software engineering got there first.
sales is next.

Software engineering was the first profession to be transformed by AI, and it happened for a simple reason: it is mostly intelligence work. AI crossed the threshold where it can do the intelligence work autonomously and leave the judgment to humans. Cursor and Claude Code didn't make engineers a little faster. They changed what an engineer is.

This is coming to every profession. The only question for each one is who builds the infrastructure.

throxy is building it for sales.

The old model sells software and leaves you to find the outcome. Apollo charges $79 a seat and wishes you luck. The new wave looks different but isn't: Clay is the best of them, and it still just enriches, automates, and beautifies the same legacy atom, the contact. It still charges 'credits' for actions, not outcomes. Faster tooling around a motion invented decades ago. We charge ~$500 per attended meeting and keep the upside. If we don't deliver, we don't get paid. We don't sell the tooling around selling. We sell the outcome, because we own the only thing that produces it: the live conversation.

And the meeting is only the entry point. We take kickers when a meeting becomes a qualified opportunity, we'll take a percentage when the deal closes, and next a percentage when it renews. We get paid the way a great sales team gets paid, except we are the sales team for everyone.

When you own the outcome, you also own the infrastructure that produces it. Every outcome you deliver makes the next one cheaper to deliver.

sales has always had a market making problem.

Selling only happens when two humans are live at the same time. Everything else, research, list building, dialing, waiting… is not selling. Across all our dials, only ~15% of the market ever picks up, and you never know when.

The result is that traditional sales reps, even the most efficient ones, are idle at least 95% of the time. We found this insight because our best SDRs only spend 30 minutes a day actually in conversation. Everyone quotes "salespeople only sell 20% of the time". What they really mean is "salespeople only attempt to sell 20% of the time". The real number is two orders of magnitude worse.

Every tool ever built for sales is organized around the contact list: the who should we call. That is the wrong atom. The right atom is the live pickup: someone is on the phone right now.

List-based selling creates a permanent supply-demand mismatch, sellers idle when nobody picks up, pickups dropped when everyone is busy. The chart below is what that looks like, and what it looks like when you fix it.

We are inverting the model, synchronous sales. Don't route a rep to a list. Route every live pickup, the instant it happens, to the best available seller for that conversation. This is market-making. For this, it requires owning the dialer, it requires owning the infrastructure.

Third-party dialers were built for a number of reps working one list for one company.

We operate at a different scale and shape: one pool of sellers, many customers, tens of thousands of dials a day. Owning the dialer means every pickup, outbound or inbound, enters a single real-time market.

A callback can match any of our customers; we know who to pitch the moment the phone rings. The best seller is always in a conversation, never on a list. We measure this as 95% recoverable inefficiency in the current stack.

Concretely: we have ~5,000 live conversations a week today. Same sellers, same customers, synchronous routing takes that to 40,000–50,000.

And because we route every conversation, we can rate every seller. Some of our SDRs convert 25x better than others. An ELO system lets us put the best ones in 25x more conversations and remove the worst entirely.

We now make so many calls that the market calls us back. Through our outbound, we receive more inbound than some Fortune 500 sales orgs. We call this inbound led outbound.

At this volume, calling stops being outreach and becomes sensing. Every conversation is transcribed and fed back into the system. Someone tells us they're leaving their job, and our system of record updates itself. Someone mentions a pain point, that's intent data no provider sells, because we're the only ones in the conversation. Someone wants to buy, we already know which of our customers to connect them with.

The loop is fully closed: agents map the target market and build the lists, our dialer talks to the market, outcomes (meetings booked, attended, closed) feed back into who we target next. Every meeting we book makes the system better at finding the next one. Traditional sales teams operate on static snapshots. We operate a live model of the market that updates with every conversation. Nobody can compete with this data and model because nobody else generates it.

Once you are the routing layer between every live buyer and the best available seller, you are the infrastructure of selling. Two motions fall out of this:

• a.We are the infrastructure: the customer's own reps plug into our routing, ranked alongside everyone else, and we take a percentage.
• b.We own the seller: we book the outcome and charge a higher percentage.

Sellers aren't in enough conversations. Nobody is fixing that. Everyone is busy building systems of record and automating the activities around selling fake work.

We are moving the atom of sales to the pickup, the first synchronous connection, and rebuilding everything around it to deliver real outcomes and revenue.

software engineering got there first.
sales is next. throxy will be the infrastructure of selling.