How Decagon Hit $1M ARR With 2 People in 6 Months

Two founders. Zero employees. $1M ARR in six months. Ashwin Sreenivas didn't follow a roadmap to build Decagon into a $1.5B company — he followed his customers. Here's the exact playbook he used.

Key Facts

• Time to $1M ARR: ~6 months with 2 founders, zero employees (Ashwin Sreenivas)
• Valuation: $1.5B in under 2 years (Pablo Srugo)
• Total raised: $230M, all via preemptive rounds (never ran a full roadshow) (Ashwin Sreenivas)
• Willingness-to-pay signal: Buyers offered $150k annually after weeks of work — repeatedly (Ashwin Sreenivas)
• Team size at publishing: ~225 employees across SF, New York, and London (Ashwin Sreenivas)

Why Over-Thinking Strategy Kills Early Startups

Most first-time founders waste months building grand strategies that fall apart on contact with customers. The fix is simple: skip the roadmap entirely and let 100+ buyer conversations do the strategizing for you.

Ashwin's first company, Helia, taught him a painful lesson about over-intellectualizing. By the time he started Decagon, he and his co-founder had a strict rule: no grand strategy before customer conversations. 'A lot of first time founders — and us included at the time — spent a lot of time over-intellectualizing the problem,' Ashwin said. 'You have this grand strategy and you bring it out to the market. And nothing works the way you thought it would.'

Instead, Decagon started with a simple hypothesis — LLMs could transform enterprise operations workflows — and tested it relentlessly with buyers. They talked to over a hundred leaders across ops, sales, and support teams. The goal wasn't to validate a thesis. It was to find where the pain was so sharp that buyers would cut a check before the product existed.

"It's really easy to fool yourself into thinking you're doing great, important work. And then you have this grand strategy and you bring it out to the market. And nothing works the way you thought it would." — Ashwin Sreenivas

The $150K Willingness-to-Pay Test That Proved PMF

True product-market fit shows up as immediate, repeated willingness to pay a significant amount. When Decagon tested customer support, buyers offered $150k annual contracts within weeks — a stark contrast to every other idea they tested.

Ashwin tested multiple ideas before landing on customer support. For most of them, the response was lukewarm — 'budgets are tight this quarter, maybe come back' or 'I'd pay a thousand dollars a month, but let's go month-to-month.' The contrast with customer support was immediate and unmistakable.

'With a couple weeks of work, we were at the point where people were like: yes, if you can deploy this thing, I will sign a $150,000 check immediately,' Ashwin said. 'And this happened repeatedly. This wasn't a one-off thing.' That repeated signal — not a single excited buyer, but the same reaction from customer after customer — was Decagon's real PMF moment.

The key diagnostic question Ashwin recommends: 'Why haven't you bought one of the other solutions already?' That question surfaces both the depth of pain and the specific gaps in existing products. Customers will tell you your competitive differentiation without you having to manufacture it.

"That stark contrast in willingness to pay. It's just direct signal of how much business pain is there truly." — Ashwin Sreenivas

• Weak PMF signal: 'Come back next quarter, budgets are tight.'
• Weak PMF signal: '$1k/month, month-to-month to test it.'
• Strong PMF signal: '$150k annually, I need this in production now.'
• Ask buyers why they haven't bought competitors — they'll tell you your differentiation.

How Two Founders Built to $1M ARR With Zero Employees

Decagon's two founders coded at night, called customers by day, and avoided hiring until ~$950k ARR. Staying small kept them fast enough to iterate on what customers actually needed — not what they assumed.

The first product took three weeks to build. It wasn't a polished platform — it was a custom build that captured enterprise workflow complexity end-to-end. For the first four months, every new customer got their own bespoke version. Scalability was not on the agenda.

Ashwin's reasoning was simple: 'The goal was to say, if we custom-built everything perfectly for one person, can we give that person a great experience? And then once you build the first three, you take a step back and say — what is common amongst these customers? And how do we build that into a great platform?'

The first hire, Amy, joined around $950k ARR. Partly by design, partly because they were so busy talking to customers and writing code they didn't have time to recruit. The lean structure also forced discipline: they only built what customers explicitly needed, not what seemed logical on a whiteboard.

"It was call customers during the day, code at night kind of a thing. And honestly, a big part of it was we were spending so much of our time talking to customers and coding that we didn't have as much time to go do recruiting calls." — Ashwin Sreenivas

Why Decagon Went Horizontal Instead of Vertical

Decagon skipped verticalization and built a horizontal enterprise platform. The insight: large enterprises across different industries — Hertz, Oura Ring, Eventbrite — all shared the same core support complexity, making verticalization unnecessary.

Conventional wisdom says AI startups should verticalize: pick one industry, go deep, own the niche. Ashwin rejected this after noticing something counterintuitive in his customer interviews. 'The problems that someone like Oura and someone like Eventbrite and someone like HearthSpace were very similar, actually,' he said. 'Let's just be a completely horizontal platform within the enterprise, but across enterprise industries.'

What made this work was refusing to compromise on depth for any customer. Rather than building a generic platform and asking customers to fit into it, Decagon went fully custom for each of the first three customers — then extracted the common patterns into a configurable platform. The result was a product that 'fits like a glove' for every customer without requiring a separate vertical product for each industry.

The differentiating capability wasn't the language model — it was the agent orchestration layer. Complex multi-step workflows, fraud checks, CRM lookups, shipping label generation — all handled conversationally, without the rigid 'if-this-then-that' trees that broke previous chatbot solutions.

"At no point did we ask any of our customers to compromise on that fully tailored feeling that they got. The hard challenge for us was: how do we build a platform where you can configure it to fit like a glove for every single one of our customers?" — Ashwin Sreenivas

Closing Enterprise Deals Before the Product Is Fully Built

Decagon closed six-figure deals with mock APIs and scraped help centers. Show buyers a working demo tailored to their specific workflow — even with fake data — and they can immediately see the value and commit.

Enterprise sales usually means long pilots, endless stakeholder reviews, and death-by-procurement. Decagon shortened the cycle by making value obvious on the first call. Before demos, they scraped every customer's public help center and built sample Agent Operating Procedures (AOPs) based on what they found. They showed up with a working demo specific to that company's workflows — not a generic product tour.

'We'd say: don't worry, we already scraped your help center. Anything we could find publicly. And we made some example AOPs for what we think some of your common workflows might be. Let's go do this together,' Ashwin explained. When a workflow assumption was wrong, they'd change it live. Instant proof of flexibility.

For capabilities requiring database access — like checking a customer's loyalty status before approving a return — they mocked the API calls. Fake data, real logic. 'You can paint a very vivid picture of what it would actually look like in practice instead of showing a pre-canned demo from an industry that has nothing to do with them,' Ashwin said. The result: buyers could mentally sign off on the product before integration was complete.

"We don't have access to your databases, obviously, and so this is fake data that we made. But the workflow is our best guess at what we think your actual internal workflow is — and if we need to change anything on the fly, let's just change it right now." — Ashwin Sreenivas

Defensibility in AI: Why the Model Is Only Part of the Answer

In enterprise AI, competitive moats come from orchestration, integrations, versioning, analytics, and testing — not just from having the best model. Companies where 90% of value comes from the model alone face brutal commoditization.

As foundation models improve rapidly, many AI products risk commoditization. Ashwin's view is that defensibility depends on how much of your value comes directly from the model versus the surrounding software. 'In some spaces where ninety percent of the value is provided just from picking the right model — the competitive dynamics are much, much higher,' he said.

Decagon invests in model fine-tuning and has an in-house research team. But the durability of the product comes from things that have nothing to do with AI: version control for deployed agents, automated conversation testing, deep legacy system integrations, and at-scale analytics. 'There's a lot of software you need to build — this is traditional, non-AI, regular SaaS software — that you need to build around something like this to make it work within the enterprise,' Ashwin explained.

The stickiest feature today, according to Ashwin, is the Agent Operating Procedure system: a natural-language interface that lets non-technical teams build complex, constrained workflows without engineering involvement. That capability compounds over time as enterprises build institutional knowledge into their Decagon configuration.

"Not all the value just comes from having the best model. That's where the actual competitive dynamics change quite a lot." — Ashwin Sreenivas

Weak vs. Strong PMF Signals in Enterprise Sales