The Investor Was Right. And Also Wrong.

At an investor sharing session recently, someone said something that stuck with me.

He was frustrated. Not with AI — with founders pitching AI. The pitch decks all sounded the same. Capability claims stacked on top of each other, demo screenshots, a slide about the bot. And at the end of it, he said something I keep coming back to: investors aren’t investing in bots. They’re investing in humans.

He’s right. He was also diagnosing the symptom, not the disease.

The real problem isn’t AI outreach. It’s context-free AI outreach.

Most AI-assisted BD fails for a simple reason. The agent goes into every conversation knowing nothing. No background on the contact. No understanding of what they’ve been thinking about lately. No awareness of what’s shifting in their market. It fires a personalised-sounding message built on surface data — a job title, a company name, maybe a recent post — and calls it intelligent outreach.

It isn’t. It’s a mail merge with better grammar.

What makes a human BD professional effective isn’t volume. It’s that they remember things. They read the trades. They notice when a contact’s company just announced a new hire or pivoted their positioning. They walk into a conversation with context, and that context shapes everything — what they lead with, what they ask, what they don’t say.

If you want AI to do what a good BD professional does, you have to give it the same foundation.

Context is infrastructure, not a prompt

This is where most implementations get it backwards.

Founders treat the conversation agent as the product. They spend weeks tuning the message, testing hooks, iterating on tone. And the agent still underperforms, because the problem was never the words. It was what the agent knew before it said them.

The stack that actually works looks like four things working together:

• A CRM spine that holds everything the agent has learned about a contact. Not just static fields — research outputs, conversation history, observed signals over time. The CRM isn’t a database. It’s the agent’s memory.
• A research agent that builds a profile on each contact before any outreach begins. Company context, recent activity, likely priorities, positioning in their market.
• A market signal monitor that tracks what’s moving in the contact’s world. Not in real time. Consistently enough that the agent’s knowledge base stays current rather than stale.
• A living knowledge base. In our case, a blog and internal wiki updated in roughly 30-minute sessions. Half observation, half commentary. Not polished content. Working notes. But those working notes become the agent’s understanding of the landscape it’s operating in.

When all of that is in place, the agent stops sounding like a bot. It sounds like someone who has been paying attention.

What the spike actually means

We ran a short outreach test recently. 50 contacts over two weeks. Roughly 30% booked a Zoom call.

That number surprised me. It’s almost certainly not the steady state — I’d expect it to settle around 10% over a longer window. But 10% is still well above the industry norm of 1-3% for cold outreach. The spike is the point: when every contact has been researched before any message is sent, and the architecture supports that research at scale, the floor is meaningfully different from what cold outreach usually means.

It isn’t evidence that AI outreach is magic. It’s evidence that preparation compounds. Every contact in the pipeline has been researched before a message goes out. Every message reflects something real about that person’s context. The agent isn’t winging it. It’s working from a brief.

That’s not different from how a good human would operate. The AI just means we can do it at a scale and consistency a solo operator couldn’t sustain manually.

Back to the investor

His frustration was real, and I think it points at something important for anyone building in this space.

The founders who pitch AI as the thing are going to keep losing the room. The founders who show what they built underneath the AI — the thinking, the architecture, the understanding of what makes a conversation earn trust — those are the ones worth backing.

The bot is not the product. The judgment that shaped the bot is.

That’s what investors are investing in. That’s what clients are buying. And it’s the difference between an AI system that closes meetings and one that just sends them.