Day 29: I Built an AI Sales Agent Because Cold Email Agencies Charge $5K/Month

Cold email works. I know this because I used to run sales for a B2B SaaS company. We had a dedicated outreach person whose entire job was building lists, writing sequences, and managing follow-ups. The cost: $4,800/month in salary plus tools. The output: maybe 15 qualified meetings per month.

When I went to start my own company, the math didn't add up. I couldn't justify a $5K/month headcount before I had revenue. But I also knew that without outbound, I'd have no pipeline. So I built Outpace instead.

Here's where we are 29 days in — real numbers, no polish.

Why Cold Email Is Broken (But Still Works)

The dirty secret about cold email: it has the best ROI of any outbound channel when it's done right, and most people do it catastrophically wrong. The two failure modes are opposite ends of the same spectrum.

Failure mode 1: The spray-and-pray blast. Buy a list, export to Instantly, add 5 generic follow-ups, send to 10,000 people. Your reply rate is 0.1%, your domain gets flagged, and you spend more time cleaning up bounces than having conversations. This is what most "cold email agencies" actually do.

Failure mode 2: Obsessive personalization that doesn't scale. Spend 45 minutes researching every prospect. Write a perfect, tailored email. Send 3 per day. You might get a 30% reply rate, but you're doing the work of a full-time SDR and your pipeline moves at 15 leads per week.

The right answer sits in the middle: personalized at scale. Research enough to be relevant, but automate enough to reach volume. This is exactly the problem AI is built to solve.

What Outpace Actually Does

The product is simple to describe and genuinely difficult to build. You tell Outpace who your ideal customer is — industry, role, company size, pain points. Then:

1. Research pipeline. Outpace pulls public signals — LinkedIn profiles, company pages, recent news, job postings — to build a prospect dossier for each contact.
2. AI email generation. Using Claude (our model of choice for quality), Outpace writes a unique cold email for each prospect. Not a template with field substitutions. An actual email that references something specific to that person or company.
3. Automated follow-up sequences. Three follow-ups at optimized intervals, written fresh for each contact based on the original email.
4. Domain warmup engine. The part everyone ignores. We manage sending domain reputation automatically — volume ramp, inbox-to-spam ratio monitoring, bounce handling.
5. Reply detection and reporting. Track opens, clicks, replies. Flag positive responses. Stop the sequence automatically when someone replies.

The goal: you describe who you want to reach, and meetings appear in your calendar. You never touch the outreach machine.

What Surprised Me About Building This

1. Deliverability is harder than the AI

The AI part — generating quality emails — was solvable in week one. The deliverability problem took three weeks and is still being iterated. Email infrastructure is a hostile environment: ISPs use machine learning to detect automation, domain reputation affects inbox placement across all major providers, and one spam complaint can tank a domain that took months to warm up.

I thought the moat was AI email quality. The real moat is the domain reputation management layer. That's unglamorous infrastructure work that has no shortcuts.

2. Founders want "someone to do it" more than they want a tool

When I talked to early users, I kept hearing the same thing: "I don't want another tool I have to manage. I want someone to just handle it." The product insight here is that the UX shouldn't feel like software — it should feel like delegating to a person.

This reframing changed how we think about the product. Every design decision gets evaluated through the lens of: does this feel like management, or does it feel like oversight? Management is bad. Oversight is fine.

3. The pricing sensitivity is not where you'd expect

I assumed the $49/month price point would be the main objection. It wasn't. The most common concern was: "Will this get my domain blacklisted?" People are more afraid of the downside risk than they are sensitive to the price.

This made sense when I thought about it. A damaged domain reputation means your entire company email — sales, support, internal comms — lands in spam. That's catastrophic. A $49/month SaaS that destroys your email infrastructure is a terrible trade.

The product implication: our biggest trust-builder isn't testimonials or features. It's a credible answer to "how do you protect my domain?"

The Numbers: Day 29 Reality

I'm going to be honest here because I think founders who share real numbers are more useful to the community than founders who only post highlight reels.

What's working:

• 200+ people have tried the beta product
• Free trial sign-up rate from landing page: meaningful but room to grow
• Users who complete setup are generating personalized emails within 10 minutes
• Zero domain blacklisting incidents in beta (this matters a lot)

What's not working:

• Free trial to paid conversion is lower than target — users love the output but haven't committed
• Email deliverability infrastructure needs more refinement for cold-start domains
• The onboarding flow assumes too much sales knowledge from non-sales founders

What I got wrong:

• I built the prospect generation feature before I fully understood the deliverability problem. Wrong order.
• I underestimated how much education work is required. "AI writes your cold emails" raises more questions than it answers.
• The landing page still doesn't adequately address the domain safety concern.

What's Next

The next two weeks are focused on three things:

1. Deliverability documentation. Clear, honest explanation of how domain warming works, what protections are in place, and what users need to provide. This is the conversion blocker.
2. Onboarding simplification. The current setup flow requires users to understand what an ICP is. Most founders don't think in those terms. We're rewriting it as: "Describe your ideal customer in one sentence."
3. First paying customer. The goal is straightforward. Ship the improvements above, get one person to pay $49, understand why they did, use that as the template.

Why I Think This Works Long-Term

Cold email as a channel isn't going away. The tools that exist today — Instantly, Apollo, Clay — are powerful but they require operators. You still need someone who knows what they're doing to run them. The cost of that operator is the same whether the tool costs $99/month or $2,000/month.

The addressable market isn't "companies that buy cold email software." It's "companies that need outbound pipeline but can't afford an SDR." That's every company under $2M ARR. There are a lot of those companies.

Outpace charges $49/month. The SDR it replaces costs $60,000/year. The math is clear. The question is whether the product earns enough trust to make that case. That's the work.

If you're building something in the sales tooling space, or if you're a founder who's thought about automating your outreach, I'd genuinely love to hear from you. What's blocking you? What would make you trust an AI to run your outbound?

Try it yourself — free 7-day trial, no credit card. The product is real, the emails it generates are real, and I'm actively iterating based on what I hear from users.