Neptyne (Winter 2023) - Startups.RIP

Overview

Neptyne was a Brooklyn-based startup that built what it called "the first natively programmable spreadsheet" — a product that embedded a full Python and Jupyter runtime directly inside a spreadsheet interface. Founded in 2022 by two former Sidewalk Labs engineers, Douwe Osinga and Jack Amadeo, the company entered Y Combinator's Winter 2023 batch with a clear thesis: analysts and data scientists know Excel doesn't scale, Python is the obvious next step, and no existing tool bridges the two cleanly. Neptyne raised approximately $2.5 million, attracted thousands of early users after a well-received Hacker News launch, and later pivoted to a Google Sheets add-on to meet users on familiar ground. Despite genuine technical enthusiasm from a power-user cohort, the company could not convert that enthusiasm into sufficient paying customers to sustain independent operation. The service shut down on March 1, 2025, with the codebase open-sourced. The founders subsequently joined Block (formerly Square), with Osinga leading Block's open-source AI tooling team — making this a soft landing rather than a pure failure, but a commercial outcome well short of the original vision.

Founding Story

Douwe Osinga and Jack Amadeo did not meet at a networking event or through a mutual investor. They worked together at Sidewalk Labs, Alphabet's urban technology venture, where both were senior engineers building data-intensive city planning tools.^[1] The problem that eventually became Neptyne emerged directly from that work.

At Sidewalk Labs, the team built models in Python and then shared results with stakeholders via spreadsheets. The final step of many analytical pipelines was, as the founders later described it, "write a .csv with the results." Stakeholders received a static file. They could not interact with the underlying model, adjust parameters, or re-run scenarios. The Python world and the spreadsheet world were separated by a one-way export.^[2] The founders experimented with bridging that gap using the Google Sheets API, but found the experience clunky enough that they believed a purpose-built product could do it far better.

Osinga brought deep technical credentials to the founding. He is described as a serial founder with four prior companies, a former Google engineer, and a published O'Reilly author.^[3] Amadeo's background spanned Sidewalk Labs, Compass (where he worked on the real estate platform), and Bloomberg (communication systems).^[4] Together, they represented an unusual combination: deep Python and data infrastructure experience alongside familiarity with enterprise workflows where spreadsheets remain dominant.

The founding thesis was direct. As Osinga put it: "Talk to any analyst or financial modeler and they'll tell you that Excel just doesn't cut it anymore. Everybody is aware of the need to move to more powerful solutions and Python is the obvious candidate. Yet collaborating with today's tools is underwhelming."^[5]

The team applied to Y Combinator's Winter 2023 batch and was admitted, providing both institutional validation and initial capital.^[6] Their angel investor roster — which included Google AI lead Jeff Dean and Google Maps co-founder Lars Rasmussen — signaled that the idea resonated with people who had spent careers thinking about developer tools and data infrastructure at scale.^[7]

The company was incorporated in 2022 and headquartered in Brooklyn, New York.^[8] No major pivots occurred at the founding stage; the core concept remained consistent from inception through the YC batch.

Timeline

2022 — Neptyne founded by Douwe Osinga and Jack Amadeo in Brooklyn, NY, after identifying the Python-to-spreadsheet workflow gap at Sidewalk Labs.^[9]
February 22, 2023 — Neptyne launches on Hacker News without AI features, to test appetite for a Python-native spreadsheet. Thousands of users try the product; a power-user cohort emerges building trading systems, fitness trackers, and Twitter bots.^[10]

Launch HN: Neptyne (YC W23) – A programmable spreadsheet that runs Python

March 27, 2023 — Neptyne announces a $2 million pre-seed round from YC and angels including Jeff Dean and Lars Rasmussen. TechCrunch covers the raise. The company is pre-revenue with 4 employees and plans to hire 6 more.^[11]
May 15, 2023 — Neptyne closes a second (Seed) funding round per Crunchbase, with investors including TRAC and Other People's Capital. Amount unconfirmed; total funding reaches approximately $2.5 million per PitchBook.^[12]
August 15, 2023 — Neptyne launches free access to 5 APIs (including OpenAI) on its free tier, alongside the paid "Neptyne Pro" tier — indicating the freemium monetization model is now live.^[13]
October 19, 2023 — Neptyne launches a Google Sheets add-on, enabling Python functions to be written and called directly from Google Sheets via the Google Workspace Marketplace — a significant strategic pivot toward augmenting the incumbent rather than replacing it.^[14]
2024–2025 — Neptyne operates for approximately two years total before the founders decide to shut down the company. Exact date of shutdown announcement unknown.^[15]
March 1, 2025 — Neptyne service fully shuts down (both core product and Google Sheets add-on). Codebase is open-sourced to GitHub. Users given time to export data and self-host.^[16]
July 1, 2025 — Douwe Osinga publicly confirms on X (Twitter) that Neptyne was acquired by Block (formerly Square), and that he is now leading Block's open-source AI tooling team (Goose).^[17]

What They Built

Neptyne's core product was a spreadsheet that ran Python natively — not as a plugin bolted onto an existing tool, but as a ground-up environment where the spreadsheet grid and a Python/Jupyter runtime coexisted in the same interface.^[18]

In a standard spreadsheet, a user types a formula like =SUM(A1:A10) into a cell. In Neptyne, a user could instead write a Python function in a panel alongside the grid, and then call that function from any cell using the same = syntax. The Python environment was persistent — variables, libraries, and state were maintained across the session, just as they would be in a Jupyter notebook. This meant a user could import pandas, fetch data from an API, run a machine learning model, and surface the results directly in spreadsheet cells, all without leaving the interface or exporting a CSV.

As a co-founder explained in the Hacker News launch thread: "What we've done with Neptyne is create an environment where teams of both types can be productive in a shared environment: the usual spreadsheet formulas are there, and so is a full Python/Jupyter runtime."^[19]

The user experience followed a familiar spreadsheet layout. The grid occupied the main workspace. A code panel — similar to a Jupyter cell — sat alongside it. Users could reference spreadsheet cells directly in Python code (e.g., sheet["A1"].value) and write results back to the grid programmatically. This bidirectional link was the core technical differentiator: the spreadsheet was not just a display layer for Python output, but an interactive input/output surface.

The founders deliberately launched without AI features in February 2023 to validate demand for the underlying Python-spreadsheet concept before layering on AI capabilities.^[20] Once that validation was established, they integrated large language model support. Osinga described the rationale: "The large language models [like OpenAI's GPT-4] that have recently produced mind-blowing results happen to be very good at writing Python code. Since Neptyne natively speaks Python, it means that the AI doesn't just help you write formulas or visualize data — you can have a dialog with the AI about the spreadsheet application in front of you and have it modify it for you."^[21]

By August 2023, Neptyne had introduced a freemium model: a free tier with credits to access five APIs including OpenAI, and a paid "Neptyne Pro" tier for heavier usage.^[22]

The most significant product evolution came in October 2023 with the launch of a Google Sheets add-on. Rather than asking users to abandon Google Sheets entirely, the add-on allowed Python functions written in Neptyne to be called directly from within a Google Sheets document via the Google Workspace Marketplace.^[23] This was a meaningful strategic shift: the standalone product required users to adopt a new tool, while the add-on met them inside a tool they already used daily.

What distinguished Neptyne from alternatives like Observable or Hex was the spreadsheet-first interface. Notebook tools like Jupyter require users to think in code cells and markdown. Neptyne's grid interface was familiar to anyone who had used Excel or Google Sheets, lowering the conceptual barrier for analysts who were Python-curious but not Python-native.

Market Position

Target Customers

Neptyne targeted a specific intersection of two large professional populations: data scientists and analysts who worked in Python but were also required to deliver results in spreadsheets, and spreadsheet-native analysts who wanted more computational power than Excel formulas could provide.^[24]

The early user cohort that emerged after the Hacker News launch illustrated both the appeal and the narrowness of this segment. Users built option trading systems, machine learning classifiers, personal fitness trackers, and Twitter bot automation.^[25] These were technically sophisticated individuals solving real problems — but they were also a self-selected group of power users comfortable enough with Python to seek out a tool like Neptyne in the first place.

The broader target — financial modelers and analysts who know Excel is insufficient but haven't yet adopted Python — represented a larger potential market. But converting that group required not just a better tool, but a change in established workflow habits, which is a substantially higher bar.

Market Size

No TAM estimates were published by Neptyne or its investors. The addressable market for spreadsheet-adjacent developer tools is large in aggregate: hundreds of millions of people use Excel and Google Sheets globally, and the subset of knowledge workers who regularly work with data is in the tens of millions. However, the relevant segment for Neptyne — users who both work in spreadsheets and are willing to write Python — is considerably smaller. The founders' own framing acknowledged this: they were targeting analysts who "know Excel doesn't cut it anymore" and see Python as the obvious next step, a population that is real but self-selecting.

Competition

Neptyne entered a market with well-funded incumbents and well-capitalized challengers across multiple adjacent categories.^[26]

In the spreadsheet-evolution category, Airtable had raised over $1 billion by 2023 and positioned itself as a database-spreadsheet hybrid for teams. Equals, Pigment, and Actiondesk each targeted financial analysts with more powerful data connectivity than Excel offered. None of these tools offered Python execution, but they competed for the same "Excel is not enough" budget and attention.

In the notebook-adjacent category, tools like Hex and Observable targeted data scientists who wanted collaborative, shareable notebooks. These products offered Python execution in a more code-centric interface, competing for the Python-native user who might otherwise have tried Neptyne.

Most significantly, the incumbents themselves began closing the gap. Microsoft's Copilot integration in Excel and Google's Duet AI in Sheets both launched AI-assisted formula and data analysis features in 2023. While neither offered native Python execution at the level Neptyne did, they reduced the perceived gap between what a non-programmer could accomplish in a standard spreadsheet and what required a specialized tool. Microsoft also introduced Python in Excel in August 2023 — a direct feature overlap with Neptyne's core value proposition, delivered inside the world's most widely used spreadsheet application.

Business Model

Neptyne operated a freemium SaaS model. The free tier provided access to the core Python-spreadsheet environment along with credits to use five external APIs, including OpenAI's GPT-4.^[27] The paid "Neptyne Pro" tier was positioned for heavier usage, though the specific pricing and feature differentiation between tiers were not publicly disclosed.

The Google Sheets add-on, launched in October 2023, was available through the Google Workspace Marketplace and presumably followed a similar freemium structure, though no pricing details were published for that product specifically.^[28]

The company was pre-revenue as of March 2023, one month after its public launch.^[29] No revenue figures or paying customer counts were ever publicly disclosed. The absence of any revenue disclosure at any point in the company's life — combined with a total funding base of approximately $2.5 million — suggests the business did not achieve meaningful monetization before the decision to shut down was made.

Traction

The February 2023 Hacker News launch drove thousands of early users to the product.^[30] The response was enthusiastic enough to surface a distinct power-user cohort: individuals building option trading systems, machine learning classifiers, personal fitness trackers, and Twitter bot automation — all within Neptyne's Python-spreadsheet environment.^[31] This cohort validated the core technical concept and demonstrated that the product could support genuinely complex use cases.

However, as of March 2023 — one month after the public launch — the company was pre-revenue and declined to disclose the size of its customer base.^[32] The team had four employees at that point, with plans to grow to ten following the pre-seed close.^[33]

PitchBook's last recorded headcount for Neptyne was three employees — below the four reported at the pre-seed announcement, and far below the ten the team had planned to reach.^[34] While the date of that PitchBook data point is unconfirmed, the figure suggests the hiring plan was not executed, which is consistent with a company that did not achieve the revenue growth needed to justify scaling the team.

No revenue figures, paying customer counts, or conversion rates from free to paid tiers were ever publicly disclosed at any stage of the company's life.

Post-Mortem

Neptyne's shutdown announcement, published on the company's blog, was notably sparse. It confirmed the decision to close, set a service end date of March 1, 2025, provided instructions for data export and self-hosting, and announced the open-sourcing of the codebase.^[35] It offered no explicit reasons for the closure, no financial disclosures, and no post-mortem. The founders did not publish a lessons-learned essay. What follows is an analysis based on available evidence.

Primary Cause: Insufficient Monetization on a Thin Funding Base

The most direct cause of Neptyne's shutdown was the failure to convert a genuinely enthusiastic user base into paying customers at a scale sufficient to sustain the business.

The company raised approximately $2.5 million in total across two rounds.^[36] For a company building infrastructure-level developer tooling — which requires significant engineering investment in reliability, performance, and security — this was a constrained runway. At a lean burn rate of $100,000–$150,000 per month (consistent with a team of three to four engineers in New York), $2.5 million provides roughly 17–25 months of operation. The company operated for approximately two years before shutting down, which is consistent with running close to the end of its initial funding without securing a Series A.

The company was pre-revenue as of March 2023, one month after its public launch.^[37] No revenue figures were ever disclosed. The freemium model launched in August 2023 — six months after the product went public — and the Google Sheets add-on launched in October 2023. That left approximately 12–16 months for the paid tiers to generate enough revenue to justify continued independent operation or attract Series A investment. That window closed without either outcome.

The team's planned growth from four to ten employees did not materialize. PitchBook's last recorded headcount was three — below even the founding team's size at the pre-seed announcement.^[38] This contraction is a strong signal that revenue did not arrive in time to fund the hiring plan.

Secondary Cause: A Niche Power-User Cohort That Did Not Scale

The Hacker News launch attracted thousands of users and surfaced a cohort building option trading systems, ML classifiers, and automation tools.^[39] This was a meaningful validation signal — but it was also a warning sign about the nature of the addressable market.

The users who self-selected into Neptyne after a Hacker News launch were, by definition, technically sophisticated individuals who read Hacker News, were comfortable with Python, and were actively looking for better developer tools. This is a real but narrow population. Converting them to paid subscribers requires a product that solves a problem they face repeatedly and professionally — not just an interesting experiment they try once.

The use cases that emerged — option trading systems, personal fitness trackers, Twitter bots — were largely individual or hobbyist applications. These are not the enterprise or team use cases that generate recurring SaaS revenue at meaningful scale. A financial analyst at a hedge fund building an option trading system in Neptyne is a compelling proof of concept, but that analyst's employer is unlikely to standardize on a two-person startup's tool for production workflows without enterprise-grade security, compliance, and support guarantees that a $2.5 million company cannot credibly offer.

Tertiary Cause: The Incumbent Platforms Closed the Gap

Neptyne's core differentiation — Python execution inside a spreadsheet, enhanced by LLMs — was a genuine technical advantage in early 2023. That advantage eroded faster than the company could monetize it.

Microsoft announced Python in Excel in August 2023, the same month Neptyne launched its freemium model.^[40] Microsoft's implementation — powered by Anaconda and running in Azure — brought native Python execution to the world's most widely used spreadsheet application, available to hundreds of millions of existing Excel users without any tool switch. Google's Duet AI integration in Sheets added AI-assisted data analysis in the same period. Both moves directly addressed the "Excel doesn't scale" problem that Neptyne was built to solve, and both were delivered inside tools that enterprises already trusted, paid for, and had IT approval for.

Osinga had anticipated this dynamic in March 2023, framing LLMs as a core Neptyne differentiator precisely because Python-native AI could modify entire spreadsheet applications through dialogue.^[41] But the same LLM capabilities that made Neptyne's Python integration compelling also accelerated the incumbents' ability to build comparable features. Microsoft and Google had the distribution, the enterprise relationships, and the engineering resources to ship Python and AI features at a pace a seed-stage startup could not match.

The Google Sheets Add-On: A Reasonable Pivot, Possibly Too Late

The October 2023 launch of the Google Sheets add-on represented a strategic acknowledgment that displacing Google Sheets entirely was too high a bar.^[42] Meeting users inside an existing tool they already used daily was a sound instinct — it reduced the switching cost to near zero and expanded the addressable market from "people willing to adopt a new spreadsheet tool" to "people who use Google Sheets and want Python."

But the add-on strategy introduced its own challenges. Neptyne's differentiation as a standalone product was the depth of integration between the Python runtime and the spreadsheet grid. As an add-on, that integration was constrained by what the Google Sheets API permitted. The product became one of many Workspace Marketplace add-ons competing for attention, rather than a purpose-built environment with a coherent user experience. And by October 2023, Google's own AI features in Sheets were already reducing the perceived gap the add-on was designed to fill.

No data on add-on adoption, conversion rates, or revenue contribution was ever published.

Outcome: Acqui-Hire, Not Pure Failure

On July 1, 2025, Douwe Osinga posted on X (Twitter) confirming that Neptyne had been acquired by Block (formerly Square):

This reframes the shutdown as a soft landing rather than a pure commercial failure. The founders' technical skills — building Python execution environments, integrating LLMs with developer tools, and designing AI-augmented workflows — were directly applicable to Block's open-source AI tooling work. The open-sourcing of the Neptyne codebase, framed as "empowering the community to continue innovating independently,"^[43] is consistent with an acqui-hire pattern in which the acquiring company wants the talent but not the ongoing product liability.

The acquisition by Block has not been confirmed by a press release, SEC filing, or LinkedIn update from either founder as of the time of this writing. The sole source is Osinga's tweet. This claim should be treated as credible but unverified.

Key Lessons

Enthusiastic early adopters from a developer community are not a reliable proxy for paying customers. Neptyne's Hacker News launch drove thousands of users and surfaced a technically sophisticated cohort building impressive applications. But Hacker News power users are self-selected for curiosity and technical sophistication, not for willingness to pay for SaaS tools or for representing enterprise buying behavior. The gap between "thousands of HN users tried it" and "sufficient paying customers to sustain the business" was never closed, and the company's pre-revenue status one month after launch suggests the team may have underestimated how long that conversion would take.
Augmenting an incumbent is strategically sound but tactically constrained. The pivot from a standalone product to a Google Sheets add-on in October 2023 was a reasonable response to the high switching cost of asking users to abandon Google Sheets. But add-ons are limited by what the host platform's API permits, compete in a crowded marketplace, and are vulnerable to the host platform shipping native versions of the same functionality — which Google did. The pivot reduced switching cost but also reduced differentiation.
Thin funding limits the time available to find product-market fit in competitive markets. Neptyne raised approximately $2.5 million to compete in a market where Microsoft and Google were both actively investing in the same problem space. That funding level was sufficient to build a compelling product and validate technical feasibility, but it provided limited runway to iterate on go-to-market, build enterprise sales capability, or weather the period between a well-received launch and meaningful revenue. Companies building developer tools in markets where incumbents can ship competing features need either a faster path to revenue or a larger initial funding base.
The acqui-hire outcome suggests the technology had value even when the business model did not. Block's acquisition of the Neptyne team — specifically to lead open-source AI tooling work — indicates that the founders' skills in building Python execution environments and LLM-integrated developer tools were genuinely valuable. The failure was on the go-to-market side, not the technical side. Open-sourcing the codebase at shutdown was both a community gesture and a signal that the underlying work had merit worth preserving.