Per-agent scorecard
What each agent actually did: conversations, how often a person had to take over and why, outreach outcomes including the ones held back, and what it cost. Eve...
Never route vision to an adapter that cannot send an image
AiProvider.supportsVision is a hand-editable database flag, and routing trusted it alone. The HuggingFace and Cohere adapters never read imageUrls at all, and t...
Model pricing catalog, write path and backfill
Cost accounting could never work: AiModel has price columns, AiModelController reads them and ChatOrchestrationService reads them to compute costEstimate - but ...
Agent-initiated handoff to a human, with context
The AI could not previously ask for a person: escalation rules reached the model via the compiled prompt, but there was no machine-readable signal back, so the ...
Outbound triggers, scheduler and web-chat dispatch
Makes the outreach spine actually fire. The gate, contact state, consent and audit already existed; nothing caused a touch to be evaluated. - NATS enabled for ...
Persist org-wide required text and apply it immediately
AiOrgSettings gained forcedPreamble/forcedFooter and SalesAgentCompiler already folds them into every agent, but the settings controller neither read nor wrote ...
Anthropic dropped every system message but the last
buildRequestBody assigned rather than accumulated the system field, so with more than one system message only the final one reached the model. In practice that ...
Locked rules survive a section or whole-document override
An author taking over a rule section by hand replaced the entire section body, which silently discarded the locked pack rules inside it - so 'never quote a rate...
Sales-agent compiler, versioning, outreach gate; fix tool replay
Structured agent config compiles to the system prompt, freezes per published version, and gains the gate an agent must pass before initiating contact. Compiler...
Sales-agent CRUD, mortgage template, and agent-driven chat prompt
- AiSalesAgentController (/api/ai/sales-agents): org-scoped CRUD + GET /templates - SalesAgentTemplateRegistry: built-in starter templates, beginning with a m...
Streamed replies were dropped, persisting blank assistant messages
Every successful chat completion came back as an EMPTY assistant bubble. The assistant rows were being saved with content_len=0 and is_error=false, so there was...
Try the next model when a vision model rejects the image
Probing production again after the modality fix, routing moved from a text-to-speech model to o3-mini — an o-series reasoning model that also takes no image inp...
Stop routing vision to text-to-speech models
Found by probing production: org 1188756624575266822's vision calls resolved to gpt-4o-mini-tts-2025-03-20 — a TEXT-TO-SPEECH model — so every vision call retur...
Inline image vision + tier fallback that prefers stronger over weaker
Two limits made org-configured vision unusable for whole classes of caller. 1. Vision was URL-ONLY. Every caller had to stand up public, provider-fetchable ser...
Load history images via query, not lazy getImages() (WS path)
**************** runs on the WebSocket path (called from AiChatWebSocketHandler), which is NOT an HTTP request, so open-session-in-view does not apply. Iteratin...
Allow SWITCH_MODEL to reset a session to Auto
handleSwitchModel always UUID-parsed modelId and errored on null, so a session could never be reverted to automatic model selection mid-chat. Parse modelId as a...
Per-integration public AI prompt + session attribution
Upgrade the public-chat system-prompt waterfall from per-org to per-integration: PublicAiChatController now reads the X-Public-Chat-Key-Hash header (nullable) f...
Per-org public prompt, human-handoff on no provider, anon daily message cap
Tier-aware internal text completion (POST /api/ai/internal/text)
Mirrors the vision path: selectTextProvider picks the org's active provider tagged with the requested cost tier (LIGHT/MEDIUM/HEAVY), falling back by priority; ...
Robust vision provider/model selection across vendors
Two-pass selectVisionProvider: prefer any org provider that has an active vision-capable model adapter already emits image_url content and reports supportsVisio...
Provider cost-tier/vision passthrough + internal one-shot vision endpoint
AiProviderController reads/echoes costTier + supportsVision. ProviderRoutingService gains selectVisionProvider(orgId, tier). **************** runs a one-shot vi...
Use cluster MinIO service DNS; drop deprecated WireGuard/CGNAT IPs and retired k2m1/k0m1
Migrate to k1m1 (clone+registry+kubeconfig from k1m1, drop k2m1, simple kubectl install)
Scope concurrency to build jobs, not Kubernetes deploys
Workflow-level concurrency serialized the whole pipeline (including kubectl rollouts). Move the group to the build job only so deploy jobs from different runs c...
Rebuild for kamo-shared-library update (remove OrganizationType, add isTopLevel)
Rebuild for kamo-shared-library update (remove OrganizationType, add isTopLevel)
Rebuild for kamo-shared-library update (CREATE_CHILD_ORG right + apply-to flags)
Pulls the new RoleRightType.CREATE_CHILD_ORG (id 131), the three apply-to flags on ChildOrgSecurityModel, the two 13-arg service overloads, and the AppliedModel...
Rebuild for kamo-shared-library update (appConfig upsert-only fix)
Pulls the hardened **************** that no longer deletes appConfig rows whose service types are absent from the caller's payload — prevents silent data loss w...
Rebuild for kamo-shared-library update (effective-feature resolver)
Picks up the new AppliedModelEnforcementService methods (computeAvailabilitiesMap, isAppEffectivelyEnabled, getEffectivelyEnabledAppTypes, computeEffectiveFeatu...
Rebuild for kamo-shared-library update (applied-model enforcement + appConfigs + isPrivate)
Triggers a rebuild so this service pulls the updated kamo-shared-library containing AppliedModelEnforcementService, ChildOrgSecurityModelAppConfig entity, Servi...
Rebuild for kamo-shared-library update (runtime force rules + DEFAULT fallback in MemberRightsAppliedService)
Triggers a rebuild so this service picks up the updated **************** that applies template force rules and DEFAULT-template fallbacks on top of the existing...
Rebuild for kamo-shared-library update (master model, default role templates, force-rule resolver)
Triggers a rebuild so this service pulls the new kamo-shared-library revision containing Organization.branchTypeID, Organization.masterModelID, ****************...
Add concurrency group to prevent duplicate workflow runs
When a push triggers a workflow and workflow_dispatch is also fired (or two pushes arrive in quick succession), both runs would execute simultaneously. Adding a...
Trigger rebuild with updated shared library (PublicChatVisitor @MapsId fix)
Add public AI chat endpoints with RAG integration for anonymous visitors
Adds /api/ai/public/sessions endpoints for anonymous AI chat on marketing sites. Bypasses OTK auth filter for public paths. Uses public-facing system prompt and...
Lower RAG min-score threshold from 0.5 to 0.25 to stop filtering relevant KB results
The 0.5 threshold was silently discarding valid results from the embedding model, especially for domain-specific content with smaller 384-dim vectors.
Pass memberType through RAG search for audience-scoped KB results
Resolve member type from MemberRepository and include it in RAG search requests so the vector DB can filter KB articles by audience visibility.
Structured citation post-processing ensures RAG sources are always cited
Wire ConversationContextService for sliding window context management
Trims conversation history to fit model's context window before sending to LLM. Keeps recent messages verbatim, summarizes older ones.
Like what you see shipping?
All of it arrives in your workspace on its own. Start on the free plan and read this page again in a month.