We Booked 614 Meetings With One Inbound Agent. Your “Contact Us” Form Is Costing You Deals.

From The Agents, our weekly podcast on how we deploy and run AI agents at SaaStr, and how to do it yourself.

The contact-us form is the most expensive lazy decision in B2B today. Someone arrives on your site ready to talk, hits a form, and waits. A human round-robins it. A rep follows up on a delay. Two or three days later, if the lead is still warm, a conversation happens. Most of the time it isn’t still warm.

We replaced ours with an AI agent. For one event, that agent booked 614 meetings at roughly $85K average ticket size, across about 2.25 million sessions and around 402,000 interactions, with almost zero complaints. We’re three people. There is no version of this we could have staffed with humans.

What follows is the entire build: the old flow we killed, the numbers, how the agent is trained, how it routes and runs campaigns and discounts on its own, and exactly how to do this yourself. This is the single most Captain Obvious move in AI go-to-market right now, and we’re still shocked how many companies, including AI startups, run a contact-us form and nothing else.

The Old Way: A Two-to-Three Day Death March

Here’s the flow we ran on Squarespace, and the flow most companies still run.

You land on the site. You want to talk about a sponsorship, so you fill out a contact form. You tell us which sponsorship you’re interested in, roughly what budget you have, and who we should follow up with. Then you wait.

On our end, someone had to round-robin that lead to an account executive. The AE picked it up on whatever delay their day allowed. They followed up. By the time a real conversation started, two or three days had passed. For inbound, that’s an eternity. The prospect has moved on, talked to three competitors, or lost the urgency that made them fill out the form in the first place.

That flow has two fatal problems. It’s slow, and it leaks. Every handoff is a place where a lead cools or disappears. And it scales linearly with headcount, which means the only way to handle more inbound is to hire more BDRs, who then quit every few months and take their context with them.

The Numbers

We launched Amelia AI last summer to fix this. She runs on Qualified, which Salesforce now owns. For one event, here’s what she did:

614 good meetings booked.****~$85K average ticket size.Run the multiplication and you can see why this is a high-ROI agent even before you count the savings on headcount.~2.25 million sessionsacross the event site.~402,000 interactionshandled directly.Almost zero complaintsabout booking this way.

They obviously didn’t all close. If every one of those 614 meetings had converted at the average, we’d be looking at tens of millions in sponsorships, which isn’t reality. But the efficiency is the point. A small team turned a flood of inbound into hundreds of booked, qualified meetings without a single BDR.

We run go-to-market with about 2 people. To handle 2.25 million sessions and 402,000 interactions and still book 614 meetings, we’d need a stack of BDRs we don’t have, working a pace they wouldn’t sustain, quitting every three months. The agent doesn’t get tired, doesn’t quit, and doesn’t have a bad week.

Why She’s Good: She’s the Most-Trained Agent We Have

The metrics come from one thing: training. Amelia AI has one of the biggest knowledge bases of any agent we run, and she’s the most-trained of all of them. Three things make that real.

She’s trained on the nuance, not just the FAQ. We have two completely different buyer types hitting the site. There are self-serve customers who can buy a ticket straight off the website, and there are sponsors, both current and potential, looking for something entirely different. We spent serious time with our forward deployed engineer at Qualified building that nuance in from launch, so she knows which kind of buyer she’s talking to and responds accordingly. She’s part seller, part customer support, part marketer, and she switches between those depending on who’s in front of her.

She’s never out of date. She crawls saastr.com and the event site in real time, every day. On top of that, anytime we push a release to one of our other agents, 10K or Annie, we push the same context to her. Anything that lives on the back end goes into her memory too. So when someone asks where a session moved or what’s on today’s agenda, she has the right answer, because she was already updated this morning, automatically, from the live site.

She has the right brain for the right context. Inside Qualified we run separate contexts: one for the main event, one for our London event, and a tighter one for in-person attendees at the event itself. We deliberately compacted her in-person memory so she isn’t dragging all of saastr.com into a question about which room a session is in. The result is faster, more accurate answers for whatever context the visitor is actually in.

That’s the whole reason she converts. She answers honestly, without drama, with fresh data instead of last year’s, and she does it instantly.

What The Inbound AI Agent Does Beyond Chat

Answering questions is the floor. The operational leverage is in three things she does on her own.

1. Smart routing weighted on real close data

When Amelia AI books a meeting, she decides which human gets it, and she weights that decision on our actual Salesforce close data. She looks at what each of us closes best and routes accordingly. Most meetings go to the rep with the broadest book, and the deals that fit a specific closer’s strengths get routed to that person.

This took tuning. For a while she over-indexed one of us with all the legacy-company meetings, because the data showed one rep closed a lot of old-school accounts, so she sent every old-school account that way. We had to correct the weighting so it didn’t route everything to one person. She now balances the book the way a good sales manager would, except she does it on every single lead in real time. At one point our own internal bot even flagged that one of us had been off the round-robin for two months and asked if we wanted back in.

No human ops team is re-weighting routing on every inbound lead based on live close rates. She does.

2. Two triggered campaigns that recover leads humans drop

She runs two re-engagement campaigns automatically, based on what you do on the site.

The sponsor campaign: if you visit the sponsor page and don’t finish, but we already know who you are from Marketo or Salesforce, she follows up on her own. She offers a meeting, and she pulls a few lookalike sponsors already active in your space to make it concrete. She also checks first and excludes anyone who’s already a sponsor, so a current customer browsing the page doesn’t get a “want to sponsor?” email. Find the right people, confirm they’re not already customers, follow up without a human lifting a finger.

The ticket campaign: if you hit the site and we know who you are but you don’t buy, she sends a VIP code. If you talked to her and she gave you a code you didn’t use, she follows up. If you didn’t interact at all but came back a few days later, she sends you a code to pull you in. That single campaign has sold hundreds of thousands of dollars in tickets, entirely from leads a human would never have circled back to.

3. Automated, guardrailed discounting

This is the subtle one, and it’s harder for humans than it looks. We hate discounts. But the data over many years is unambiguous: it works better to mark a price up and offer a structured discount, because that’s how buying psychology works. The problem has always been the humans. Reps forget codes, make up discounts on the spot, and when they smell a deal slipping they panic. They go from 20% off to 25% to 30% to 34%, chasing a buyer who was never going to be moved by price, and all they do is train the market to wait for a bigger discount and create margin problems.

Amelia AI removes that entirely. She gives the right discount, on the right schedule, inside hard guardrails, and she can step it over time up to a cap. It works like a real-time, lightweight CPQ driven by rules instead of a rep’s mood. No forgotten codes, no panic spiral, no drama. For self-serve buyers, that consistency is worth real money. It wouldn’t replace a human on a true enterprise negotiation, but for the long tail it’s better than any rep, because it never flinches.

It’s Not Just Us

If you think a high-volume inbound agent is a SaaStr AI quirk, look at the operators and the data.

Kyle Norton, CRO of Owner.com, describes the same shift toward letting the buyer choose their level of engagement, from fully self-serve to fully handheld, and meeting them there with a blend of AI and humans.

The example he points to is stark: Perplexity serves thousands of enterprise customers with a sales team you can count on one hand. That ratio is only possible when an agent handles the front of the funnel. Kyle is also blunt that AI agents are now better than a mid-pack AE or SDR, which is exactly the tier doing inbound qualification at most companies.

**ICONIQ’s 2026 GTM data, drawn from 150+ B2B revenue leaders, explains why the cost of a slow contact form is rising. **

Top-of-funnel has held up, but the funnel is breaking at the bottom: demo-to-close conversion is down 5 to 10 points year over year, sales cycles run 3 to 4 weeks longer, and pipeline coverage has slipped. When every deal is harder and slower to close, a two-to-three day response lag on warm inbound is no longer a minor inefficiency. It’s where deals die. The same data shows teams with strong AI adoption hitting quota at 67% versus 59% for everyone else.

Speed and coverage are the whole ballgame on inbound, and a form has neither.

How to Actually Do This

The playbook is not complicated.

Pick a vendor and commit. Qualified is what we use, but it’s not the only one that does this well. The choice of tool matters far less than training it properly. Don’t run an eight-vendor bakeoff. Pick one and go deep.

Train it on every buyer type you have, separately. We have two, ticket buyers and sponsors, and we built the distinction in from day one. If you have self-serve and enterprise, or SMB and mid-market, the agent needs to know which conversation it’s in. Generic training produces a generic chatbot.

Connect it to your real data and keep it fresh. Crawl your own site daily. Push every relevant back-end change into the agent’s memory. An inbound agent answering from stale data is worse than no agent, because it answers confidently and wrong.

Give it the jobs beyond chat. Routing rules weighted on your close data. Re-engagement campaigns for the people who don’t finish. Discount guardrails if you discount at all. That’s where the agent stops being a chatbot and starts being a rep.

Go incognito on your own site first. Before you build anything, go through your own inbound flow as a stranger. Fill out your own contact form. Time how long it takes to get a real response. Fix the thing that embarrasses you most, and that’s your starting point.

The Playbook in Bullets

The classic contact-us form is slow and it leaks. Every handoff is a place a warm lead cools. It scales only by hiring BDRs who quit.One inbound agent booked 614 meetingsat ~$85K average, across ~2.25M sessions and ~402K interactions, with a three-person team.**Training is the entire moat.**Make it the most-trained thing you own: every buyer type, crawled daily, updated with every back-end change.Run separate contexts. A tight, venue-specific or product-specific brain answers faster and more accurately than one giant memory.**Route on real close data.**The agent re-weights every lead in real time the way a great sales manager would, but on every single lead.**Recover the drop-offs automatically.**Two triggered campaigns, sponsor re-engagement with lookalikes and ticket VIP codes, pull back leads humans never circle back to.**Automate discounting inside guardrails.**It beats a rep who panics and spirals from 20% to 34% off.It’s not just us. ICONIQ 2026: conversion is down across the funnel, so a slow form costs more than ever.Go incognito on your own site, fill out your own form, and fix what embarrasses you first.

We go deep on builds like this every week on The Agents. If you sell to operators who are actively deploying agents, that’s the room, and we’re taking a small number of sponsors: saastr.ai/media-sponsors.