The CEO of an AI Agents company we genuinely admire reached out recently and asked why we didn’t select them as a vendor. It was a fair question — they have a great product.
The honest answer? Their VP of Sales argued with us. Wouldn’t do any work up front. Wouldn’t give us access to the features we needed.
And most importantly — wouldn’t commit to helping train our agent. That was the real issue.
So we went with someone else. And honestly, the competitor we picked, their software probably wasn’t any better. In fact, it was less feature rich.
But they did the work and got the AI agent working. That mattered far more.
We Now Have 20+ AI Agents Running SaaStr
Let me share what criteria we actually used for picking which vendors made the cut.
We picked the ones that helped us the most.
That’s it. That’s the whole framework.
Not the ones with sales folks that wanted to “get on a call” before showing us anything.
Not the ones bashing their competitors on every demo.
Not the ones sending 100 LinkedIn InMails a week.
The ones that actually rolled up their sleeves and helped us get our AI Agents running. This includes the FDE team at Salesforce, who were exceptional partners throughout the process.
We’ve Rejected Leading Vendors — Not Because of Product Quality
Here’s something that surprised even us: We said no to two leading AI vendors — not because their products weren’t good, but because their sales teams couldn’t explain deployment details or fought against our requirements.
One vendor’s sales team didn’t know their own product deeply enough to answer basic integration questions. Another was actively combative when we pushed back on their standard implementation approach.
That cost us months of lost time. But here’s the lesson: Sales team quality predicts post-sale success. If they fight you or can’t explain things before the contract, imagine what happens after.
The vendors that won? They showed up ready to work. They helped us understand what was possible. They did the upfront investment to prove they could handle our complexity — a decade of SaaStr data, massive integration requirements, and zero engineering resources on our end.
The Deployment Gap Is Real
Here’s what I’ve learned after deploying 20+ agents across our operations:
The greatest AI Agent in the world — without top-tier help deploying it — is not that valuable for many complex workflows.
Read that again.
You can have the most sophisticated AI agent on the market. The best underlying models. The slickest interface. But if you can’t get it deployed properly into your existing workflows, trained on your specific data, and integrated with your systems? It’s just expensive shelfware.
You need both:
- A trusted vendor with a product that actually works
- AND a team committed to helping you get it live
At least for the complex stuff. The workflows with significant training requirements. The integrations that touch multiple systems.
Our Results (The Receipts)
We’re not theorizing here. Here’s what our AI agents have actually delivered:
AI Inbound (Qualified + Agentforce):
- $1,010,000 in closed-won revenue
- $2,500,000 currently in pipeline
- 71% of October closed-won deals came from AI-qualified inbound (historic average was 29-34%)
- Response time dropped from 24-48 hours to under 2 minutes — even at midnight
AI Outbound (Artisan SDRs):
- 19,326 messages sent over 6 months
- 11-43x the volume of human SDRs
- 6.67% overall response rate (vs. 2-4% industry average)
- Warm campaigns hitting 12.13% positive response rates
- An AI SDR booked a six-figure sponsor meeting on Saturday at 6:02 PM
Delphi (Digital Jason):
- 139,000+ advisory conversations
- Now making product recommendations, reviewing VC pitch decks, drafting comp plans
The Investment:
- ~$500K+ annual across all AI agents
- 30% of our Chief AI Officer’s time managing agents
- First 1,000 emails were manually reviewed before we could trust the system
The difference between good and great? We’re the #1 performing customer for both Artisan and Qualified across their entire customer bases. Not because the tools are magic — but because we invested heavily in training and daily optimization.
Most B2B and SaaS Companies Haven’t Adapted Here
This is the part that surprised me.
We’re in 2025. AI agents are the hottest category in enterprise software. Everyone is talking about autonomous workflows and agentic AI.
And yet most older B2B and SaaS companies are still running the same sales playbook from 2015:
- Gate everything behind a demo
- Qualify before helping
- Make customers prove they’re “serious” before doing any work
- Treat pre-sales engineering as a cost center to minimize
Meanwhile, the vendors winning in the AI agent space? They’re doing the opposite. They’re leading with value. They’re helping prospects get agents running before the contract is signed. They’re investing heavily in customer success and deployment support.
The Brutal Truth About AI Agent Deployment
Here’s what no one tells you:
So many AI SDR + GTM implementations fail. Not because the technology doesn’t work — but because companies set it and forget it.
We went through 47 iterations to stop our AI from being too aggressive on pricing. It takes about 30 days of daily training to get it right. Our first 1,000 emails were manually reviewed. Today we still spend 20-30 minutes daily spot-checking.
All agents have hallucinated. All of them. The key isn’t achieving perfection — it’s building systems to catch errors before they reach customers.
Your AI works 24/7, but your humans don’t. We had to build triage systems to avoid information overload. Without filtering, agents create decision fatigue, not efficiency.
The companies that understand this — that deployment support is the product — are the ones we chose.
The New Playbook for Selling AI Agents
If you’re building or selling AI agents, here’s what we’ve learned as buyers:
1. Help first, sell second. The vendors that won our business were the ones that helped us deploy a proof-of-concept before we committed. They showed us what was possible with our actual data.
2. Your deployment team is your competitive advantage. Product parity is coming fast in AI. What’s harder to replicate is a world-class team that can get complex agents live in weeks, not months.
3. Stop gatekeeping features. When prospects ask for access to capabilities they need to evaluate your product properly, say yes. The vendor that argued with us about feature access lost a six-figure deal.
4. Pre-sales is marketing now. The work your team does before the contract is signed is the most important marketing you’ll ever do. It’s proof that you’ll be a good partner after the deal closes.
5. Sales team quality = implementation quality. If your sales team fights customers or doesn’t know the product deeply, run. We’ve learned this the hard way.
The CMO Story (Why This Matters Beyond Vendor Selection)
This same principle applies to how executives are learning AI.
A CMO I’ve worked with for years — I got her her last two jobs — reached out recently. She’s ready for her third role.
I had to be direct with her: “I don’t have anything for you right now.”
Why? She hasn’t deployed an agent herself. She doesn’t know how the training went, what issues came up, or how to iterate on prompts.
I told her: “Go deploy an agent. Tell me how it worked. Tell me how the training went. Tell me the issues you ran into. Come back to me, and I’ll have two jobs for you.”
The executives who are getting hired right now? They’re the ones who did the work themselves — hands-on keyboard, not just encouraging their team to try Perplexity.
AI Agents Aren’t Just Software
We’re in a moment where AI agents can genuinely transform how companies operate. But there’s a massive gap between “AI agent” as a concept and “AI agent deployed and creating value.”
The vendors that close that gap — the ones that help customers get live, not just get on a call — are the ones that will win.
And the ones still running the old playbook? They’re going to keep losing deals to competitors who understand that in the age of AI agents, helping is the new selling.
And that VP of Sales who argued with us and lost the deal — along with all the referrals that would have come from it?
He’s gone now.
