Q1 2026 Platform Review

The platform shipped four major releases this quarter. Each release moved a distinct part of the stack: Q1.1 reworked the request router, Q1.2 added a vector cache between the API and the embedding service, Q1.3 rewrote the billing pipeline, and Q1.4 introduced a new admin console.

Architecture-wise, requests now flow: User → CDN (edge cache + SSL termination) → API gateway → Application server → either the vector cache (for AI features) or the relational database (for everything else). The application server is the focal piece — it owns routing, auth, and tenant resolution. The vector cache is new this quarter and is a Redis cluster with custom embeddings stored as binary blobs.

Revenue split for Q1: subscriptions 62%, usage-based billing 24%, professional services 9%, marketplace fees 5%. This is the first quarter usage-based billing crossed 20%, driven by the AI feature rollout in Q1.2.

Monthly active users grew from 41,200 in January to 47,800 in February to 53,500 in March. The growth in February tracks the launch of the admin console; March growth tracks the AI features going GA.

For Q2 priorities we evaluated 9 candidate projects on impact and effort. High impact / low effort: SSO support, audit log retention, and webhook reliability. High impact / high effort: multi-region deployment, custom model fine-tuning. Low impact / high effort (do not pick): a complete UI redesign and a homegrown analytics warehouse.

The deploy process now runs as: Engineer pushes to feature branch → CI runs tests + builds artifact → review approval gate → staging deploy → smoke tests → production canary (5% traffic) → full rollout. The canary stage was added in Q1.4 after a Q4 incident.
