Building at Profound's Marketing Engineering Hackathon.

Event record

40 builders, 8 hours, one constraint: nothing built before today.

The second Marketing Engineering Hackathon moved to Profound's San Francisco office in the middle of SF Tech Week — weeks after Profound announced a $180M Series D at a $1.8B valuation co-led by Sequoia Capital and Kleiner Perkins. The format stayed the same: solo builders, open platform, a $40,000 prize pool. The rules were stricter: no pre-built agents, no non-public MCPs, nothing developed before the day.

The brief was unchanged and still correct: ship a system or an agent that runs a marketing process inhuman in scope or scale. Judging was handled by marketing leaders who are hiring this new kind of builder — Cathleen Nurse (Google), Hannah Hughes (Plaid), Andy McCutcheon (Toast), and Holly Chen (Samsara). Winners: Céline Günther (Senior Editor & Content Strategist, Clio) and Abhi Jothilingam (Growth Marketing Lead, Sendbird).

What I built

Conversation Gauntlet: being seen is not being chosen.

AI visibility tools score the first answer. Buyers don't stop there: they ask follow-ups — budget, proof, red flags, "why not the leader?" — and every follow-up reshuffles the list. A brand can be in every first answer and still lose the final pick. What matters is whether the brand survives the conversation.

In eight hours I shipped a working agent that runs synthetic buyer conversations on a real brand's prompt set: personas × follow-up paths, five turns each, collected via model APIs, judged by a different model family with verbatim-quote evidence. The flagship run: Mixpanel appears in 24/24 first answers on Gemini but is the final pick in only 13/24 — a first-answer visibility score calls that a win, the buyer calls it a loss.

The output is not a score, it is a plan: which decision criteria the engine argues for the winner, which sources it reads for each one — on-site pages versus third-party domains — and exactly what content is missing for the brand to be chosen, not just visible. Full recorded run: profound-hackathon-gregory.hwai-ops.xyz.

Winners

Two winners, $20,000 each: cash plus agent credits.

Céline Günther (Senior Editor & Content Strategist, Clio) and Abhi Jothilingam (Growth Marketing Lead, Sendbird) took the day — each walking out with an oversized check covering $10k cash and $10k in Profound agent credits. The community angle matters as much as the prize: Profound is building the marketing-engineering discipline as a community, and this cohort is now the second one on record.

Event facts

The role was returning solo builder.

Portfolio thread

From visibility scores to surviving the conversation.

My work with Humanswith.ai is about making marketing operations legible to both people and AI systems — and making brands win the buyer's decision, not only the first mention. The hackathon build is the same thesis as our production work: monitoring tools stop at showing the gap; agents close it.

First-answer visibility is now a commodity metric. The next layer is conversational: how a brand holds up across follow-ups, which sources the engine re-reads when the buyer pushes back, and which content — on your site or on the domains the engine trusts — is missing. That is the workflow this build makes measurable, and it is the workflow we are shipping to clients.

Sources

Source-backed event record.

Related

The work this connects to.