Niching down

Some forks in the road are hard. You build a love for what you build and it's hard to decide to kill what you love.

The time has come to kill copywriting on Wafrow and just focus on automated images. There are 3 reasons why I'm making this call:

1. All recent customers have bought into wafrow for its image capability. The translation product is not deeply integrated into customer's existing workflows. I'd either need to build

   • extensive developer SDKs / tooling and
   • copy versioning with maker / checker capability to match the leaders or not play this game at all. I've picked the latter

2. My edge in solving the copywriting problem is not high. I've upgraded the underlying LLMs around 8 times in 5 months, added audience personas and refined my input layer significantly but wafrow is not an R&D shop building its unique LLM. My effort leads to a 10-15% better output, not 100% better which could compel an existing user of another tool to switch

3. The future of automated images is better placed. In a world where core image generation APIs are having a 'ghiblified' moment, using that output as one layer and deterministically adding someone's name, a discount banner, etc. is still an amazing marketing opportunity.

If you're a wafrow customer, this blog post is clearly not a surprise. Hit me up on slack in case you need any help during the migration phase.

Onwards 🚀