Chapter 50 : The Product Question
[Gardner Analytics Office. October 2014, Afternoon, Strategy Session]
The door-table had been upgraded. Not the table itself, still a door on sawhorses, but the chairs were real now, purchased from an office supply warehouse that delivered in bulk. Twelve chairs surrounded the table. Seven were occupied: Ethan, Sarah, Priya, Marcus, Diana (sales), Derek (sales), and James (the 7-rated engineer who'd turned down Hooli with a GitHub commit history).
Monica's voice came through the conference phone in the center of the table, her audio slightly compressed by the speaker's cheap driver but her authority fully intact.
"The GPT-1 model works," Monica said. "Everyone in this room knows that. The question isn't whether the technology is real. The question is what we sell."
Sarah had prepared a slide, projected onto the wall from a portable projector Marcus had rigged to a ceiling mount. Three columns:
OPTION A: Research API Sell model access to researchers and developers. Monthly subscription. Low touch. Scales with minimal support.
OPTION B: Enterprise Documentation Sell AI-powered documentation generation to businesses. Contract-based. High touch. Immediate revenue.
OPTION C: Consumer Chatbot Build a conversational AI product for end users. High risk. Viral potential. Long time to revenue.
Priya spoke first. She'd been at the company for six weeks and had already become the person whose opinion carried the most technical weight after Ethan's. Sarah had accepted this gracefully. Her 9.5 rating sat beside a 9's perspective instead of competing with it.
"Option A is the safest path for the technology. We expose the model through a clean API, let researchers and developers build applications on top, and collect usage data that improves our training. The revenue is subscription-based, predictable, recurring, scalable without linear headcount growth."
"The revenue is also tiny," Diana countered from the sales end of the table. She'd been the company's first sales hire, the 6-rated former SaaS rep who'd defied her Talent Resonance number by landing three pilot agreements through sheer tenacity. "Research API pricing is ten to fifty bucks a month per user. To hit a million in annual revenue, we'd need twenty thousand subscribers in a market that doesn't know we exist."
"Option B is where the money is," Sarah said. She tapped the Enterprise Documentation column. "Companies spend millions on technical writing. Documentation, internal knowledge bases, support articles, product manuals. It's expensive because it requires domain expertise and consistent quality. Our model can produce first drafts that technical writers edit rather than create from scratch. Cuts production time by sixty percent. We've run the numbers."
Monica's voice cut through. "Option B has my vote. Revenue is the milestone, and enterprise contracts are the fastest path. A single mid-size company could pay fifty to a hundred thousand annually for documentation AI. Ten contracts and we've hit seven figures."
"Option C is the future," Ethan said. The room turned to him. He'd been quiet through the first round of debate, listening, letting the team build the argument before adding his perspective. "A consumer chatbot, conversational AI that people interact with directly, is where this technology ultimately matters. Not as a writing tool. As an intelligence."
"That's a five-year vision," Sarah said.
"I know."
"We have a twelve-month milestone."
"I know that too." Ethan picked up the marker from the whiteboard ledge: the blue one, always the blue one. He drew a timeline on the wall beside the projector screen. "We ship enterprise documentation now. We prove revenue. We satisfy the milestones. And underneath, we keep building toward the consumer product. The enterprise revenue funds the research. The research produces the chatbot. The chatbot changes everything."
The room absorbed the dual-track strategy. Diana was already calculating contract values in her notebook. Derek was listing potential clients from the startup ecosystem. Priya was annotating the research API option, not as the primary product but as a secondary revenue stream that could run in parallel with enterprise sales.
Sarah studied the timeline Ethan had drawn. The dual-track approach was elegant but expensive, two product lines meant two engineering efforts, two marketing strategies, two support structures. The team was thirty-two people. They'd need fifty to execute both tracks.
"We focus on enterprise," Sarah said. "Build the documentation product. Launch in three months. The research API runs as a beta alongside. Consumer chatbot stays in the roadmap as a placeholder for post-Series B."
"Agreed," Ethan said.
Monica confirmed from the phone. "Agreed. I'll position the enterprise angle with James Wright for the next board update. Revenue from documentation AI is a story investors understand."
The strategy session ran another forty minutes. By the end, the whiteboard showed a product roadmap that Sarah had drawn in red, launch timeline, feature milestones, customer targets, revenue projections. The Gardner on the chalkboard downstairs was a sandwich. The Gardner on the whiteboard upstairs was a company that generated text for enterprises. In secret, it was building toward conversational AI that would leave the Siri comparison far behind. That comparison had haunted Ethan's first pitch meeting at Basecamp nine months ago. Next to this, it was a bicycle against a rocket.
Diana packed up her notebook. "I'm starting outreach tomorrow. I have twelve companies on my list. If I get three pilots in the first month, we're ahead of schedule."
"Get four," Sarah said. "Ahead of schedule isn't enough. We need to be so far ahead that nobody questions the timeline."