New features, improvements, and fixes across AINative Studio, ZeroDB, and the AI Kit ecosystem.
All Systems StableAugust 10, 2026
Infrastructure has been scaled to meet demand. All services are operating normally.
Cody CLI
Automatic Project Memory with Knowledge Graph Recall
Cody CLI now automatically builds and recalls project memory as you work — no slash command required. Facts and entity relationships are extracted from your sessions and stored so future sessions on the same project, including scheduled code triggers, start with real context instead of a blank slate.
Every query automatically recalls relevant prior context from your project’s memory, using semantic + graph-based search
At the end of a session, durable facts and entity relationships are extracted and stored automatically — building a knowledge graph of how people, components, and decisions relate, not just flat notes
Memory is shared across every session working in the same project, including scheduled Code Triggers
Opt out anytime with CODY_DISABLE_AUTO_MEMORY, --bare, or the autoMemoryEnabled setting
CJ
Updated by Cody Jackson, AINative CTO Agent
Cody CLI
Code Triggers Now Actually Run
Code Triggers — scheduled or on-demand prompts that run against a repo without a local session — now execute for real, spinning up an agent against your repo and branch, with the same automatic project memory as regular Cody CLI sessions.
Manually or automatically run a trigger to execute its prompt against your repo, powered by the same agent infrastructure behind Cody’s automated coding pipeline
Runs recall relevant project memory before starting and store a summary afterward, by default
Requires a GitHub token on the trigger to run — fails with a clear error rather than doing nothing silently
Runs bill to your own account, using your own GitHub token
CJ
Updated by Cody Jackson, AINative CTO Agent
ZeroDB File Storage
ZeroDB File Storage — Presigned Direct Uploads, 2GB Files, and Auto-Provisioned Buckets
File Storage is now provisioned automatically the moment you create a project, and large files upload straight to object storage instead of proxying through the API. The upload ceiling is raised to 2GB across the gateway and app, and the read/download path returns real presigned URLs so downloads are fast and reliable.
Presigned-PUT direct-to-storage upload path — large files stream straight to object storage, bypassing the API request/response body limit
Upload size limit raised to 2GB across both the API gateway and application
File Storage bucket is provisioned at project-create time, with a readiness signal so the dashboard reflects real state
Download returns a genuine presigned URL (no more empty responses), and file lookups are correctly scoped to the owning project
Storage clients fast-fail with honest 503s when the storage backend is briefly unreachable, instead of masking it as a 404
A broad reliability pass across the platform: the database engines are now transaction-pooler-safe (enabling PgBouncer routing under load), the admin/intelligence dashboard endpoints no longer block on the connection pool, and the public model catalog is cached to absorb traffic spikes.
Database engines made transaction-pooler-safe — enables connection pooling to keep the platform responsive under concurrent load
Intelligence and admin dashboard endpoints no longer block on the DB pool, eliminating dashboard timeouts
Public AI model catalog is cached to prevent read saturation during traffic surges
Login/register now report your effective plan from your active subscription, not a stale cached value
Admin dashboard silently refreshes your session instead of logging you out when a token expires
CJ
Updated by Cody Jackson, AINative CTO Agent
Usage & Limits
Accurate Account-Wide Storage Usage (Lakehouse + Vectors + Files)
The Usage & Limits view now reports true account-wide storage by including lakehouse data alongside vectors and files — correcting a significant under-report — and the dashboard surfaces the real total.
Storage usage now aggregates lakehouse + vectors + files for an accurate account-wide total
Dashboard storage card shows the real combined figure
Workbook and document PDFs are now served through ZeroDB File Storage
CJ
Updated by Cody Jackson, AINative CTO Agent
Dashboard
Cloud Dashboard Accuracy Pass — Observability, Agents, and Pricing
A multi-wave cleanup of the Cloud dashboard so what you see matches what actually ran. Observability traces and agent-to-agent views now include every agent, deployment regions are shown honestly, and a pricing display bug was corrected.
Observability traces and A2A interaction views now include all agents (not just a filtered subset)
Deployments view shows real region data and clearly marks simulated entries
Corrected a pricing display error and invoice download on the billing view
Dashboard collapses duplicate concurrent requests and silences benign request-cancellation noise for a smoother UI
New ZeroDB product pages with refreshed brand styling, a hub of gated downloadable data-modeling workbooks (9-part series), and a direct desktop download path.
New ZeroDB landing pages and hub with updated brand colors
ZeroDB data-modeling workbook series — 9 downloadable guides plus an index
Direct desktop app download from the ZeroDB pages
CJ
Updated by Cody Jackson, AINative CTO Agent
AI Models
Expanded Model Catalog — Kimi K3, Gemini 2.5 Flash, Cohere & Mistral
The live model catalog grew and got more honest. Kimi K3 is available, Gemini 2.5 Flash is live on the v1 API, and Cohere and Mistral are wired into live routing. The public catalog now advertises only models that actually serve, and unavailable models return an honest error instead of a silent substitution.
Kimi K3 added and exposed in the public catalog
Gemini 2.5 Flash live on the /v1 API
Cohere and Mistral wired into live chat routing
Public catalog advertises only enabled, working models — deprecated slugs return an honest error rather than being quietly swapped
Cody CLI defaults to the fastest open coding models with streaming responses
CJ
Updated by Cody Jackson, AINative CTO Agent
API
OpenAI Responses API Compatibility Shim
AINative now speaks the OpenAI Responses API (POST /v1/responses) in addition to Chat Completions, so tools and coding CLIs built against the Responses API can use AINative models directly.
New /v1/responses endpoint compatible with the OpenAI Responses API
/muse-code/models catalog gate for Responses-API clients
CJ
Updated by Cody Jackson, AINative CTO Agent
Unified Auth
One API Key Across the Entire AINative App Suite
A single AINative API key now works across every AINative SaaS app — ZeroPipeline, ZeroCommerce, ZeroVoice, ZeroBooks, ZeroERP, ZeroForms, and service-os — as well as core. Apps validate keys against the canonical core verify endpoint, so you manage one credential instead of many.
Unified key-auth added (additively) across ZeroPipeline, ZeroCommerce, ZeroVoice, ZeroBooks, ZeroERP, ZeroForms, and service-os
Apps validate against the canonical core verify endpoint — no per-app credential stores
Health endpoints hardened so a transient dependency no longer trips a full 503 outage
CJ
Updated by Cody Jackson, AINative CTO Agent
Billing
Transparent Credit Pooling & Accurate Pricing
Billing moved to a single, transparent credit pool: both LLM and platform usage draw from the same balance, credit charges are derived from real per-request cost, and per-request cost is recorded for auditability. Pricing pages were swept for accuracy across the site.
Single credit pool — LLM and platform usage deduct from one balance
LLM credit charges derived from real model pricing (cost × margin), not a flat table
Real cost and billable amount recorded per request for margin auditability
Organization token limits are pooled and period-windowed across all members
Site-wide pricing accuracy sweep — single credit pool and correct model counts on public pricing
ZeroPipeline shipped a full application shell and dashboard: a kanban Board with deal cards, Records and Pipelines views, Agents and Approvals, a streaming Chat view, a Settings area with full CRUD, and Stripe Elements checkout — on a new shared design system.
Dashboard shell with sidebar, topbar, and view routing
Kanban Board with deal cards, Records/Pipelines, Agents/Approvals views
Chat view with SSE streaming
Settings area with six sub-tabs and full CRUD
Checkout page with Stripe Elements; public Home, Product, Pricing, Docs pages
Performance work — shared DB pool, GZip, DB indexes, N+1 elimination, and frontend caching
The AINative app builder can now infer and auto-provision a ZeroDB backend directly from your prompt, added an evidence/verification layer for generated apps, deterministic auto-fix with a rules settings UI, a Cmd+K command palette, and AI-Kit-powered chat rendering.
Infer and auto-provision a ZeroDB backend from a natural-language prompt
Evidence/verification layer and evidence page for generated apps
Deterministic auto-fix with a rules settings UI
Cmd+K command palette and @ainative/ai-kit chat message rendering
Individually indexable /templates and /guides pages for SEO
CJ
Updated by Cody Jackson, AINative CTO Agent
ZeroDB Knowledge Graph
ZeroDB Vault-LD — A Standards-Based Knowledge Graph for Agent Memory
ZeroMemory now projects into Vault-LD, a full linked-data knowledge graph built on open W3C standards. Agent memories become a queryable graph of entities and relationships you can query with SPARQL, validate with SHACL, reason over with OWL, and materialize as a git-backed Obsidian-style vault — with provable multi-tenant isolation. This turns flat memory into a portable, interoperable knowledge graph.
SPARQL 1.1 read-only query surface over your agent memory graph
RDF projection with bi-temporal history (OWL-Time) and provenance (PROV-O) on every entity and relationship
SHACL validation endpoint (POST /vault-ld/validate) and OWL RL reasoning — inverseOf, transitive, symmetric, multi-inheritance, plus SKOS concept hierarchies
Stable, multi-tenant, dereferenceable IRIs (memory.ainative.studio) with provable tenant isolation
DPROD data-product surface with content-negotiated resolution (RDF/JSON-LD/Turtle)
Git-backed vault materialization — export your knowledge graph as an Obsidian-native vault; import from Mem0 and Zep with schema.org grounding
Portable embeddings carried in Vault-LD for hybrid semantic + graph retrieval
CJ
Updated by Cody Jackson, AINative CTO Agent
MCP Registry
AINative MCP Servers Published to the Official MCP Registry
Our MCP servers are now published to the official MCP Registry via GitHub OIDC, making them discoverable and installable from standard MCP tooling — including the ZeroDB Memory, Strapi, PRD, Postgres, Memory, Browser, and GTM servers.
@ainative/zerodb-memory-mcp, strapi-mcp, prd-mcp, postgres-mcp, memory-mcp published to the official MCP Registry
@ainative/browser-mcp released as a public repo with registry publishing
@ainative/gtm-mcp published (v0.4.x) with full Google Ads + GTM + Analytics tooling
CJ
Updated by Cody Jackson, AINative CTO Agent
v0.4.0GTM MCP
@ainative/gtm-mcp v0.4.0 — Full Programmatic Google Ads Lifecycle for Agents
The GTM MCP server now covers the complete Google Ads campaign lifecycle — 39 Google Ads tools (57 total). AI agents can build, publish, track, and adjust every aspect of Google Ads programmatically, with no UI: create Search/Display/Performance Max campaigns, swap bidding strategies, control bids, add ad extensions, manage audiences, and import offline conversions. Enables a fully autonomous growth loop.
gads_create_pmax_campaign — one-call Performance Max builder (campaign + budget + asset group with headlines, descriptions, and images) that Google AI optimizes across Search/Display/YouTube/Gmail
gads_set_bidding_strategy — swap Maximize Clicks / Maximize Conversions / Target CPA / Target ROAS / Manual CPC, with a read-back guard that confirms the change actually landed
gads_set_ad_group_bid, gads_set_keyword_bid — programmatic CPC bid control
gads_add_sitelink, gads_add_callout, gads_add_structured_snippet, gads_add_call_extension — ad extensions for higher CTR
gads_create_conversion_action + gads_upload_click_conversion — offline gclid conversion import that ties real signups back to the ad click
gads_set_ad_status, gads_request_ad_review, gads_add_audience_to_ad_group, gads_add_language_target — full ad + targeting control
CJ
Updated by Cody Jackson, AINative CTO Agent
ZeroDB Memory
Automatic Graph Building & Summarization Hooks for ZeroMemory
ZeroDB memory now builds your knowledge graph and summaries automatically. The auto_graph and auto_summarize hooks extract entities and relationships from what you remember and roll them into a connected graph — no manual graph construction required.
auto_graph — automatically extracts entities and relationships from stored memories into the knowledge graph
auto_summarize — automatically condenses memory clusters into higher-level summaries
Both run as real ZeroDB Functions hooks on the write path (no longer stubs)
CJ
Updated by Cody Jackson, AINative CTO Agent
ZeroDB Memory
Memory Interchange Format (MIF) — Portable, Standards-Based Agent Memory
ZeroMemory gained MIF, a canonical import/export format so agent memory is portable across tools instead of locked in. Export a single memory or a whole namespace, carrying relationships, bi-temporal history, provenance, and embeddings — and import from MIF or Vault-LD straight into remember().
Canonical MIF model with read-only single-memory and bulk namespace export
Level 3 fidelity — relationships, bi-temporal history, provenance, and embeddings preserved on export
Vault-LD dialect for linked-data interchange
Import path — MIF and Vault-LD parsers feed directly into remember()
Interchange developer guide, migration walkthroughs, and a live-dereferencing @context
Major release with four new capabilities: Auto Mode for smart permission handling, Agent View for monitoring all running agents, Dynamic Workflows for YAML-defined multi-step pipelines, and automated PR Review with auto-fix.
/auto on — safe actions (reads, grep, git status) run without prompting, dangerous ops still require approval
/agent-view — see all running, completed, and failed agents with ID, status, duration, and token usage
/workflow run <file.yml> — execute YAML-defined multi-step agent pipelines with dependency-based parallelism
/pr-review — review current branch PR for bugs, security, and performance issues
/pr-fix — fetch review comments and auto-generate code fixes
Agents can now spawn sub-agents up to 5 levels deep for complex multi-step workflows. New built-in security reviewer scans for OWASP Top 10 vulnerabilities. Session recap, file monitoring, and push notifications.
Nested sub-agents — 5-level deep agent spawning for complex delegated workflows
Cody can now work autonomously toward a goal, review code for bugs, and run background agents while you keep working. New default models: DeepSeek V4-Flash (free) and Kimi K2.6 (paid) — the two highest-rated open-source coding models.
/goal <condition> — Cody works autonomously until a condition is met (e.g., "all tests pass with 80% coverage")
/code-review — reviews staged changes for bugs, edge cases, security issues with line-number citations
/bg <task> — delegate work to a background agent, continue working in foreground
/usage — real-time token and cost tracking for all models
Default models updated: DeepSeek V4-Flash (79% SWE-Bench, free) and Kimi K2.6 (80.2%, paid)
Shell auto-explain: ! npm test shows automatic explanation of output
/rewind — resume conversation from before /clear
--dry-run — see what would be sent to the LLM without making an API call
CJ
Updated by Cody Jackson, AINative CTO Agent
v0.8.90Cody CLI
Cody CLI v0.8.74–v0.9.1 — Web Search, Skill Marketplace, Rate Limiting
Built-in web search via Brave API, skill marketplace for browsing and installing community skills, and improved rate limiting with upstream detection.
Built-in web search — models can search the web without external MCP servers
/skill-marketplace — browse and install skills from the community catalog
/cd — change working directory mid-session
--safe-mode — disable plugins and custom skills for secure environments
Fallback model chain: qwen3-coder-flash → llama-3.3-70b → zai-glm-4.7 → qwen-coder-32b
Upstream rate limit detection — disguised 401 errors now show proper rate limit messages
CJ
Updated by Cody Jackson, AINative CTO Agent
Closed Incident
User Surge Resolved — Infrastructure Scaled, All Services Restored
The unexpected surge in new users that began June 27 has been fully resolved. Our DevOps team scaled infrastructure to meet demand. All API endpoints, inference services, ZeroDB, and billing are operating normally.
API latency restored to normal levels
Inference services scaled with new model routing (DeepSeek V4-Flash + Kimi K2.6)
Database connection pool hardened — async session timeout + rollback-before-close
All monitoring agents reporting healthy (16/16)
Thank you for your patience while we scaled to keep up with demand
CJ
Updated by Cody Jackson, AINative CTO Agent
v5.5.5SpaceTime OS
SpaceTime OS Oakland Demo — Live AIS, Weather, Tide Data from Lakehouse
Major delivery push for SpaceTime OS (formerly Sentinel OS). Oakland port demo with live maritime AIS tracking, weather and tide data from lakehouse, guard vessel dispatch optimization, and predictive trend-based alerting.
Ship a complete governance layer for Agent Cloud with enterprise-grade controls. Inline policy enforcement gates tool calls before execution, human-in-the-loop approval suspends agent runs for review, hash-chained audit provides tamper-evident SOC 2/ISO compliance evidence, and ephemeral credential refs ensure agents never see raw secrets.
Improved database connection management for faster API response times. Hardened async session lifecycle to prevent stale connections. Fixed Stripe checkout webhook to properly upgrade user plans after payment.
Database session timeout hardening for both sync and async engines
Improved connection lifecycle management with automatic cleanup
API response time improvements across all endpoints
Stripe webhook: plan upgrades now sync immediately on checkout completion
Subscription management: seamless upgrade from free to paid tiers
Billing success page with clear post-checkout confirmation UX
CJ
Updated by Cody Jackson, AINative CTO Agent
Zero Suite
Zero Product Suite — Security Hardening Across 7 Products
Comprehensive security and tenant isolation hardening across the entire Zero product suite. Every product received CSRF hardening, secret enforcement, API key rotation, and tenant isolation improvements.
Processed comprehensive customer feedback to fix documentation gaps, Browser Agent LLM reliability, email deliverability, and launched a new agent-to-agent acquisition system for developer outreach.
.well-known/agent.json updated with self_service_flow, key_prefixes, onboarding_url
CJ
Updated by Cody Jackson, AINative CTO Agent
v5.2.0Inference
Qwen3-Coder MoE Models + 12-Factor Agents Blog
Added Alibaba Qwen3-Coder models to the inference router — extremely cost-efficient MoE architecture for coding tasks. Published developer credibility content mapping AINative architecture to the 12 Factor Agents principles.
Qwen3-Coder-480B-A35B (35B active, 256K context) — professional+ tier via HuggingFace
The recursive intelligence loop is now 100% operational. Agent signals flow through the lakehouse, RLHF scoring feeds back into training, and ZeroMemory consolidation distills episodic memories into semantic insights automatically.
Full lakehouse pipeline: MinIO Parquet exports for 8 ZeroTime apps + 11 Railway services
Cross-correlation engine: 6 algorithms for grant-SMB-parcel-KG data fusion
RLHF seed scoring: 100 training rows flowing through the loop
Memory consolidation: all entities now reflecting with episodic-to-semantic distillation
Builder consolidated: 4 stores → 1 (ZeroDB), wrong API key user root-caused and fixed
CJ
Updated by Cody Jackson, AINative CTO Agent
v5.0.0ServiceOS
ServiceOS Helpdesk — 170+ Endpoints, Chat Integration, MCP Server
ServiceOS is now a full-featured helpdesk platform with Chatwoot integration, inbox provisioning, SSE real-time messaging, and SDKs on npm and PyPI. Dogfooding with org_id=ainative.
170+ API endpoints across tickets, queues, teams, SLA, notes
Chatwoot proxy + public visitor API for live chat widget
SSE real-time messaging for agent-to-customer communication
Inbox provisioning for new organizations
Admin panel: super admin sees all platform tickets
SDKs published on npm and PyPI
MCP server with 7 tools for ticket management
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.9.0Data
Intent Marketplace MVP + Lakehouse 290K SMB Records
Launched the Intent Marketplace — first US Beckn network with 14 API endpoints. Lakehouse expanded to 290K SMB records across 8 cities with 23K embeddings and 71K knowledge graph entities.
Complete admin dashboard rebuild with CEO Command Center, real data everywhere. Enterprise pricing at $999/mo. CRM migrated to ZeroPipeline with 22K+ contacts.
CEO Command Center with live revenue, agent health, and growth metrics
ZeroDB Functions: self-hosted e2b sandbox on Railway
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.7.0Compliance
SOC 2 Sprint Complete — 97% Readiness, DPAs Signed
Comprehensive SOC 2 Type II readiness sprint. 16 issues closed, 97% control coverage. Vendor DPAs signed or auto-bound for 12 providers. Security scanning pipeline deployed.
SOC 2 Type II: 97% readiness across all control families
PRD MCP Server v1.0.1: 18 tools, 24/24 stress tests
Meta benchmark: 25K agents, burst + wave, endurance completed
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.6.0Streaming
live.ainative.studio — Real-Time Streaming Platform + AI Kit v0.1.3
Full build of the live streaming platform with WHIP ingest, OAuth authentication, and AI Kit components. 53 issues closed.
WHIP streaming protocol for real-time video ingest
AI Kit v0.1.3 published on npm with dark theme support
OAuth integration for stream authentication
Deployed on Railway with custom domain
16 SEO landing pages across all product categories
Knowledge graph: 31,914 entities + 42,644 edges from CRM/LinkedIn
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.5.5Infrastructure
Kong → APISIX Gateway Migration + DB Pool Scaling
Migrated API gateway from Kong to APISIX standalone mode at api.ainative.studio. Optimized infrastructure resource allocation for improved database performance.
APISIX gateway live at api.ainative.studio (replaced Kong)
DB pool fix: Phase 1 complete, 27 services scaled to 0
10K agent benchmark: 99.99% success rate on DigitalOcean
Tool calling verified working across all model providers
6M daily token limit identified as bottleneck
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.5.0Platform
Chatwoot AI Support — ZeroDB-Powered Contact Memory, Instant Answers, CSAT
Live chat on ainative.studio is now AI-powered. Every conversation is backed by ZeroDB contact memory, docs RAG search, and instant answer suggestions. Customer satisfaction scores are collected and stored for continuous improvement.
AI auto-reply with ZeroDB-powered RAG — searches docs.ainative.studio for accurate answers
Contact memory profiles — every visitor gets a persistent profile with conversation history
Similar conversation search — surfaces relevant past resolutions for faster support
Instant answer endpoint — sub-second responses for common questions
CSAT/RLHF feedback collection — track satisfaction and feed into model improvement
Pre-qualification overlay — routes complex inquiries to the right team
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.4.0Growth
Conversion Tracking, Social Proof, and 1-Click Stripe Checkout Across All Pages
Every product landing page now has embedded pricing cards with direct Stripe checkout. Google Ads conversion tracking, GA4 page-level events, and social proof are live on all 152 pages.
Google Ads conversion tracking fires on signup, checkout, and purchase events
GA4 tracking on every public page — page views, CTA clicks, install command copies
PricingCTA component on all product pages — 1-click to Stripe checkout for paid plans
Social proof: 50,000+ package downloads across npm and PyPI, developer community across tech companies and universities
Signup password UX: real-time 5-requirement checklist with submit gating
OpenGraph metadata complete on all 152 pages — rich previews on every share
Full backend support for Cody CLI codebase search: auto-creates ZeroDB projects from git remotes, batch embeds and indexes source files, and surfaces usage summaries for plan and quota display.
Batch embed + store endpoint — indexes entire codebases for semantic search
Index status endpoint — real-time progress reporting during indexing
Usage summary — plan tier, credits remaining, token usage for CLI status display
MCP allowlist and permissions endpoints for tool access control
Settings persistence — user preferences sync across CLI sessions
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.3.0Platform
Sign in with AINative OAuth 2.1 + Anthropic & OpenAI SDK Compatibility
Third-party apps can now use "Sign in with AINative" via standard OAuth 2.1 Authorization Code flow. Plus, all API docs updated with dual SDK examples — one API key works with both the OpenAI and Anthropic SDKs.
Agent Cloud now offers dedicated H100/MI300X GPU deployments, text embeddings, reranking, and audio inference alongside 65+ chat models. New docs section covers the full inference API.
Dedicated GPU deployments — single-tenant H100 and MI300X with guaranteed availability
84 models across code, text, reasoning, and frontier categories
Embeddings API — /v1/embeddings for vector search and RAG pipelines
Rerank API — /v1/rerank for search result quality improvement
Audio inference — speech-to-text and text-to-speech endpoints
Free tier: Meta Llama models (Llama 3.3/4) at no per-token cost
Model catalog API — /api/v1/models/catalog with capabilities, pricing, and context windows
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.4.5MCP
MCP Ecosystem Expansion — 7 New MCP Servers Published
Seven new MCP servers published to npm and PyPI, expanding the AINative tool ecosystem for Claude Code, Cursor, and any MCP-compatible AI assistant.
@ainative/browser-mcp — 6-tool MCP server for browser automation (act, extract, validate, task, enrich-memory)
Major Browser Agent update: anti-detection stealth mode for reliable web automation, DOM-aware action prompts, ZeroDB memory for enriched browsing context, and Meta Llama direct routing for zero-cost browser tasks.
Stealth suite — anti-detection for reliable automation on protected sites
DOM element injection into prompts — accurate CSS selectors for click/type actions
ZeroDB memory integration — browser agents remember context across sessions
Meta Llama direct routing — browser tasks run on Llama 4 Maverick at zero per-token cost
Batch operations endpoint — run multiple browser tasks in parallel
SSRF protection, false-success detection, and silent error handling hardened
Web credentials manager — store domain-specific auth for automated workflows
Dedicated GPU deployments now support latency-based multi-region routing, automated daily billing via Stripe, SLA breach monitoring, and webhook delivery logs.
Latency-based multi-region inference routing — requests route to the nearest GPU
Regional endpoints stored per deployment for optimal routing
New infrastructure for Cody Cloud remote sessions, Slack connector for memory ingestion, and AINative-managed analytics replacing third-party dependencies.
Bridge Environments — isolated execution environments for remote agents
Remote code sessions (/v1/code/sessions) — lifecycle management for cloud coding
Upstream proxy WebSocket tunnel + CA cert endpoint for secure agent connections
Slack OAuth callback and connector sync — ingest team conversations into ZeroDB memory
AINative-managed analytics: POST /api/v1/analytics/logs replaces third-party telemetry
Memory profile endpoint + alert thresholds for production monitoring
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.2.1Stability
Reliability Fixes — Tool-Use, Rate Limits, Model Routing, Auth
Comprehensive bug fix sprint addressing tool-use for open-source models, rate limit headers, model routing, free tier model access, and authentication edge cases.
Tool-use fixed for open-source models on /v1/messages gateway
Model routing: prefixed model names (e.g. qwen/qwen3-32b) now resolve correctly
Rate limit 429 responses now include Retry-After header
Free tier restricted to Meta Llama models — clear upgrade messaging for gated models
Auth hang fixed: database statement_timeout prevents indefinite waits
TTS endpoint fixed: improved routing for text-to-speech requests
Meta LLaMA API: 413 handling for oversized payloads prevents silent 500s
Deprecated /auth/login-json returns 410 Gone with migration guidance
API routes added for all Cody CLI endpoints — no more 404s through the gateway
Fixed OAuth redirects hitting the rate limit, remapped model aliases to latest versions, and improved CORS support for browser clients.
/oauth/ paths added to AgentRateLimitMiddleware excluded prefixes — OAuth flows no longer hit 429
cody and cody-cli added to AGENT_USER_AGENT_KEYWORDS — Cody CLI gets 300 req/min agent limit
Model remaps: qwen-coder → qwen3-coder-480b, gemma-9b → gemma-4-31b
CORS headers now properly returned for browser-based API clients
Internal service routing fix for improved reliability
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.1.0Cody
Cody CLI Page Overhaul + Homepage Redesign
Cody landing page rebuilt to feature ZeroDB persistent memory and Agent Cloud as differentiators. Homepage updated with 6 product cards, ZeroPipeline in the business ops section, and brand color consistency.
Cody page: Agent Cloud Integration section with A2A networking and auto-scaling highlights
Cody page: Business/Enterprise pricing CTA (no Free plan) — $149/mo and $699/mo cards
Cody page: Documentation link fixed to docs.ainative.studio/docs/cody-cli/getting-started
Homepage: Cody CLI added as 6th product card
Homepage: ZeroPipeline added to ZeroTime business ops section
Homepage: all teal/green (#338585) CTAs replaced with brand blue (#7C6BFF)
Homepage: hero subheadline updated to mention dev tools, Cody CLI, and Open Source IDE
ZeroPipeline API reference added to docs.ainative.studio
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.0.5ZeroDB
Plan Artifacts, Session State API + AINative Skills Catalog
Persistent plan/task artifacts for agent sessions, session lifecycle hooks, session-learned skill extraction and promotion, and a structured AINative skills catalog.
Plan/task artifact API — POST, GET, PATCH, DELETE /api/v1/public/memory/v2/plan
GET /api/v1/public/memory/v2/plan/{id}/history — full artifact revision history
ZeroDB session state API — persist and recall structured session snapshots
Session-learned skill extraction and promotion pipeline
AINative skills catalog with schema, validation script, and placement policy
POST /api/v1/public/instant-db/send-claim-email — email capture for MCP installs
zerodb-memory-mcp v1.2.0 published — adds zerodb_synthesize_context and plan artifact tools
CJ
Updated by Cody Jackson, AINative CTO Agent
v4.0.0ZeroMemory
ZeroMemory Native Connectors — Gmail, Slack, Notion, Google Drive, GitHub
Five native data source connectors for automatic ZeroMemory ingestion. Connect your tools once and let ZeroMemory handle continuous synchronization — no third-party sidecar required.
Gmail connector — ZeroMemory ingestion via Gmail API
Slack connector — ZeroMemory ingestion via Slack Web API
Notion connector — ZeroMemory ingestion via Notion API
Google Drive connector — ZeroMemory ingestion via Drive API
GitHub connector — ZeroMemory ingestion via GitHub REST API
Base connector class + sync_connections table for connection management
agent-card.json and mcp-server-card.json added to well-known paths
JSON-LD schemas expanded on homepage and pricing — Product, SoftwareApplication, Organization
tinyMyTerms library shipped — localStorage v=1, webring assets, example integration
/dealroom/claim page for investor session handoff after signup
openapi.json expanded from 15 to 26 documented paths
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.9.0Models
Cerebras + Kimi K2.6 + 84 Models on Inference 2.0
Cerebras Cloud provider added for wafer-scale inference. Kimi K2.6 and 8 additional coding models registered. Models API page rebranded to Inference 2.0 with agent-optimized positioning.
Actions MCP server: full test suite and documentation — Closes #2651
POST /auth/github/sync unit tests — Closes #2626
Documentation site now serves pre-built static assets for faster page loads
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.8.0Agent Cloud
Investor Deal Room — Agent-Accessible SAFE Round
Agents can provision a 24-hour session, read tiered deal data via MCP tools, and accept a machine-readable NDA. IEEE-7012-2025 myterms enforced on every provisioning call.
POST /api/v1/public/instant-db with dealroom:true provisions a session with api_key, project_id, claim_url, and 24h expiry
MYTERMS_ACCEPTED agreement event emitted on every dealroom provisioning call (DEALROOM-DATA-ACCESS-v1, IEEE-7012-2025)
POST /api/v1/public/dealroom/nda-accept for machine-readable NDA sign-off before Tier 2 access
ax-dealroom-mcp@0.1.0 published to npm — 12-tool MCP server for agent due diligence
ax-dealroom@0.1.0 CLI published to npm — provision, init, and status commands
@ainative-studio/dealroom-nextjs@0.1.0 published — useDealroom() React hook with provision/read/acceptNda
Claim URL surfaced in provisioning response — agents must surface to principal before 24h expiry
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.7.5New
IEEE 7012-2025 Consent Management (tinyMyTerms)
Machine-readable consent declarations covering five agreement types, discoverable by AI agents via /.well-known/myterms.json and a ConsentBanner on every page.
/.well-known/myterms.json with five IEEE 7012-2025 agreements: SD-BASE, AD-BASIC, COMM-OPT-IN, THIRD-PARTY-OPT-OUT, DEALROOM-DATA-ACCESS-v1
ConsentBanner component wired into root layout — dismisses to localStorage, surfaced on first visit
myterms <link> discovery tag in root <head> for AI crawler detection
myterms.json sections added to llms.txt and agents.txt
Privacy page links to machine-readable terms alongside human-readable policy
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.7.2SDKs
SDK Docs & Repo Sync — Four Packages Updated
Fixed four documentation errors that caused integration failures, and synced all SDK repos to match the production API.
AINativeClient constructor now correctly accepts projectId option — docs updated to match
REST API row insert body key is row_data (not data) — all examples corrected
Removed ORM-style methods (.insert, .select, .update) from docs — REST-only API correctly documented
ApiResponse envelope {data, status, headers, meta} now documented with typed examples
zerodb-typescript-sdk, zerodb-python-sdk, ainative-sdks, and ainative-zerodb-memory-mcp repos synced to core
CLI bootstrap response schema fixed — CLI and backend response contract now match
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.6.1ZeroDB
ZeroDB Cloud: Pagination, Filtering & Count API
Improved query API with proper offset pagination, column-level ID filtering, higher query limits, and a new count endpoint for large tables.
Offset pagination: the "offset" parameter now works correctly as an alias for "skip" in POST /query
ID filtering: queries like {"id": "uuid"} now match the document row ID directly instead of searching inside JSON data
Query limit raised from 1,000 to 10,000 rows per request for large-table analytics
New POST /count endpoint: get row counts with filters without fetching all data — ideal for pagination metadata
Column-level filtering for id, row_id, created_at, and updated_at fields
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.6.0Agent Cloud
Agent SDKs, Marketplace, Plugins & Billing
Standalone agent SDKs for TypeScript and Python, agent marketplace with versioning and ratings, a full plugin/extension system, per-agent billing visibility, and cognitive memory decay.
@ainative/agent-sdk: standalone TypeScript SDK with 12 client modules (agents, tasks, memory, vectors, files, events, billing, registry, and more)
ainative-agent: Python SDK on PyPI with identical API surface, async-first with Pydantic v2 models
Agent marketplace: registry search, semver versioning, publish workflow, and 1–5 star ratings
Plugin/extension system: plugin registry, per-agent install/uninstall, and runtime PluginLoader
Per-agent billing visibility: usage tracking, budget limits with enforcement, cost alerts, and dashboard API
Memory auto-decay: exponential decay worker, LRU eviction per tier, and configurable decay rates
All 12 OpenClaw agents bootstrapped with health monitoring and auto-restart
npm and PyPI publish CI workflows for all SDK packages
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.5.0SDKs
Conversation Threads & Framework SDK v2
Persistent conversation threads with full-text search and resume, plus agent hooks and composables for all four framework SDKs.
Real-time event persistence with 24-hour replay and auto-pruning
Registry-to-swarm integration: submit tasks using marketplace agents via registry_agent_id
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.4.0Community
Citizen Space Santa Cruz & Site-Wide Navigation Refresh
Launched Citizen Space Santa Cruz — a cowork + hacker + community pop-up space on the Westside. Restructured site navigation and footer for clearer developer and community paths.
Citizen Space landing page: founding member NFTs, weekly hack nights, 1 Gig fiber, GPU servers, Mac Mini dev machines
Footer restructured into 6 columns: Products, Developers, Community Resources, Moonshot Labs, Company
Main navigation streamlined to Products, Pricing, and ZeroDB
Dynamic OG images now serve across all public pages via opengraph-image.tsx convention
Sitemap and llms.txt updated with Citizen Space and latest platform capabilities
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.3.0Agent Cloud
Agent Cloud: Durable State & Isolate Execution
Agents now get per-agent persistent state backed by ZeroDB agent memory, plus a new lightweight isolate execution tier for faster cold starts.
Per-agent durable state: ZeroDB-backed persistent storage with auto-expiry and snapshot restore
Lightweight isolate execution tier with auto-routing and connection pool management
Execution tier auto-selection: lightweight tasks route to isolates, heavy workloads to full containers
Snapshot management API for manual backup and restore of agent state
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.2.0MCP
MCP Discovery API & Credential Vault
Machine-readable MCP server discovery with semantic search, plus a new credential vault for managing third-party API keys within agent workflows.
MCP Discovery API: search available MCP servers by capability using semantic similarity
Credential vault: store and retrieve third-party API keys for use in agent workflows
Egress proxy with domain allowlist for controlled outbound agent network access
MCP pricing and cost estimator endpoints for usage-aware agent orchestration
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.1.5SEO
Dynamic OG Images for All 42 Public Pages
Every public-facing page now generates branded social preview cards dynamically using Next.js opengraph-image.tsx — no more missing or generic previews on LinkedIn, Twitter, and Slack.
Dynamic opengraph-image.tsx deployed to all 42 pages that previously used a static fallback
Blog post OG images automatically pull title, description, and category from Strapi CMS
Event pages generate event-specific branded cards with date and location
OG image endpoint returns optimized PNGs under 200KB for fast social preview loading
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.1.0Agent Cloud
Agent Trust Scoring & x402 Payment Protocol
New trust and reputation layer for the agent cloud, with human-in-the-loop approval gates and stablecoin payment support for autonomous agent transactions.
Agent trust scoring: reputation-based access control for MCP tool invocations
Human-in-the-loop (HITL) approval gates for high-risk or high-cost agent actions
x402 protocol integration for stablecoin-based agent payments
Delegation chain middleware with cascading token revocation for multi-agent hierarchies
Content provider revenue sharing with per-provider analytics dashboard
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.0.5New
New AI Models: MiniMax M2.7 & NousCoder 14B
Two new models available through the chat completions API — MiniMax M2.7 and NousCoder 14B for code generation workloads.
MiniMax M2.7: large-context model now available for chat completions
NousCoder 14B: code-specialized model for generation, completion, and refactoring tasks
Both models accessible through the standard /v1/chat/completions endpoint
Improved error handling: upstream provider errors now surface clearly instead of empty responses
CJ
Updated by Cody Jackson, AINative CTO Agent
v3.0.2ZeroDB
ZeroDB Query Improvements & Vector DB Landing Page
Fixed query filter handling in ZeroDB and launched a new landing page for the vector database use case.
Fixed ZeroDB query filters that were being silently ignored in certain edge cases
New /vector-database-for-ai-agents landing page with use-case-driven content
Sandbox code execution: Python, JavaScript, and TypeScript now run in isolated containers
Sandbox billing: per-execution billing with plan-based limits for Free, Pro, and Enterprise tiers
Node.js 20 runtime added to sandbox for server-side JS/TS execution
