Invisible Technologies is one of a growing field of AI vendors selling into large enterprises, and its own marketing team has become a live test case for running marketing as an AI-agent operation. In a recent episode of Unicorn Marketers, we sat down with Angela Winegar, CMO of Invisible Technologies, to get specific about how her team decomposes workflows into AI, logic, and human-judgment steps, where AI-driven customer research quietly produces false positives, and why she thinks marketing is turning into a relationship-first, sales-adjacent function. Topics Discussed: A concrete framework for splitting any 10-step marketing workflow into roughly 3 generative-AI steps, 5 deterministic steps, and 2 steps that require human judgment Why AI sentiment analysis on sales call transcripts can mistake a checked-out "yep, sounds good" for genuine buyer engagement, and how Invisible's research tooling forces a human review of the exact flagged moments How Invisible's marketing agents are structured, with each marketer owning their own agent stack inside a shared repo rather than a centralized build A weekly product launch cadence that is roughly 90% automated from PRD to cross-channel content, built to keep pace with near-continuous feature releases Why Angela reframes the AI-and-headcount question around time reallocation rather than team size, moving hours out of lead-scraping and data cleaning and into designing harder-to-replicate, high-value events The short shelf life of a viral marketing tactic, using a real-time example of a $6,000-a-day plane-banner campaign she expects marketers to saturate within one or two quarters Why Angela expects marketing to increasingly resemble a "forward-deployed" function, with marketers spending real time on relationship-building work that used to sit with sales or PR GTM Lessons For B2B Marketers: Decompose workflows with a specific ratio, not a vague split. Angela's rule of thumb: in a 10-step marketing workflow, expect roughly 3 steps where generative AI adds real value, 5 steps that are purely deterministic logic, and 2 steps that genuinely need human judgment. Use that ratio as a starting audit, not just a general "keep humans in the loop" principle, before assigning agents to a process. Don't trust AI's read on customer sentiment without spot-checking the source clips. Angela's team found that AI scraping sales calls can tag a disengaged "sounds good" as a positive signal, so they built tooling that surfaces the exact 10 to 20 transcript moments behind each AI-flagged insight for human review. If you're running AI over call recordings or support tickets for sentiment, build in a step that resurfaces raw clips, not just aggregate scores. Treat customer PR relationships as core GTM work, not a favor. Angela spends real time helping customer PR and comms leads build joint stories, because Fortune 100 comms teams are fielding pitches from what she calls "a billion AI startup vendors" and have no default reason to cover any one of them. She frames this shift as marketing starting to resemble a forward-deployed function, borrowing the term from the "forward deployed engineer" role she calls the hottest job in Silicon Valley right now. Build these relationships proactively, before you need the coverage. // Sponsors: Front Lines — We help B2B tech companies launch, manage, and grow podcasts that drive demand, awareness, and thought leadership. www.FrontLines.io The Global Talent Co. — We help tech startups find, vet, hire, pay, and retain amazing marketing talent that costs 50-70% less than the US & Europe. www.GlobalTalent.co // Don't Miss: New Podcast Series — How I Hire Senior GTM leaders share the tactical hiring frameworks they use to build winning revenue teams. Hosted by Andy Mowat, who scaled 4 unicorns from $10M to $100M+ ARR and launched Whispered to help executives find their next role. Subscribe here: https://open.spotify.com/show/53yCHlPfLSMFimtv0riPyM