That voice up top is mine, cloned on Fish Audio. It answers from my real work record, and it won't make things up (I checked). Honestly, it's the kind of thing I'd want a developer to build in their first hour with your API.
Why Fish: there are quite literally a handful of people who have worked on this problem and really understand demand for voice AI products. I'm one of them. Leading B2B growth at a rocketship sounds amazing.
| YOU ASKED FOR | WHAT I'VE DONE |
|---|---|
| 8+ years in B2B, building a function | 10 years: ABM agency, Google, Deepgram. Took Deepgram's paid program in-house, ran it solo for ten months, built the operating system, then brought on the next agency. |
| Marketing to developers | Voice AI APIs at Deepgram. One "go to console" button took signups in a campaign from 4 to 84 in six days (we rolled it out everywhere). |
| Owning a pipeline number | Co-own the paid OKRs on SQLs, activations and paying logos. Q3 paying logos were tracking ~4x the target by mid-quarter. |
| Self-serve and enterprise | Built DocuSign's PLG program on Google Ads (+95% qualified trials). At Deepgram, moved bidding from signups to activations to business email (~58x the value). |
| Analytical depth | Wrote the forecast behind a ~2.4x budget. Moved spend off a keyword that never activated: 2.4% → 18.5% activation in ten weeks. |
| Modern stack | HubSpot, Salesforce, GA4, Looker, Dreamdata, Clay, n8n. I build my own tools with Claude Code (including this page). |
| Bias to ship | Five weeks with no direct manager: launched a programmatic channel, a newsletter sponsorship and a G2 program anyway. |
| NICE TO HAVE | |
| Voice AI / API-first | Paid launches for Aura-2 TTS, Voice Agent API and Flux. |
| Better-funded incumbents | Ran the program that took Deepgram #3 → #1 on G2 in Voice Recognition, ahead of Google. |
| APAC and Europe | Helped run our first German and Spanish webinars (50% live attendance vs. ~38% norm). Expanded paid into Japan, Korea, India and the Nordics. |
| Open-source credibility | Haven't run an OSS community. I'd build on fish-speech's, not pretend to it. |
My sneaking suspicion was that our best signup keywords were tourists: people expecting a consumer product who bounced once they realized they had to code. They were. I argued against my own headline metric, moved the money to developer intent, and activation went from 2.4% to 18.5%.
Work-email signups were ~17% of paid-search volume and over 90% of the revenue. So that's what we bid on now.
Moved them to value-based bidding: +160% and +53% conversion performance YoY. It became a framework for other enterprise accounts.
A landing page test I called negative. An incrementality test I called unmeasurable. A growth number I corrected down. If a claim doesn't hold up, we change what we assert.
Based in Denver, and very open to regular travel to SF.