I built every one of these myself: no engineering team, no long build cycle. Each one started as a real problem someone on the team had, and each one is still running in production today. Not a slide. Not a demo. The actual workflow, doing the actual work.
Most people talk about AI in marketing as personal productivity: faster drafts, quicker research. That’s real, but it’s not what I’m building. Connect AI to the right data and it stops being one person’s shortcut. A whole sales team gets a clearer picture of where to focus. Personalisation that used to need a much bigger team becomes possible at scale. Hours of manual analysis become a 15-minute workflow anyone can run. That’s the difference between a tool and infrastructure, and it’s what’s below.
17+ account executives across five APAC markets, each managing large account books, with no consistent weekly signal on what to prioritise. Intelligence was assembled by hand, inconsistent rep to rep, and dependent on one person. When that person was offline, nothing went out.
An account intelligence system that pulls intent signals from Demandbase and Salesforce, scores accounts by likelihood and timing, and sends each AE a personalised Monday brief: ranked MQAs, closed-lost reactivations, new buying signals, recent engagement. Regional leads get roll-ups. It runs without me, and the team always has it.
Started with 400+ accounts across ANZ, Japan, GCR, India, and SEA. The APAC SDR team ranked it their top tool for weekly prioritisation, and it’s since grown into a global hub, EMEA didn’t copy the model, they’re running on it.
Enterprise outreach for a major Shopify event needed real personalisation, not a template. That meant reading each merchant’s audit results, understanding their gaps, and writing copy tied to their commercial context, for every account. At scale, that was weeks of work.
A Cursor workflow that turns one audit dataset into 18 personalised merchant assets: flyers, LinkedIn DM copy, and call scripts, each mapped to that merchant’s score, pain points, and stage. Two to three weeks of work became two days.
Personalised outreach across the full account list, delivered ahead of the event, that led some of the day’s best conversations.
Regional sales teams needed more than a generic deck to open enterprise conversations: proof, specific to that market, that a merchant’s setup could be better. Building a live demo store, a real audit, and a video comparison for every region by hand wasn’t something any team could keep up with.
Co-built an AI-generated outbound experience: a live demo store, personalised AI audit results, and side-by-side video comparisons, with copy and imagery localised for each region. What used to be a bespoke build per market became a repeatable, AI-assisted production line.
Month 1: 6 demos live. ROI will be determined after six months.
Events, partner, integrated campaigns, and customer success each worked from their own tools and trackers, with no shared home for campaign assets, status, or playbooks. Work got duplicated, and nobody outside a given team could see what was live, or what could support sales outreach.
An AI-built hub that brings events, partner, integrated campaigns, and customer success into one shared home: campaign assets, status, and playbooks in one place.
40% adoption in the first month, across events, partner, integrated campaigns, and customer success.
The gap between AI that’s impressive in a demo and AI that a sales team trusts and uses every week is wider than most people expect. Here’s what I’ve found actually matters.
Not "how do I use AI?" but "what recurring workflow is slow, manual, or inconsistent and costs the team something meaningful?" The tool comes second. The problem comes first.
Generic AI output doesn’t earn seller trust. Every system here pulls from Salesforce, Demandbase, or internal data the team already believes. AI is the engine. Trusted data is what makes it credible.
If I can’t describe the inputs, outputs, owners, and review points on paper first, I’m not ready to build it. Automation makes a bad process faster. It doesn’t make it good.
AI speeds up research and synthesis. Judgment on customer sentiment, account strategy, and commercial decisions stays with a person. That’s the design, not a limitation.
Success isn’t "AI produced an answer." It’s "a seller read it, trusted it, and acted on it." If the output requires interpretation or cleaning before it’s useful, the workflow isn’t finished.
Done something more than twice? It becomes a workflow, a template, or a dashboard. The goal is infrastructure that runs whether I’m there or not.
Shipping isn’t the finish line. I watch who actually uses it and where it stalls, and change it based on that, not on my own assumptions. If nobody’s using it, I stop building on it. Otherwise it’s just building for the sake of it.