What's new
Changelog
Real, shipped updates to ZeroDB — no filler, no internal cleanup.
Subscribe via RSSTerms of Service published; Bucket Storage card now shows real data
Added a real Terms of Service page. Also fixed the dashboard's "Bucket Storage" card, which was stuck on "Not available yet" due to a field-name mismatch with the backend — it now shows your real per-project S3/Lakehouse storage once a usage snapshot has run.
One-click AI tool setup, real workspaces, and a smarter Knowledge Graph
Added one-click "Connect your AI tool" configs for Claude Code, Cursor, VS Code, and Windsurf, plus a Connection Details card on every project. Workspaces now have a real creation flow and switcher, with the Projects list filterable by active workspace. Knowledge Graph gained contradiction detection and resolution, entity-type (ontology) management, and a "Most Connected Entities" view. Postgres projects got schema, backup, and restore management in the console, and Cloud Functions got ready-to-run templates with example code.
Fixed fake vector dimension numbers and a broken Generate Embeddings call
Vectors and Memory Replay no longer show a fallback "768" or "1" dimension count when the real value is unknown — they now say so honestly. Fixed Generate Embeddings sending a single text string instead of the texts array the API expects, which was silently failing. Also added timeouts and a visible retry affordance to Lakehouse catalog loading, so a slow or failed load no longer looks like it hung forever, and fixed TurboQuant Compression's settings toggle so a backend no-op is caught and reported instead of showing a fake success message.
Console visual refresh: brand alignment and clearer information hierarchy
Brought the console's logo, accent color, and typography in line with the marketing site, and reduced visual density on the Dashboard and other pages for a clearer information hierarchy.
Plain-language labels across Vectors, Memory Replay, Knowledge Graph, RLHF, and API Keys
Replaced technical jargon with plain-language explanations throughout the console — Vector Dimensions, Memory Replay's Session/Agent ID and Role fields, Knowledge Graph relationships, API Key scopes, RLHF naming, and Semantic Search's Hybrid/Rerank options all now explain what they do in context, not just what they're called.
Fixed Memory Replay creation errors and surfaced workspace context on projects
Fixed a 422 error (with an unreadable error toast) when creating a new Memory Replay entry, and added a retry affordance when Lakehouse fails to load. The project selector now surfaces which workspace each project belongs to.
New blog section, more accurate pricing page, and a smoother checkout
Added a /blog section with curated ZeroDB posts. Fixed the pricing page's checkout flow and discount visibility, corrected a storage-number mismatch between the homepage and pricing page, and fixed the Feature Comparison table showing fabricated data instead of real capability information. Sign-up now defaults to the Register tab and names your pending plan when you arrive from a pricing link.
Fixed account registration and the 3-day trial checkout flow
Signing up now reaches Stripe checkout for the 3-day trial as intended. We also found and fixed a registration bug where new accounts could get stuck without a clear next step — if you hit this recently, please try signing up again.
Knowledge Graph and Lakehouse now listed among ZeroDB's core capabilities
The homepage capability grid now includes Knowledge Graph (GraphRAG) and Lakehouse (DuckDB SQL analytics) alongside Vector Search, NoSQL Tables, and the rest of the unified API — bringing the full lineup to 10 capabilities.
Pricing page redesign and billing-accuracy fixes
Refreshed the pricing page design and fixed the yearly-billing toggle so every plan shows its correct annual price, with savings messaging that matches the real numbers.
New pages for Functions, Lakehouse, and RLHF
Added dedicated documentation pages for ZeroDB Functions (auto_embed, auto_summarize, auto_graph event hooks), Lakehouse (DuckDB SQL analytics with no ETL pipeline), and RLHF data collection.
Site-wide visual refresh
Rolled out a refined visual design across the marketing site, added site-wide top navigation, a branded 404 page, and new onboarding pages (Quick Start, ZeroMemory, GraphRAG) plus SDK/integration docs for MCP, LangChain, LlamaIndex, and the React/Next.js/Python SDKs.