One data point stuck with me this week: 43% of SaaS companies already charge for work performed, not access. The rest of the day brought launch after launch—Gemini reaching 1 billion users, Alibaba with a new model, Meta entering the coding-agent race. But I set today’s edition aside to talk about money, because money is what decides what survives.

In brief

  • In 2026, 43% of SaaS companies already use some form of hybrid pricing, moving away from the old seat-based access model to charge for completed tasks or delivered outcomes.
  • The projection is that this will pass 60% of companies by the end of the year.
  • When AI only answered questions, charging for access to the tool was reasonable. Now that it executes work end to end, the value is in the completed task, not the login.
  • AI pricing decisions are moving into product, where teams need to understand cost per model call, value generated per task and how that becomes a plan that makes sense for the customer.

When the value is the completed task

Every time I review a product’s pricing model, I reach the same conclusion: price is not an implementation detail; it is core strategy. It is an AI product management decision, not an adjustment to leave until the end of the cycle.

I saw a data point this week that reinforces this. In 2026, 43% of SaaS companies already use some form of hybrid pricing, moving away from the old “seat-based access” model to charge for completed tasks or delivered outcomes. The projection is that this will pass 60% of companies by the end of the year.

It makes perfect sense when you stop to think about it. When AI only answered questions, charging for access to the tool was reasonable. Now that it executes work end to end, the value is in the completed task, not the login.

This connects directly to the world I work in. On the credit product side, we have been thinking about price in terms of risk and outcome for a long time, not only access. Seeing this logic arrive so explicitly in AI products is a sign that these two worlds are moving closer together faster than I expected.

One point I especially liked: AI pricing decisions are moving into product because technology changes too quickly to leave this only to finance or sales to sort out later. The people building the product need to understand cost per model call, value generated per task and how that becomes a pricing plan that makes sense for the customer.

For people who work in product, technology or business, this shift is worth watching closely. For AI Product Managers, pricing decisions today shape product architecture tomorrow.

If you are curious and want to understand this change better, here is the article.

The rest of the radar

Gemini app passes 1 billion monthly users — Shows that adoption speed has become the central competition metric among AI platforms. Read more

OpenAI enables unlimited chats on free ChatGPT and launches GPT-5.6 Luna/Sol — Signals a price/access war in the free tier, putting pressure on monetization strategies and paid plans. Read more

ChatGPT integrates restaurant bookings through Yelp, OpenTable and Resy — A concrete example of agentic commerce: AI executing end-to-end transactional tasks inside the chat. Read more

OpenAI launches “Sign in with ChatGPT” in beta as an identity layer — A new cross-platform login standard that could become infrastructure, just as “Sign in with Google” became for SaaS. Read more

Alibaba launches Qwen3.8-Max, its most capable model yet — Expands low-cost, high-capability model options outside the US, relevant to multi-model routing and cost. Read more

Google expands the apps connected to Gemini (Granola, Otter.ai, Wix, Ticketmaster and more) — Expands Gemini’s native integration ecosystem, opening surfaces for third-party products to plug into AI. Read more

Gemini 3.6 Flash and 3.5 Flash-Lite become generally available in production — Low-cost, low-latency “flash” models are the foundation for AI features in high-volume products. Read more

Meta launches Muse Code, a coding agent powered by Muse Spark 1.2 — Another entrant in the coding-agent race, with a cheaper tier in exchange for training data. Read more

The LLM observability and evaluation market grows and consolidates (Langfuse, LangSmith, Braintrust, Arize) — LLM observability and evaluation are becoming production prerequisites, not differentiators. Read more

That is what stayed with me today. More tomorrow.