Find the signal
in noisy systems.

When clients ask "what does ChatGPT say about us?" the honest agency answer was usually a screenshot and a shrug. Every commercial GEO tool was expensive, narrow, or both, and none of them answered the strategic question: is our brand showing up in the answers our customers are getting?

The agency needed a defensible GEO capability across several anchor clients, each watching different signals. Buying a SaaS tool meant inheriting someone else's prompt set, their model coverage, and their idea of what mattered. Building meant proving the agency could ship something usable, and ship it fast.

I ran a structured build-vs-buy evaluation in parallel with a working prototype. The framework scored vendors against an honest list of agency requirements. The prototype proved the cheaper, more flexible path was real.

The monitor parses generative AI responses for brand mentions and citation patterns across the major AI surfaces, and rolls them into a dashboard a strategist can actually read. Model coverage was wider than the commercial benchmarks, including platforms most vendor tools don't track.

The first thirty days of a new utility account are where most of a customer's questions show up. Billing, programs, how to read the first statement. SCE needed a multi-channel onboarding journey to move new customers into its key programs. But if you over-communicate, you risk them tuning you out.

New SCE accounts produced a disproportionate share of contact center volume and depressed CSAT in the same window. The company had just stood up a CDP, and this was a chance to show how it could orchestrate a multi-channel campaign that adapted to customer segments and the actions they'd already taken.

I led the strategy and rollout for a CDP-driven onboarding pilot. We treated the first thirty days as a deliberate sequence rather than a one-time handoff. The journey adapted to each customer's segment and the actions they'd taken, and the key rule was simple: only message them about programs they hadn't enrolled in yet.

The work crossed CX, data, and digital teams. The CDP became the routing layer. The journey was the product.

Wildfire season turns a utility from a back-office service into an emergency communicator. Public Safety Power Shutoffs leave millions of customers asking the same three questions at once: is this happening, is this about to happen to me, and when do the lights come back on?

We had to help customers understand what was happening when the lights went out, and help them prepare for outages ahead of time instead of only reacting once one hit. We also had to satisfy strict regulation about what content has to be on the site and how it's presented.

I led the redesign around one question every customer was already asking: is this happening now, soon, or in the future? All content was organized within that structure, surfacing the most urgent information first, like the outage map, which alone drew more than 90% of the site's traffic. Now is what's true this second. Soon is what's coming and when. Future is what we know about resolution.

The framework set the rules for tone, content density, and what surfaced first on the page. The outage map led, with wildfire updates and program information organized behind it, so the site told one coherent story instead of contradicting itself.

TheStrat is a technical analysis methodology based on price action. It uses candle patterns across multiple timeframes to help traders and investors make decisions. Because it's built on universal truths of price action, it's an objective system.

TheStrat traders use up to six timeframes at once to triangulate setups. Doing that with off-the-shelf scripts means stacking five or six indicators and hand-configuring them per instrument, while duplicate and orphaned labels clutter the chart at the exact moment a trader needs to read it.

I built TheStrat Suite as a closed-source Pine Script v6 indicator that consolidates six timeframes into one workspace. The architecture follows a rule I landed on after the first prototype: only draw what should exist. The suite creates a label only when the signal state calls for it, instead of drawing everything and hiding what doesn't apply. The chart stays clean as the timeframe configuration changes.

The product has two tiers. The free Lite version sits at the top of the funnel: a single-timeframe view, without cross-timeframe continuity or alerts. The paid Suite handles the full multi-timeframe workflow: full timeframe continuity (FTFC), magnitude and exhaustion targets, stop placement, and the alert system. Distribution runs through TradingView, with subscription management on Whop.