Rob Ferguson watched competitors copy Automatic’s app icons and RDO’s rotated icon, and says Jev-style features reached 95–97% parity within a week. He argues the durable moat is direct customer understanding and service investment, not the feature itself. That burden is harder for competitors to replicate than a quick copy.
Rob Ferguson tells founders to burn through Anthropic, OpenAI and AWS subsidies while learning their intelligence needs, then come to Fireworks when inference spend justifies customization. He points to Cursor distilling a coding-only model for better cost per intelligence and speed. That transition is the path to model independence.
Rob Ferguson says clickable AI prototypes raise the quality of product conversations, but the worst teams stop there. He calls 'done' the most seductive word in English and says the real work is 50–60 iterations to make it reliable. Teams must define done before showing the prototype.
Rob Ferguson traces software from literal 'not hardware' to SaaS and now to 'work': teams sell the actual outcome on the other side. That requires forward-deployed people who understand hidden contracts, customer data and quality expectations before writing requirements. This is the most human part of AI product work.
Rob Ferguson says Fireworks keeps zero data retention and couldn't calculate a discount large enough to justify training on customer data. He warns closed labs discount what your data is worth when absorbing it, and people underestimate what they give away. That asymmetry matters when choosing model partners.
Randy Silver, drawing on music editing, says the best artists produce far more than should ever be published. As AI lowers the cost of making things, choosing and concentrating on the right output becomes the critical skill, not volume. That editorial judgment is the differentiator.
Rob Ferguson says features are far too easy to copy, so management must restructure teams to engage customers directly. He warns against sitting in the design pantheon and says the durable thing is the relationship on the other side, even with seasonal businesses. Forward-deployed teams are the management response.
Rob Ferguson says 10,000 GPUs represent roughly $1 billion, and a small model may need 80-plus GPUs plus replication. Because idle time drives up costs for everyone, Fireworks buys across 20 clouds and runs serverless to keep clusters busy. That operational load is the real barrier at the high end.
Rob Ferguson says Fireworks had already published how to train a classifier for $11, yet Jev broke through with cookbooks, community and a clear point of view. Competitors copied the feature quickly, but not the attention or ongoing engagement. That launch work is the harder barrier.
Rob Ferguson cites an advised startup that saw cold email open rates fall from 4% to zero because recipients assume AI-generated spam. He says rising above the noise is much harder now, and formerly useful channels simply disappear. This makes distribution strategy more urgent for founders.
Rob Ferguson says the state of the art is never one model; models have affinities such as JavaScript versus Python. In 'oracle modeling,' teams run every permutation of tasks and find a mixture wins, so the harness and UI context matter too. That makes measurement across combinations essential.
Rob Ferguson cites reinforcement learning finding 4–6% better video compression after PhD-level work plateaued, because the outcome was easy to score. He advises breaking work into small verifiable parts; if you can measure it, you can improve it with AI. That is the great secret to AI, he says.
