Gigacatalyst wants your customers to build the feature they keep asking for

The oldest way to lose an enterprise software deal is also the quietest. A buyer runs the demo, likes most of it, then points at one thing the product can't do - a report shaped a certain way, a field their team lives by, a tab that isn't there. The salesperson says the honest thing: "I'll file a request." That request joins a queue. The queue has a name in most companies - the backlog - and the backlog is where deals go to cool off. Gigacatalyst, a company in Y Combinator's Spring 2026 batch, is built on the theory that the queue is the problem, not the feature.

Its answer is unusual enough to need a second read. Gigacatalyst embeds an AI agent directly inside another company's SaaS product. A non-technical user - a sales rep, a customer-success manager, sometimes the customer themselves - types what they want in plain language. About a minute later, a working feature appears inside the product, styled like the product, logged in as the same user, obeying the same permissions. No ticket. No sprint.

01 / The ProblemEvery buyer has a workflow no product ships

The company frames the pain as three related leaks. First, competitive deals lost when a vendor lacks one specific capability a buyer insists on. Second, weak adoption when a product almost fits a team's workflow but not quite, so people quietly stop logging in. Third, engineering backlogs clogged with one-off requests that are individually small and collectively bottomless. Each leak is familiar to anyone who has sold or built B2B software. What is new is the claim that one mechanism can plug all three.

The reason these problems usually stay unsolved is arithmetic. A one-off feature for a single account rarely earns a place on a roadmap serving hundreds of accounts. It is too small to prioritize and too frequent to ignore, so it sits. Gigacatalyst's move is to take the request out of the queue entirely and hand it back to the person who made it.

02 / How It WorksAn agent that reads your product first

The interesting engineering is not the text box. It is what happens before it. Gigacatalyst performs what the company calls agentic API discovery: the agent parses the host product's own endpoints, query parameters and data structures so it understands what the software can actually do. Only then does it translate a sentence like "build a KPI dashboard for maintenance performance" into a real, running microapp. The apps execute in sandboxes and inherit the host product's authentication and role-based access controls, so a user can't build their way past permissions they don't already have.

The examples the company cites are deliberately mundane, which is the point: a KPI dashboard, a photo-upload tab that runs OCR, a custom lifecycle tracker. These are not moonshots. They are the small, specific things that decide whether a team adopts a tool or abandons it.

03 / The NumbersSix weeks, and what came out of them

Gigacatalyst's early-deployment figures are the spine of its launch. Within six weeks, the company says, it unblocked more than $1M in pipeline and prevented roughly $100K in churn, while more than 800 features shipped - work it estimates at around 2,400 engineering hours. Across 946 users, it reports 90.8% adoption and 89% day-30 retention, with 670 microapps built entirely by people who don't write code. These are self-reported and early; they describe a promising start, not a proven category.

04 / The CustomerUpKeep, and a thousand daily users

The clearest evidence Gigacatalyst offers is a customer, not a metric. Ryan Chan, CEO of UpKeep - a maintenance-management platform, itself a YC alum (W17) that reached Series B - is on the record with a strong endorsement.

The detail worth sitting with is the "over 1,000 customers." UpKeep's own users are building and running these workflows. If the pattern holds, it flips a familiar relationship: the software vendor stops being the sole author of its product and becomes the platform other people build on.

05 / The MarketWhere it sits, and who it's up against

Gigacatalyst is aimed at vertical and enterprise SaaS vendors and sold to the people who feel feature gaps most sharply - CROs, VPs of Sales, Chief Product Officers - then used day to day by sales, sales-engineering and customer-success teams. It also reaches back into the sales motion itself: the website leads with turning customer requirements into configured demos and proof-of-concept environments, the setup work that normally eats a sales engineer's week.

The competition is partly other software and partly the status quo. No-code and low-code builders like Retool or Bubble ask a team to leave the product and assemble something separate. In-app copilots answer questions but don't ship features. Automation tools connect apps rather than extend them. And the incumbent Gigacatalyst most directly challenges isn't a product at all - it's the internal engineering backlog and the headcount thrown at customization. The distinction it presses on is that its output lives inside the host product, on the host product's data, under the host product's rules.

06 / The FounderA writer who argued himself into a product

Gigacatalyst is led by Namanyay Goel, its founder and CEO, who started programming at 13 and has spent roughly fifteen years building developer tools and apps. He is also unusually well read for a first-time-batch founder: his essays on AI and engineering drew more than four million views in 2025, and one - "AI is Creating a Generation of Illiterate Programmers" - passed 1.3 million on its own, with pickups in the New York Times, Futurism and Inc. The company operates as Giga Next Inc., and its earlier identity, Giga AI, still lingers in a LinkedIn handle.

There's a small irony worth noting. A founder who spent 2025 writing skeptically about AI hype is now shipping AI that writes software for people who can't. The reconciliation is in the product's caution: sandboxes, inherited permissions, features built on real APIs rather than hallucinated ones. It reads less like a bet on magic and more like a bet on plumbing.

07 / The ModelSelling time back to the people who lost it

Gigacatalyst is business-to-business software sold to SaaS vendors who embed it in their own product. Its pitch is framed almost entirely in the currency its buyers track: pipeline unblocked, churn prevented, engineering hours saved. That framing is doing quiet work - it positions the tool as revenue protection rather than an IT expense, which is the difference between a line item a CRO fights for and one that gets cut. Whether the early numbers generalize past a handful of design partners is the open question, and the honest one.