Pricing and token economy of AI features – the product aspect

Most of the product managers are excited about integrating GenAI capabilities into their products. Very few, however, ask the hard question early enough: What will this actually cost at scale - and will the economics still make sense? Usually, that question only comes up once the feature starts getting real adoption at scale… and someone in the organization receives the bill. At our upcoming meetup in collaboration with Workato we’ll tackle this extremely important, highly relevant, and still under-discussed topic with must-see talks designed to save you both money and a lot of operational headaches:

1. Token Economy: Building AI Products That Scale by Jane Vandale, Senior solutions consultant @ Workato: As AI moves from feature to infrastructure, token economics becomes a product decision—not a FinOps afterthought. Every inference call has a real marginal cost, and poorly designed agentic systems can quickly become expensive, opaque, and difficult to govern. This talk explores where token costs actually leak and introduces a practical framework for deciding what truly requires an LLM and what should be handled through fast, deterministic execution. Using real production examples and five core techniques—routing, packaging, right-sizing, filtering, and pre-fetching—we’ll show how product and engineering teams can significantly reduce token usage while improving reliability, transparency, and scalability. The key question for every AI feature: does this need inference, or does it simply need execution? 2. Ignorance Is Expensive You Have $100. Now What? by Daniel Shani, Product Manager @ Sharegain We built an AI feature that worked, and still couldn't ship it — the unit economics said no. This talk is about what we did next: three product decisions that cut our cost per answer by 20x, from decomposing the product so the model only handles what earns it, to shrinking what the model needs to know, to replacing document search with a knowledge graph. I'll also share the simple test we run before any AI feature enters the roadmap, which works on a fixed budget that never grows. Expect opinions about pricing models you may not agree with.

3. Credits Create Pressure: Why Your AI Pricing Model Is Working Against You by Gilad Livnat\, Prodict Lead\, Ex Monday.com Every AI product with a credit balance is fighting the same invisible enemy: its own pricing page. Credit-based pricing looks fair on paper - you pay for what you use. In practice, it quietly trains users to dread the button they're paying to press. The psychology behind this isn't new. It was discovered by accident about 150 years ago, in a place with no regulation, no banks, and a currency problem nobody planned to solve - and it's been hiding in plain sight in every "pay per use" product since. This talk breaks down the three behavioral biases baked into every credit system, shows exactly where they're costing you trust and engagement, and asks the question most product teams never do: is credit-based pricing solving your problem, or just moving it onto the user?

If you’re building AI features, shipping GenAI products, or planning to scale AI in production, this meetup will give you practical frameworks, hard-earned lessons, and actionable strategies from teams already operating at scale. We’re looking forward to meeting you there. ProductX.biz Team

Please register via the Luma registration page [here](https://luma.com/r0wvfgob). [https://luma.com/r0wvfgob](https://luma.com/r0wvfgob)

*The address will be shared after your registration is approved.