsement, naming it "the leading open-source implementation of a research assistant," lands squarely inside that vertical ramp (Aug 13 vs 3.4k stars by Aug 17). The two July 25 launch essays primed the narrative ("the AutoGPT that actually finishes, in 3 minutes"), but the curve was already climbing steeply by the time LangChain posted, so LangChain accelerated and legitimized rather than single-handedly started it. What it was NOT: Hacker News (Show HN flopped at 4 points, and that was a year later) and Product Hunt (only #13 of the day, 117 votes) were both non-events. Confidence: high (upgraded from medium-high). Beyond the star-curve shape and the dated LangChain endorsement, the X engagement data now measures the split directly: the founder's solo Jul 10 launch tweet drew only ~1,606 views, whereas the Aug 14 LangChain-collaboration tweet drew ~35,638 views (~22x), 144 likes and 110 bookmarks, and lands inside the same window the stars went vertical (3.4k by Aug 17). That confirms LangChain's involvement (the blog plus the collab announcement) carried the reach, while the AutoGPT-wave framing primed discovery. Residual uncertainty: I still lack a granular daily star-history graph, so the exact day-by-day attribution split between organic AutoGPT-wave discovery and the LangChain push remains an informed inference, but the direction and rough magnitude are now measured rather than assumed.

Techniques & tactics

  • Open-source-as-distribution: MIT-licensed (later Apache-2.0) GitHub repo was the product, the funnel, and the marketing all at once; no gated launch.
  • Fix/improve a famous upstream (wedge): positioned squarely as "the AutoGPT that actually completes the task," borrowing AutoGPT's enormous mid-2023 attention and converting frustration with it into installs.
  • Hero benchmark numbers in the README and launch posts: ~3 minutes per research, 100% completion, 20+ sources aggregated, 1,200-2,000+ word reports, and a falling cost story (~$0.1, then ~$0.005) used as a recurring milestone-marketing beat.
  • Academic-paper credibility: README cites Plan-and-Solve, RAG, and (for the multi-agent version) STORM papers to signal rigor to a technical audience.
  • Peer/ecosystem endorsement, then ecosystem integration: secured the LangChain blog and a LangChain OpenAI-adapter + LangSmith integration, plugging into the largest LLM-dev distribution surface of the moment.
  • Distribution-surface breadth: PyPI package (6-line code snippet), Docker image, Colab notebook, hosted demo, and an embeddable demo video lowered trial friction across every developer entry point.
  • Build-in-public / milestone marketing: founder essays, an embedded Star History chart, and repeated "first / most-starred deep research agent" framing.
  • Docs-as-SEO and i18n: dedicated docs site plus Chinese and Japanese READMEs widened organic reach.
  • Open-source flywheel into a commercial product: the OSS repo doubled as top-of-funnel for the Tavily search API (Tavily badged as the recommended retriever inside GPT Researcher's own README), and the OSS brand was later cited directly in Tavily's $25M raise.

Sources