Cold outbound is one of the least forgiving places to prospect. There’s no existing relationship, no warm intro, and no reason for a prospect to care unless the timing is genuinely right.
We iterated on outbound campaigns consistently for years, and the numbers didn’t move. The issue wasn’t the messaging or the effort behind it — it was that we were showing up at arbitrary moments instead of meaningful ones.
We turned to Prospecting Agent to help fix that. It monitors accounts for relevant signals and drafts personalized outreach when it finds one. We built it for customers, then put it to work on our own pipeline. Here’s what happened.
Where We Started
At HubSpot, most of our automation work is reactive. We partner closely with sales to find the processes that are predictable and repeatable, then build automation into outreach where it helps both prospects and reps. Day to day, that looks like automated sequences firing at contacts already showing intent.
But nurturing people who are already engaging only gets you so far. So we also build outbound campaigns to generate net-new pipeline — reaching contacts whose intent is low as a way to create demand rather than wait for it.
Each month, we ran one dedicated campaign for the outbound audience: pick a segment, build a list, write the messaging, QA both, and send. That adds up to about 10 hours of work per campaign.
It was also selective by necessity. We could realistically reach 10-20% of a rep’s cold book in any given cycle. The rest sat untouched.
We experimented constantly, and conversion rates stayed flat. We were reaching cold contacts on our schedule, not theirs. No triggering signal. No way to tell us who was ready.
Sales worked it hard. But a process that only touches a fraction of available pipeline has a ceiling. We hit it.
What We Did
In April 2026, we ran a small pilot with one sales team. We enrolled nearly all of each rep’s cold book into Prospecting Agent, a significant jump from the 10 to 20 percent we’d been working manually.
The agent monitored those accounts continuously. When it found a timely signal, such as a funding round, an executive hire, a product launch, or a spike in research activity on our site, it generated personalized outreach automatically. No rep research. No message crafting from scratch. The agent’s job was to find the moment, not manufacture one.
Sales feedback was immediately positive. Reps noticed their cold books were being worked without manual effort, and the messaging landed because it was tied to something real, not a monthly send cycle.
The Results
Over the pilot, we saw more than a 10x increase in booked meetings compared to our manual process. Those meetings turned into deals, then pipeline.
That matters because our baseline wasn’t weak. We compared against our most targeted, carefully crafted campaigns and still doubled conversion.
We expanded to more regions, and conversion rates held.
We also introduced a human touchpoint during the pilot. When a prospect engaged with our content, a rep called. That combination — AI finds the signal, a human makes the call — is what moved our pipeline. Reps focused on the conversations worth having, not the research required to get there.
More than 700 sales reps now use Prospecting Agent. It finds a timely reason to reach out to 78% of enrolled contacts with no manual work from sales.
How We Weighed the Cost of Credits
We’ll be honest: This was new for us too. Enrolling a full cold book into an AI agent and letting it send is a real leap. We had the same questions our customers have: What does this cost, and is it actually worth it?
So we didn’t scale on faith. We started with one team, and we watched the numbers. When conversion doubled, we expanded — carefully, region by region — and watched again. The results held every time we widened the aperture. That’s what earned our confidence. Not a projection, but a pattern we saw repeat.
Careful experimentation is what turned credits from a cost we worried about into an investment we could stand behind.
What We Learned
Here are the key takeaways that helped us go from small pilot to global impact:
-
Translate best practices from your best sales reps. Before you begin, find out what messaging already works. Partner with your top reps to learn what resonates and what doesn’t, then use that to set expectations for the agent’s messaging.
-
Review outputs before you scale. Before expanding to new regions, we had real sales reps compare the agent’s messages with manually written outreach for tone and quality. The agent writes the copy; humans confirm it sounds right. Human review keeps the output from sounding generic.
-
Enroll broadly. The old model forced selectivity because research was manual. Prospecting Agent removes that constraint. Enroll your full cold book and let the agent decide when outreach makes sense, not whether it’s worth researching.
-
Let the signal drive timing. The difference between cold email and timely outreach is the moment, not the copy. We stopped sending on a schedule and reached out when something happened at an account. Audit your current outbound cadence and ask: Is each send tied to a signal or just a date on the calendar?
-
Add a human at the point of engagement. When a prospect responds to Prospecting Agent content, that’s when a rep calls. The agent warms the account; the rep takes it from there. Don’t skip this step. It’s what turns interest into pipeline.
How Prospecting Agent Changes the Way You Sell
The old model asked reps a question they couldn’t answer at scale: Which accounts are worth the research this month? The better question is one AI can answer continuously: Which accounts have a reason to hear from us right now?
We led with cold outbound because it leaves the least to work with. No existing relationship, no warm intro, no reason for a prospect to care unless the timing is genuinely right. Clear that bar, and the warmer scenarios only get easier. If Prospecting Agent could move cold pipeline, it could handle the rest.
But the same engine that watches for a buying signal and drafts a relevant message doesn’t only apply to cold accounts. It applies to the dormant leads sitting in every CRM — the “not now” conversations that went quiet. It applies to unworked MQLs, to sourcing a full buying group instead of a single contact, to any moment where relevance and timing decide whether outreach lands.
That’s the shift. Not better copy, not bigger lists. Better timing — at a scale no manual process can match. It’s true for cold outbound, and it’s just as true for the warm leads and dormant accounts most teams never get back to.
Prospecting doesn’t fail because salespeople aren’t working hard enough. It fails because relevance and timing are nearly impossible to get right at scale by hand. Let AI watch for the moment and free your reps to have the conversation, and both sides of the equation improve.
Marketing gets time back. Sales focuses where they’re needed. And the pipeline moves.