Pinecone vs June
Side-by-side trajectory, velocity, and editorial themes.
Pinecone widens from vector DB to retrieval app platform with Marketplace and BM25.
Pinecone shipped two structurally significant launches in early May: a public Marketplace for building and operating knowledge apps directly on Pinecone, and full-text BM25 search via a typed document model that unifies dense, sparse, text, and metadata fields. Alongside, the company introduced a $20/mo Builder plan for solo developers and added Frankfurt and Singapore regions.
Pinecone is widening from vector database to managed substrate for retrieval-driven apps, covering both the storage primitive — vectors, BM25, and filters in one document model — and the surrounding application stack of templates, evaluations, and end-user chat. The Builder tier signals deliberate cultivation of solo developers as a top-of-funnel into the same platform.
Expect deeper opinionated tooling around Marketplace — more connectors, agent SDK glue — and a push to make hybrid retrieval the default rather than a separate code path. SDK coverage for the new document and full-text endpoints is the obvious next gap.
June's last visible push was a tight May 2025 B2B sprint — Custom Objects, SQL traits, PostHog integration.
June is product analytics for B2B SaaS, and the only visible release activity in the input is a concentrated four-week sprint in May 2025: SQL computed traits, PostHog as a data source, increased computed-trait limits, and the GA of Custom Objects after a two-month rollout. Each release is paired with small fixes (Slack alerts, HubSpot reverse sync) suggesting a stable maintenance cadence around the headline launches.
The May 2025 batch is internally consistent: every release widens what June can model (Custom Objects), how flexibly customers can compute on it (SQL traits), or how easily it slots into existing data plumbing (PostHog source). All three target the B2B-SaaS persona that wants more than user/account analytics. After this burst the changelog goes quiet in the input — it's not clear from the entries alone whether the product moved to a slower cadence, switched publishing channels, or paused.
The entries don't support a confident prediction about what comes next. If publishing resumes from the same direction, the obvious extensions are deeper integrations with reverse-ETL or warehouse-native sources and richer pre-built health-score templates on top of SQL computed traits.
See more alternatives to Pinecone →
See more alternatives to June →