Apr 18, 2026
Zapier Won No-Code. Claude Has the Models. Why Did Benchmark Bet $50M on Gumloop?
Breaking Down Gumloop's Series B
Raise Report
💡
At a Glance
- Company: Gumloop, an AI-native workflow automation platform
- Co-founders: Max Brodeur-Urbas and Rahul Behal
- Funding: $50M Series B led by Benchmark (March 2026); ~$70M in total
- Investors: Benchmark, Nexus Venture Partners, First Round Capital, Y Combinator, Shopify Ventures, BoxGroup, The Cannon Project
- Growth: Team expanded from 2 at Series A to 15 at Series B
- Key Product: A visual canvas that lets non-technical users build AI workflows from editable Python nodes, with custom nodes generated from natural language; Gumstack extends the governance model across every AI tool running inside the enterprise
1. Why Benchmark's $50M Series B in Gumloop Deserves a Closer Look
On March 12, 2026, Benchmark led a $50 million Series B round in the workflow automation startup Gumloop. The company had fifteen employees. Y Combinator had backed it from the W24 batch. First Round wrote a seed check, and then wrote another at Series A. Nexus Venture Partners led the Series A and came back again.
Clearly, this early-stage startup had already been catching the eye of top-tier VCs. But the name leading this Series B, Benchmark, carries a gravity of its own.
Benchmark is the firm behind eBay, Uber, Dropbox, and, more recently, Airtable and LangChain. It keeps its fund small on purpose, writes a few concentrated checks a year, and sends the partner who signed the term sheet onto the board. This time, the partner was Everett Randle. He had joined Benchmark in October 2025, five months before the Gumloop term sheet, after a stint at Kleiner Perkins, where he worked on Glean and Harvey. Gumloop was his first deal at the new firm.
First deals at Benchmark are rarely accidents.
Which raises a question worth sitting with, because the market Gumloop has walked into does not, at first glance, have room for it. Zapier holds the incumbent position, profitable with $420M in ARR, 30,000+ integrations, and adoption by 69% of the Fortune 1000. n8n just raised $180M at a $2.5B valuation. OpenAI, Anthropic, and Google are all shipping their own agent builders. A fifteen-person startup does not usually draw a first check from Benchmark into a field that crowded.
So, what are the investors seeing that isn't obvious yet?
2. How ChatGPT Reopened the Enterprise Automation Market
To understand the bet, back up.
In December 2010, four co-founders launched IFTTT, a small service that let a non-engineer connect two web apps with a single rule: if this, then that. The following year, three University of Missouri friends started Zapier and applied the same idea to business users. Popular Science called the new category "letting normal people program." The New York Times called it automating the internet. The phrase "no-code" did not yet exist.
Zapier reached profitability by 2014, sits at a $5 billion valuation following a 2021 secondary sale, and hit ~$420M in ARR by 2026, surpassing 30,000 integrations. By any normal measure of market success, it is a massive company.
Plenty of tools followed, each taking a swing at the same demand. Make, n8n, and dozens of others. But none of them solved the one problem that mattered most: making automation feel approachable. The perception that giving instructions to a computer was hard work never really went away.
Then, on November 30, 2022, ChatGPT shipped. It reached 100 million monthly active users by January 2023, the fastest any consumer application had grown. At launch, ChatGPT had no obvious link to the automation market. But once people had experienced getting high-quality answers from an AI through a familiar chat interface, they arrived at a thought they had not had before: I can tell AI to do my work.
The irony is that most of what they wanted was simple automation. ChatGPT made them believe AI was the only way to get it.
The capital has followed that belief. Incumbents pivoted hard. n8n rode roughly 10x revenue growth in a single year and raised $180M at $2.5B, led by Accel with NVIDIA's venture arm participating. Make, acquired by Celonis in 2020 for over $100M, added AI agents and MCP support. A new wave of AI-native challengers emerged behind them, with Gumloop among the most visible.
That is the story of one lane. Zoom out, and it was never the whole market.
Enterprise automation has been three other things for just as long. RPA (UiPath, Automation Anywhere) automates legacy desktop applications for back-office work. iPaaS (MuleSoft, Boomi, Workato) integrates enterprise systems for IT departments, a market Gartner sized at roughly $11 billion in its 2026 Magic Quadrant, where even Zapier registers only as a Niche Player. BPM (Pega, Appian) models complex business processes end-to-end. Each has its own incumbents, its own buyer, its own decades of history. Together, they dwarf the no-code lane Zapier pioneered.
Now the walls between them are coming down. In Forrester's 2026 predictions, principal analyst Leslie Joseph put it directly:
"The enterprise automation space is moving toward adaptive, AI-driven workflows, as the focus shifts from deterministic to cognitive autonomy. Individual automation markets like RPA, iPaaS, and BPM have all but converged."
"The challenge for 2026 will be to figure out how to combine adaptive intelligence with proven controls, balancing innovation with trust."Leslie JosephPrincipal Analyst of Forrester
Gartner reached a similar conclusion, launching the inaugural Magic Quadrant for Business Orchestration and Automation Technologies in October 2025, a new unified category that folds BPM, RPA, iPaaS, low-code, and AI into one platform market.
What is emerging is a single giant enterprise automation market with no clear winner. Gumloop enters it from the smallest of the four origin lineages, as the youngest challenger, built AI-native from day one. That is the position Benchmark bet on, with Ev Randle's first check of his tenure.
Which leaves the practical question. Customers still type "Gumloop vs Zapier" and "Gumloop vs n8n" into Google every day, and the comparison posts published by both sides show how clearly the battle lines are drawn.
So what, if anything, genuinely separates Gumloop from the tools that came before?
3. What Makes Gumloop Different from Zapier, n8n, and Claude Code
This is where the product has to do its own talking.
Axis 1: AI-Native, which Zapier doesn't know how to be
Gumloop was designed from the ground up around AI. The origin is specific. In April 2023, Max Brodeur-Urbas, a former Microsoft Azure Linux engineer who had just been denied re-entry to the United States, was watching AutoGPT's Discord server fill up with questions like "what is GitHub" and "how do I use a terminal."
"Whenever anyone would ask in the Discord server for help setting up their local environment, I would send them a link to AgentHub, which was the first version of Gumloop. After building it for a couple of days, I was like, oh, this could be like GitHub for agents."Max Brodeur-UrbasCEO of Gumloop
Max's AgentHub story makes one thing clear: Gumloop was built around AI agents from the start. The founding idea was to give people an accessible interface for AI agents. Today's product looks very different from that first version. So where does Gumloop's AI-native claim actually show up?
Three places to look.
First, Gumloop's agent and workflow pages open with a natural language prompt. The screenshot above is the first thing users see when they open Gumloop. A text input sits front and center, so anyone unsure where to start can just ask in plain language.
The same principle carries into the workflow tab, where most of Gumloop's actual work gets built. Users aren't left staring at a blank canvas, wondering what to assemble first. If they don't know where to begin, they can begin by asking. This is why Gumloop has been described as a child of Zapier and ChatGPT.
Second, if the prebuilt library is not enough, users can build custom nodes in natural language. The AI writes code underneath. Any trigger a person can imagine becomes a node, not just the ones a product manager at Zapier pre-decided were worth shipping. Every niche workflow that no vertical SaaS would bother to build, Gumloop can reach.
Third, and most revealing: the built-in AI node library. Tasks that are genuinely effective when delegated to AI, like categorizing, web searching, analyzing, and scoring, are already named, visible, and selectable as discrete functions. The user does not need to know that AI can do this. The product already knows.
Axis 2: Code Nodes, which Claude Code cannot do
However, Gumloop's deeper differentiation, in Max's own framing, runs in a surprisingly different direction.
"I often compare Gumloop to it's more like Python than it is to like a vertical SaaS solution. There are some general things you do with it, but normally it is a tool that you use to build something really specific and really niche."Max Brodeur-UrbasCEO of Gumloop
To understand this, we have to go back to its beginning. Max built an agent platform first. Then he watched users get frustrated. His diagnosis, in his own words:
"I realized the agents weren't useful, which was the aha moment. I kind of gave them what they were secretly asking for, which is just reliability, predictability."Max Brodeur-UrbasCEO of Gumloop
What followed was a platform built on roughly 90 percent deterministic workflow structure and 10 percent AI judgment. The philosophical inverse of autonomous agents.
The natural endpoint of that philosophy, surprisingly, is code. Not no-code. Brodeur-Urbas now describes Gumloop as an "O code" platform, meaning users should be "delightfully surprised" to find there is code underneath. He has also compared the product to Python itself. Not a black box. A code manager where the engine runs on AI but the structure can be inspected and assembled.
In practice, every custom node in a Gumloop workflow has code underneath it. That code can be opened, read, and edited directly. A user can modify exactly the part they want and leave the rest untouched. Surgical precision, not wholesale regeneration.
This is the structural difference from CLI-based coding agents like Claude Code. When Claude Code generates a workflow, the output is a single artifact. Gumloop cuts the workflow into node-sized pieces. Each piece is individually inspectable, individually editable. The complexity does not disappear. It becomes addressable.
The shared principle
Both axes express the same design logic. Don't hide complexity. Cut it into units a person can handle.
The product that resulted sits at an unusual intersection: the probabilistic range of AI agents and the auditable predictability of deterministic automation, simultaneously. Most tools pick one. Gumloop treats the tension as a design constraint to solve for.
The combination of AI's generative range and automation's predictability is exactly what every enterprise wants. To make sure Gumloop could grow from a clever individual tool into something companies deploy at scale, the team built a governance layer called Gumstack. Administrators can track traffic across workflows and manage access permissions from a central view. The ambition to grow into a serious B2B platform is visible in that build.
4. So Did They Actually Win Enterprise Customers?
Roughly 80 percent of Gumloop's users have no technical background. Salespeople, operations managers, finance teams, and HR coordinators. That is the audience Brodeur-Urbas watched adopt AutoGPT, and the audience he has said repeatedly he is building for.
In the twelve months after launch, he personally logged approximately 1,100 customer calls. Most ran for thirty minutes. Many were, by his own admission, longer than they needed to be. What came out of them was an internal diagnostic he calls "customer cringe." If he found himself cringing on a call, hedging, redirecting, saying "no, click there, click there," the product had failed at that moment, not the user. The cringe was the metric. The 1,100 calls were the practice.
Four pieces of evidence stand out that Gumloop is building close relationships with its customers and producing real results.
(1) Randle's due diligence
During the Series B process, one customer's CTO told Randle directly how Gumloop had been chosen. He had given employees full access to Gumloop and to two competing products at the same time, with no guidance or mandate. Six months later, staff were using Gumloop daily or weekly. The competing tools sat untouched. Randle pointed to a minimal learning curve as the reason. Users, he said, "can go in and start making agents and workflow automations immediately."
"When we asked enterprises why Gumloop has emerged as the primary AI platform for their employees, they consistently pointed to the product’s balance between powerful capabilities & ease of use, as well as the team’s customer obsession & dedication to service."
"The incredible adoption & expansion we’re seeing within enterprises firsthand is a testament to the depth of this team’s commitment to their users & customers."Ev RandleGP of Benchmark
(2) Shopify and the Executive Who Took It With Her
Shopify rolled Gumloop out company-wide and has since processed approximately 17 million automated actions across thousands of custom workflows. The pattern underneath that number is more interesting than the number itself.
Ritu Khanna, formerly VP of Global Partnerships and Monetization at Shopify, left in 2025 to join Gusto as Chief Commercial Officer. She brought Gumloop with her. A senior executive changed companies and packed the tool alongside everything else she knew worked. That is not a story about lucky distribution.
(3) Gusto's Measurable Wins
At Gusto, a sales coaching agent built on Gumloop produced approximately $1.5 million additional ARR within three months. A partnership review agent cut rep prep time from 30 to 45 minutes down to under five, and raised win rates by 31 percent on the deals it touched. An outreach agent booked meetings at roughly twice the rate of human-run campaigns.
(4) Gumloop itself
Gumloop itself is arguably the cleanest proof. At Series A, the company had two founders and two summer interns. At Series B, fifteen people. The team runs on its own product. Max estimates that everyone on the team has automated "millions of tasks." The efficiency is not a claim. It is the concept demonstrated.
"We're like... some of the biggest users. I think everyone on the team has automated literally millions of tasks."Max Brodeur-UrbasCEO of Gumloop
However, three caveats are worth noting.
First, Gumloop has not disclosed ARR. At comparable stages, companies like Harvey and n8n have published that number. Silence is not a neutral signal.
Second, every named enterprise customer is a tech-forward company: Shopify, Gusto, Ramp, Instacart, Opendoor, and Samsara. If the real prize is non-technical workers across industries that do not already have an AI strategy, the evidence from traditional enterprise has not yet arrived.
Third, Max has acknowledged that B2B sales require more human effort than he originally planned for, and has begun hiring accordingly. Whether the efficiency ratio that got Gumloop to Series B will hold as the team scales remains an open question.
5. Inside Gumloop's Cap Table: Y Combinator, First Round, Shopify Ventures, and Benchmark
By the numbers, Gumloop is still a small and uncertain startup that emerged through the cracks of a new era. And yet, by March 2026, investors had placed roughly $70 million behind it. Who are they, and what did they see?
The funding history table from Section 1 deserves a second look. Four signals stand out.
(1) The Angels Who Said Yes First
The angel lineup at Seed is not random. Mullen (Instacart), Ferdowsi (Dropbox), and Ofstad (Airtable) each built products used at scale by non-technical users. They know what a tool for that audience looks like when it actually works. Their participation before any institutional syndicate is set is the judgment of people who can evaluate the product, not just the market.
(2) Who Doubled Down
The follow-on pattern is more telling than the lead changes. First Round wrote at Seed and again at Series A. Y Combinator participated at every round from pre-seed through Series B. Nexus led Series A and came back for Series B. These investors have access to internal data the public does not see, and they kept writing checks.
Y Combinator in particular is worth pausing on. The firm wound down its $700M Continuity Fund in March 2023, citing the growth-stage program as "a distraction from the core mission." Participating in every subsequent round of a portfolio company is not default behavior anymore. It is a deliberate choice. That sharpens the signal.
(3) The Customer on the Cap Table
Shopify Ventures is the most unusual entry. It is not a financial bet. It is a dependency bet. A company using Gumloop as operational infrastructure invested in the company providing that infrastructure. The implication that Gumloop is replaceable within Shopify's stack writes itself.
(4) The Harvey Thesis, Applied Again
Of all these signals, the most interesting is still the latest one: Benchmark and Randle. Benchmark makes concentrated investments. When they write a check, they have a specific thesis about why this company can define a category.
Randle's most consequential bet before joining Benchmark was Harvey, the legal AI company he worked on while at Kleiner Perkins. When Randle backed Harvey, the legal AI market looked, on paper, already won. Thomson Reuters and LexisNexis held decades of proprietary data, existing relationships with virtually every lawyer in the world, and the kind of structural advantages that typically decide enterprise software categories. The conventional read was that incumbents would adapt, absorb, and win. Harvey's bet ran in the opposite direction: a team building fast with direct customer relationships could establish a position before the incumbents fully adjusted.
Harvey today: approximately $11 billion valuation after a March 2026 round co-led by Sequoia and Singapore's GIC, roughly $190 million ARR, and relationships with the majority of the Am Law 100.
The Gumloop investment reads like the same thesis in a different market. A category that looks settled. Large players holding structural advantages. A small team whose product spreads through use rather than outbound sales.
"Enterprise automation is a massive pot of gold. I think it's the biggest category in enterprise AI."Everett RandleGP at Benchmark
Randle is an investor who believes speed and direct customer relationships are the essential weapons of a startup at this stage. That conviction is the same bet every investor on this cap table has made. It is also the same bet Gumloop itself is running on.
6. Can Gumloop Hold Its Moat Against OpenAI, Anthropic, and Zapier?
However, whether speed and customer orientation can constitute a genuine moat is an old argument without a settled answer. The companies with foundational technology (OpenAI, Anthropic) and those with established networks (Zapier, n8n, Make) are formidable in different ways. Neither set is going to sit still.
Max knows this better than anyone. He has said openly that on the first day he started building, he could have listed a hundred reasons why Zapier would do it better, why OpenAI would crush him, why the whole thing would be pointless.
"I could have, on day one, explained to myself 100 reasons why Zapier would do this better than us. We wouldn't have created anything new."Max Brodeur-UrbasCEO of Gumloop
But the more important context is structural, and the structure has shifted.
ChatGPT did not just launch a new product. It broke a belief that had capped this market for fifteen years: that giving a computer instructions is inherently hard work. When that belief broke, the automation market opened in a way it had not before. The incumbents riding that wave, the challengers pivoting into it, and the new entrants born from it are all riding the same swell. Their positions differ. The wave is the same.
That shift is what created a real opening for Gumloop. Fifteen years without a single category winner. When the ground moved, new players got a chance that structural logic had previously denied them. Speed and customer orientation are Gumloop's bets on how to capitalize on that opportunity.
The question that remains, whether this wave produces a new category winner or gets absorbed by the model providers themselves, is not Gumloop's question alone. It is the question facing every automation company born into this generation of AI.
Asked about exactly that risk, Brodeur-Urbas answers the way a founder has to answer:
"I think there's a million reasons not to build every startup. It's very easy to hear an idea and say, what's the moat? Why won't X company do this? Why won't Y company do that?"
"I think the people who are really asking themselves that and obsessing over those questions will never build anything."Max Brodeur-UrbasCEO of Gumloop
For a founder, that is the right orientation. Whether it is also a sufficient theory of competitive advantage is what the next two to three years will answer.
Watch the in-depth interview with Gumloop CEO Max Brodeur-Urbas on the EO YouTube channel!
💡
Further Reading
- Gumloop
- YC Companies - Gumloop
- Why we built Gumloop (Gumloop blog, July 2024)
- Gumloop lands $50M from Benchmark to turn every employee into an AI agent builder (Techcrunch, March 2026)
- Gumloop sticks $50-million USD Series B round to let employees build their own AI agents (Betakit, March 2026)
- Benchmark's Newest General Partner Ev Randle on Why Margins Matter Less in AI & Why Mega Funds Will Not Produce Good Returns (20VC, Nov 2025)
Explore more