Hello Scott and Brian

BuyerSight's human-plus-AI promise lives in its prompts. I test prompts the way you'd test a campaign.

Before applying, I ran three versions of a research and first-line prompt on 25 real hotel developers, the buyers behind your Jet Hotel Solutions case study. Here's what changed between versions, and how I'd run that loop for every client.

James Praise · GTM Engineer

7.16/8blind rubric score for my third prompt version, up from 2.96 for a standard prompt

3.36%reply rate across 1,116 sends on my own outbound campaign, with a 1.25% bounce rate

27%lift in outbound reply rates from the Clay stack I built at FunnelEnvy

The Pitch: Where I'd Start

BuyerSight pairs human strategy with AI research and personalization, against tools that automate the whole job. That difference only shows when the AI writes lines a prospect recognizes as being about them, and this role owns that work: writing the prompts, running experiments and documenting what they teach in a knowledge base.

I'd start with exactly what I already drafted: a three-version prompt test on 25 hotel developers for the Jet Hotel Solutions segment, scored blind. Read the test →

From there, every client segment would get the same loop before a campaign goes live. Two or three prompt versions run on a sample of real prospects, get scored blind against a rubric, and earn a short knowledge base entry on what changed and why. The winning prompt becomes that client's default, and every send still gets a human read.

The Proof

The JET prompt test ran 25 real hotel developers through three prompt versions. A standard prompt scored 2.96 out of 8 and padded its lines with praise the sources didn't support. The third version scored 7.16 by building each line on one detail unique to the project, and it wrote nothing for the three companies with no hotel project.

At FunnelEnvy, I built the AI-enabled outbound stack on Clay that lifted reply rates 27% and enterprise meeting bookings 31%.

I also built the Prospect & Outbound Engine, a Python pipeline across 10+ sources with its own verifier and a preview tool that checks every email before it sends. On a home care campaign for MIA Acquisition, it took 1,796 scraped businesses to 1,116 sends at a 3.36% reply rate and a 1.25% bounce rate.

Let's talk about BuyerSight.

If this is the right problem for you, I'd like to talk about it directly.

Let's Talk My Resume