LangChain Raised Two Rounds Before Its First Paid Product

LangChain packaged the steps around a model call

A developer building a chatbot that answers questions from company documents has to find the relevant passages, put them into a prompt, send the prompt to a language model, and sometimes let the model call a search tool or a database. In late 2022, each team wrote that code itself. LangChain, a Python library, gave those steps standard building blocks: connectors to models and data sources, ways to run several model calls in a row, and “agents” that let the model choose which tool to use next.

It started as a side project on a personal account

The repository was created on 17 October 2022 under Chase’s personal GitHub account, hwchase17, and he merged his first pull requests that day. On 24 October, he restarted the history with a commit titled “initial commit”. Chase was a machine learning engineer at Robust Intelligence, a startup selling AI risk tools. He later described the first version as a single package of about 800 lines, built from patterns he saw at meetups where people were experimenting with language models.

For its first weeks it was mostly Chase’s project. He opened 264 of the 398 pull requests in 2022. The repository has 535 surviving stars (stars not since removed) dated before 30 November 2022, the day OpenAI released ChatGPT.

Ten Hacker News submissions drew 11 points or fewer

On 6 November 2022, Riley Goodside, a prompt engineer, submitted LangChain to Hacker News as “Chained, composable LLM (GPT-3) prompts in Python”. It got 3 points. The next nine stories mentioning LangChain, before the evening of 17 January, drew between 1 and 11 points each.

Stars had begun rising before the posts that took off: 85 are dated 16 January 2023, against 19 to 41 a day from 12 to 15 January. On the evening of 17 January, US Pacific time, two posts went up an hour apart. A demo combining GPT-3.5 with Wolfram Alpha through LangChain reached 107 points, and a link to the repository reached 372 points and 76 comments. GitHub dates 339 stars to 17 January and 636 to 18 January, the second most of any day in the repository’s history. In the seven days from 17 January, the repository gained 1,805 surviving stars, against 366 in the week before.

GitHub star history for langchain-ai/langchain, 12–24 January 2023, collected on 3 November 2026. Removed stars are not counted.

Benchmark’s partner met Chase before there was a company

Miles Grimshaw, a general partner at Benchmark, said he had been following language model research through 2022. On The Gradient podcast in September 2023, he described what happened next: he “ended up seeing LangChain in … fall 2022 come out and started spending time with Harrison.” Chase, he said, “wasn’t even sure he was going to make it into a company.”

Chase and Ankush Gola, a colleague from Robust Intelligence, started the company in February 2023, Chase later wrote. Grimshaw said the two spent a lot of time together, and when Chase said he wanted to form a company, “we said we should partner up together.”

Grimshaw later described the case in terms of who would use the library. He described “an order of magnitude more” developers arriving who were “application engineers”, not machine learning experts, and two things they needed: “the tooling to make that build experience easier, and then the confidence to put it into production.” Asked by TechCrunch in June 2023 whether LangChain was a software-as-a-service business, he said: “Yeah, exactly.” He also said Benchmark had not invested in companies building the models themselves, and had “instead focused on the developer layer above that.”

On 4 April 2023, LangChain announced a $10 million seed round led by Benchmark. The announcement cited “over 20K stars on GitHub, 10K active Discord members” and “over 350 contributors”. The stars and contributors are consistent with the repository: 18,406 surviving stars predate that day, and 354 people had a merged pull request, in line with the post’s “over 300 contributors on the Python repo alone”. The post described no paid product. It closed by asking teams “interested in evaluating and running your LangChain applications in production” to get in touch.

Sequoia’s link ran through a company it had backed

On 13 April, Business Insider reported that Sequoia was leading a round of $20 million to $25 million at a valuation of at least $200 million. A month later, the same publication quoted investors who found the valuation surprising “given the company’s minimal revenue and tough competition.”

Sequoia had a connection through Chase’s employer. When it announced the round, it wrote that Chase had been “a standout machine learning engineer at our portfolio company Robust Intelligence”, and Gola “a star colleague from Robust”. In late March 2023, Chase presented at Sequoia’s first AI Ascent, a one-day event for about 100 people. Sequoia partner Pat Grady later recalled that this was “when we were just starting to get to know you better”, and that Chase spent the day writing code. Chase recalled it as the day OpenAI released plugins for ChatGPT, which OpenAI announced on 23 March.

LangChain did not announce the Sequoia round until 15 February 2024, when it called it a $25 million Series A. Forbes reported the same day that Chase had closed $20 million from Sequoia, at a $200 million valuation, before the Benchmark seed was announced.

April 2023 brought 13,436 surviving stars, more than any other month before or since. It was also the month with the most people opening pull requests: 399.

GitHub star history for langchain-ai/langchain, October 2022 to December 2023, collected on 3 November 2026. Removed stars are not counted.

LangSmith was built for the problem after the prototype

On 18 July 2023, LangChain opened LangSmith in a closed beta. LangSmith records every step an application takes, including the exact prompt sent to the model and what came back, so developers can find where an answer went wrong. It also lets them collect example questions and rerun them after each change to see whether results improved.

The launch post explained why: something that “works ~30% of the time is good enough for a Twitter demo, but not nearly good enough for production.” It named design partners and early users, including Boston Consulting Group, Snowflake, Streamlit, Klarna and Fintual. Chase later wrote that LangChain made LangSmith “completely separate” from the library, so it worked with any framework or none.

Sequoia’s own interviews found little demand for such tools yet. In June 2023, it published findings from 33 companies in its network: 38% were interested in a framework like LangChain, and fewer than 10% were looking for tools to monitor model outputs, cost or performance. Sequoia wrote that “interest in these areas may increase” as more large companies adopted language models.

Developers criticised the library in mid-2023

Chase has written that around the summer of 2023, “we started to get a lot of negative feedback about langchain.” He listed breaking changes, hidden prompts, package bloat and outdated documentation, and said the hardest complaint to fix was that people wanted more control. On Hacker News, a submitted Reddit thread titled “Langchain Is Pointless” drew 386 points that July, and Max Woolf’s essay “The Problem with LangChain” drew 268.

Surviving stars fell from 13,436 in April to 2,916 in September. The record does not show how much of that decline came from criticism and how much from interest fading after the first rush.

LangChain’s answer was a second library. LangGraph, launched in January 2024, gave developers step-by-step control over what their applications did. langchain 1.0, released in October 2025, was rebuilt on top of it.

LangSmith went on sale with the Series A announcement

LangSmith became generally available on 15 February 2024, the day the Series A was announced. LangChain reported more than 80,000 signups, more than 5,000 monthly active teams and more than 40 million traces (records of each step in an application run) logged in January, and named Rakuten, Elastic, Moody’s and Retool as users. Forbes called LangSmith the company’s first paid product.

Sequoia’s announcement described the plan it had backed. It compared LangChain to React and Next.js, frameworks developers use to build websites, and wrote that developers also pay for “services such as GitHub, Datadog, Vercel and Heroku to deploy and monitor their applications into production. LangChain’s ambitions are similarly expansive.” It called LangSmith “the company’s latest ‘big bet’”.

Reported revenue came from LangSmith, and stars did not move with it

In July 2025, four sources told TechCrunch that LangSmith had brought LangChain to $12 million to $16 million in annual recurring revenue. LangChain didn’t respond to TechCrunch’s request for comment. On 20 October 2025, it announced $125 million at a $1.25 billion valuation, led by IVP, and reported 90 million combined monthly downloads of langchain and langgraph.

The repository’s stars no longer moved with these events. Surviving stars have run at 25,000 or more a year since 2024, with 28,074 in 2026 by 3 November, and neither funding announcement brought a jump: February 2024, the Series A month, had 2,253, fewer than the months on either side, and October 2025, the Series B month, had 2,173, between September’s 1,988 and November’s 2,356. LangSmith, the product the revenue was reported to come from, is closed source, so its growth does not appear in any public repository.

Stars are surviving stars dated to each year. Authors are people with at least one pull request merged that year, excluding GitHub app bots. Share is the part of that year’s merged pull requests written by its ten most active authors. 2026 runs to 3 November.

Fewer people get their work merged

The number of people with a merged pull request has fallen every year since 2023, while a handful of authors write a growing share. Two decisions narrowed outside contributions further. On 29 April 2025, LangChain moved community integrations to a separate repository. Since 9 March 2026, pull requests from outside contributors are closed automatically unless they link to a maintainer-approved issue assigned to their author; the rule replaced a policy that asked contributors to disclose AI use. Of the 4,053 pull requests people have opened so far in 2026, 30% have been merged, against 78% in 2023 and 2024.

What a small team can take from LangChain

These are our inferences from the record above, not claims by LangChain or its investors.

The first Hacker News submissions said little about where the project would go. LangChain drew 3 points in November 2022 and 372 points two months later, after ChatGPT’s release.

Both lead investors had a connection to Chase that predated their rounds: Grimshaw had followed the project since fall 2022, and Sequoia had backed his employer. Because the work was public, Grimshaw could follow it for months before Chase decided to start a company.

Neither investor’s public explanation cited revenue, and Sequoia’s was published after LangSmith had launched. Grimshaw described a large new group of developers and the tools they would need to reach production; Sequoia described a framework followed by paid services. The company built the paid product after it was funded.

LangSmith grew from a problem the library’s users kept hitting. Because it worked without the library, teams that moved away from the library could still pay for it.

Stars tracked attention in LangChain’s first year and little after it. From 2024, they ran at 25,000 or more a year, with no jump in the months of either round, while LangChain reportedly reached $12 million to $16 million in annual revenue.

Sources and method

This article covers 2022–2026. GitHub’s repository star history API, collected on 3 November 2026, counts only stars still present (“surviving stars”), not historical displayed totals or users. Its 212 weeks sum to the repository’s 148,959 stars at collection. Daily, monthly and yearly groupings use the returned week timestamps and day offsets; GitHub does not document the time zone of its day boundaries. The repository was first hosted under Chase’s hwchase17 account, which now redirects to langchain-ai/langchain.

Pull request figures cover all 25,184 pull requests in the repository, retrieved through GitHub’s GraphQL API on 3 November 2026. We excluded the 1,008 opened by GitHub app accounts, such as dependabot and github-actions, and 30 whose author accounts have been deleted. A person counts as having a merged pull request in a year if one of theirs was merged that year. Pull requests moved to other repositories, such as integrations after April 2025, are not counted.

Hacker News points and comments come from the Algolia Hacker News API in November 2026. The Gradient published no transcript of episode 90. We transcribed Grimshaw’s answer, from about 19:45 in the episode, with two speech-to-text tools and checked that they agreed on every word quoted.

Funding dates are announcement or report dates, not necessarily closing dates. Business Insider and Forbes give different amounts for the Sequoia round; we give LangChain’s announced $25 million and note the reports. Signups, teams, traces, downloads and customer names are as LangChain reported them. Revenue is TechCrunch’s report from unnamed sources, not a company figure.