Takeoff founder Aakash Thumaty on going from $0 to near-8-figures in revenue before being acquired by Sierra

Takeoff's acquisition by Sierra

Aakash Thumaty built Takeoff in a little over a year, grew it from effectively $0 to near eight figures in revenue, and sold it to Sierra. The speed is the story.

The founding thesis

Thumaty drew a hard line between two types of agents. Human-in-the-loop agents — Copilot, Codex, Claude Code — deliver value to the person prompting them. They can run for hours, but a human kicks them off and evaluates the result. Autonomous agents are different. Buyers should feel like they are multiplying their labor force, with agents handling end-to-end work independently. Takeoff called these long-horizon agents and aimed them exclusively at revenue-generating use cases.

The revenue focus was deliberate. Thumaty's pitch to a lending company was simple: give us your lowest-quality leads. To a healthcare company: give us the patients most likely to churn. If the agents recover that lost revenue, Takeoff earns its place. If they don't, the customer has risked almost nothing.

“We entered this calendar year at effectively $0 in committed revenue. By the time we got acquired by Brett and Sierra, we were at near 8 figures in revenue. Our buyer was always the CEO or a c-suite member — when you are selling revenue, you are selling to the CEO. Our first $3–7 figure customers already had customer support vendors; with us, they were spending at least three times more.”

The pricing model

Takeoff charged on outcome, structured like a commission plus base units tied to token, voice, and SMS consumption. For a lending company, the trigger was loans originated or funded. The customer's logic was equally clean: paying hundreds of dollars in cost of goods sold to generate thousands of dollars in margin is a trade worth taking indefinitely. That framing is what let an unknown founder close multimillion-dollar contracts repeatedly.

Customers spending with Takeoff were also, in most cases, already paying a customer support AI vendor like Sierra or a competitor, typically between a few hundred thousand and a million dollars annually. Thumaty says Takeoff customers were spending at least three times more with Takeoff than with their support vendor, and in one case eight times more. The reason is category. Revenue-generating software sits in the same bucket as Google Ads or Facebook Ads — spend until the return line flattens.

How the agents actually work

Takeoff built what Thumaty calls a domain-specific language for agent construction — not a pretrained or fine-tuned model, but a harness that can operate across multiple services and handle a full workflow end to end. His framing is blunt: the inference API is a commodity. The value is in what runs on top of it.

He walks through the loan example in detail. An API call triggers the agent to fund a loan for a specific borrower. The agent then handles the initial rate-quoting call, a follow-up after document collection, outreach to third-party e-notaries and underwriters, resolution of edge-case borrower questions, and a final call to get the loan funded. The agent must be "always on" and responsive to anything that happens on behalf of, or in relation to, that central entity. That is materially different from a standard tool-call chain.

Thumaty's rule: if you only automate a horizontal slice of a workflow, you are useless, because everything above and below the slice still has to be coordinated manually. Own the whole thing or don't bother.

The commercial implication of that depth is that every customer's CEO is in Thumaty's top five iMessage contacts. Selling revenue means selling to the person graded on revenue. The agents have to understand the customer's business — gross margins, conversion rates, time to fund, first-contact to revenue — well enough that the CEO starts asking Takeoff questions about their own company. That, Thumaty says, is the signal that the relationship has worked.

The Sierra deal and Horizon

Thumaty says the conversation with Bret Taylor began with a shared view of where the market was heading, but with Takeoff already further along that timeline — proving end-to-end autonomous revenue generation was possible before most of the industry had accepted it was. The strategic logic from Sierra's side, as Thumaty describes it, was getting to where Takeoff already was rather than building toward it independently.

Sierra and Takeoff have launched a joint product called Horizon, positioned as a step-function jump in capability rather than an incremental improvement on existing agent or support tooling. The framing is deliberate: Horizon is designed to be bought by CEOs, not by IT or a VP layer below them, because it is sold on revenue impact.

Thumaty says sales calls with Taylor are notably effective — customers begin imagining specific board-deck outcomes before the meeting ends.