To build its internal "Cloudflare OS," the company set up an email address that employees thought was a powerful AI. It was actually a human team fulfilling requests and, more importantly, using the prompts to identify and systematically document the "jobs to be done" needed to train the real AI.

SaaStr's initial AI, a clone of founder Jason Lemkin for giving advice, unexpectedly received many questions about events and sales. This user behavior revealed a clear need for dedicated go-to-market AI agents, pivoting their AI strategy from a simple experiment to a core business function.

The quickest path to market is a pilot where you sell the desired outcome, not the software. Initially, perform the work manually with AI assistance behind the scenes. This validates customer value and pinpoints the most repeatable patterns to productize.

When Bespoke's chatbot broke during a holiday, the founder became the bot for a week. This revealed that less efficient, more conversational interactions significantly increased user engagement. This insight, contradicting the goal of pure efficiency, became a key product differentiator.

To avoid over-engineering, validate an AI chatbot using a simple spreadsheet as its knowledge base. This MVP approach quickly tests user adoption and commercial value. The subsequent pain of manually updating the sheet is the best justification for investing engineering resources into a proper data pipeline.

At HubSpot, a 100+ person chat team, unburdened by sales quotas, became the top lead source simply by answering user questions. They were trained to guide conversations toward leads without being 'salesy,' a model that can now be replicated and scaled with AI chatbots.

To demo his AI coaching platform before it was built, the founder voice-cloned a prospect, manually created a scripted video conversation, and presented it as a working product. This high-effort simulation convinced the leadership team and secured a $10,000 pilot.

To ensure product quality, Fixer pitted its AI against 10 of its own human executive assistants on the same tasks. They refused to launch features until the AI could consistently outperform the humans on accuracy, using their service business as a direct training and validation engine.

To build its internal AI system, Cloudflare set up an email address that employees thought was an advanced AI. A human team fulfilled the requests behind the scenes, allowing them to precisely map the company's key 'jobs to be done' before building the actual automation.