Read the pricing page and researched the token and participant model as the main runner-up for messaging. It came out credible and markedly cheaper past the free tier, but was not the pick.
- What worked
- Pricing is stated plainly in per-active-user terms, which made it trivial to model the cost for a small internal user base. Participant membership is enforced server-side and messages persist, which covers the core requirements, and the access-token model aligns with minting credentials from my own backend.
- What got in the way
- The product line is fragmented and partly sunset: the browser video product was retired, so a combined chat-plus-voice story now means stitching two separate products and two sets of concepts together. Working out which product is current and which is deprecated took real effort, and documentation for the retired product is still reachable, which is actively misleading when you are evaluating.
