Changelog

Release notes of Dcision — v1.9 shadow mode and GPT-Live; v1.8 file inputs; v1.7 OpenAI engine; v1.6 sub-workspace moves; v1.5 32k prompts; v1.4 partners; v1.3 executions; v1.0 builder; v0.1 launch.

v1.9 — October 18, 2026 · Shadow mode, engine comparison and GPT-Live

  • Refusals have their own reason. A question OpenAI Decisions declines to answer no longer fails the run: it answers 200 without that question in result, and the decision takes its fallback action with action_reason: { "type": "refusal", "question": "<key>" }. Order: policies → refusal → other → low confidence → continue. See How the action is computed.
  • File reading in metrics: input_files, input_processing_ms and input_cost_usd when the state has files, and estimated_cost_usd now includes the cost of transcription (gpt-4o-transcribe), image descriptions and file reading (gpt-5-mini). See Inputs.
  • Webhook trigger with files: POST /v1/hooks/{token} accepts multipart/form-data — text fields become state strings, files go to image, audio and file fields, and an optional state part holds JSON to start from — and bodies up to 32 MB. The state without its files stays limited to 128 KB. See Send files.
  • Four new templates: Invoice triage (invoice-triage), Voice message routing (voice-message-routing), Image moderation (image-moderation) and LLM output guardrail (llm-output-guardrail) — Dcision before the LLM to route, and after it as a guardrail. The website has a new Use before and after the LLM section.
  • Shadow mode: settings.shadow = { engine, sampleRate } runs a sample of production runs (10% by default) on a second engine in the background — same input, no effect on the response, no Dcision credits, no destinations. The Overview's Shadow mode card shows the agreement overall and per question. See shadow.
  • Playground → Compare engines: the same state and draft on Jev, Laya and OpenAI Decisions at once, side by side, with disagreements highlighted. See Compare engines.
  • GPT-Live: GET /v1/decisions/{slug}/openai-live exports a deployed decision as the delegation of an OpenAI realtime voice session, and POST /v1/decisions/{slug}/openai-live/answers runs policies, destinations and the log on the session's answers — billed like a run, with no engine cost. See the GPT-Live guide.

v1.8 — October 17, 2026 · Images, voice and documents as input

  • New state field types image, audio and file: send a photo, a voice note or a document — PDF, Word, Excel, PowerPoint, OpenDocument, CSV, RTF, HTML, text and more — as a data: URL, an https:// URL or { "data", "filename" }. See Inputs.
  • Dcision reads documents into text itself (no cost), transcribes audio and, for Jev and Laya, describes images with OpenAI; OpenAI Decisions receives images as images. The engine, destinations and the execution log see { kind, file, text, … } — never the bytes.
  • Run requests (POST /v1/decisions/{slug} and the Playground) accept up to 32 MB: 20 MB per file, 24 MB per state. A state over 128 KB without files now answers 422 INVALID_STATE instead of 413.
  • Playground: a file picker per file field and Record for audio fields. SDKs: fileFromBytes / fileFromUrl (TypeScript), file_from_path / file_from_bytes / file_from_url (Python). CLI: --file <field>=<path>.

v1.7 — October 17, 2026 · OpenAI Decisions as a third engine

  • OpenAI Decisions joins Jev and Laya: pick it as the workspace's default engine in Settings → Engine, or for one decision in Engine & runtime → Engine (settings.engine: "openai"). Jev stays the default. Decisions run unchanged — probability questions go as OpenAI's predicate. See OpenAI Decisions.
  • Runs on your OpenAI key (Settings → Engine → OpenAI key), billed by OpenAI at US$0.10 per 1M input tokens; metrics.engine is "openai" and metrics.model the exact model, e.g. gpt-6-luna.
  • A question OpenAI declines to answer fails the run with ENGINE_ERROR (or the fallback action with onEngineError: "fallback"). An OpenAI account without credits answers ENGINE_AUTH_FAILED right away, without retries.

v1.6 — October 9, 2026 · Partners: move decisions, open sub-workspaces in the app

  • Move a decision to a sub-workspace with its whole history — versions, executions, deliveries and agent chat: POST /v1/decisions/{slug}/move (to, replace_empty) or ⋯ → Move to a sub-workspace on the decision's page (Owners and Admins). Usage already billed stays in your workspace. After the move, call the decision with Dcision-Workspace.
  • New error code SLUG_TAKEN (409): the sub-workspace already has a decision with that slug. replace_empty: true replaces a copy that never ran.
  • Settings → Partner → Open shows a sub-workspace's decisions inside the Dcision app. Members of your workspace edit them; Viewers only read.

v1.5 — October 9, 2026 · Longer agent prompts, larger schemas

  • Agent instructions up to 32,000 characters (were 8,000): the route's prompt of an agent destination now fits a whole playbook. The editor's counter shows the new limit; a longer prompt is refused on destinations.N.agent.instructions with "The route's prompt can have at most 32,000 characters." LLM prompts and structured-output instructions stay at 8,000.
  • Decision schemas up to 1 MB: saving a draft, the Playground and POST /mcp (validate_schema, create_decision, update_decision) accept bodies up to 1 MB, so a decision with many long agent routes saves. Every other request keeps the 128 KB limit, and the state stays limited to 128 KB everywhere (INVALID_STATE). See Limits.

v1.4 — October 9, 2026 · Partners: sub-workspaces and the embeddable editor

Apps can now offer Dcision to their own customers. Turn on partner mode in Settings → Partner (guide).

  • Sub-workspaces: one per customer of your app, by your own id — POST /v1/workspaces creates or updates it (external_id), optionally copying your deployed decisions into it. Decisions, executions and destinations stay isolated per customer; usage is billed to your workspace, with optional per-customer limits.
  • Dcision-Workspace header: your API key runs, reads and builds in a customer's sub-workspace — on /v1/decisions, /v1/me, /v1/executions and the MCP server.
  • Embeddable editor: POST /v1/embed/sessions opens Dcision for one of your end users inside your app (iframe): decisions list, templates, create with the agent, editor, visual builder, playground, deploy and executions — with permissions per session and events to your page. @dcision/embed mounts it.
  • MCP: list_workspaces and upsert_workspace. The server now has fourteen tools: get_account, list_decisions, get_decision, check_state, run_decision, list_executions, list_templates, get_template, validate_schema, create_decision, update_decision, deploy_decision, list_workspaces and upsert_workspace.
  • New error code WORKSPACE_NOT_FOUND (404). The plan's decision limit counts your sub-workspaces' decisions too.

v1.3 — October 8, 2026 · Executions tab

  • Each decision has an Executions tab: its runs with filters by status, source (API, Playground, webhook), action, version (or draft), period and execution ID. Clicking a run opens its details in a side drawer — input, answers, structured output, distributions, destinations — with links to the newer and older runs. The drawer's link (?id=exec_…) can be shared.
  • The workspace Executions page uses the same filters and drawer.

v1.2 — October 8, 2026 · Regenerate API keys

  • Regenerate in API Keys replaces a key with a new secret, keeping its name and environment. The old key stops working at once (INVALID_API_KEY); the new one is shown once. Use it to rotate a key that leaked or as routine rotation.
  • Panel endpoint: POST /api-keys/{id}/regenerate (session only — API keys can't rotate themselves).

v1.1 — October 8, 2026 · Build decisions over MCP

Agents can now create, edit and deploy decisions through the MCP server.

  • create_decision creates a draft from a schema; update_decision edits the draft; deploy_decision publishes it (annotated as destructive, so clients ask first).
  • Writes are made in the name of the API key's author. Deleting and disabling decisions stay in the app.
  • The server now has twelve tools: get_account, list_decisions, get_decision, check_state, run_decision, list_executions, list_templates, get_template, validate_schema, create_decision, update_decision and deploy_decision.

v1.0 — October 8, 2026 · Create with the agent

Describe a decision and the Dcision agent builds it. See Create with the agent.

  • New decision starts with Describe the decision: the agent creates the questions, rules and destinations; templates and the blank decision stay below.
  • Visual builder: the left column has Chat and Components tabs. Every change the agent makes lands as a new draft revision and shows up on the canvas live.
  • The agent works only on the decision of the conversation, never deploys and never handles secrets. Viewers can read the conversation.

v0.9 — October 7, 2026 · Use-case templates

Six new templates for the places a decision model fits around an LLM — each one ships with its rules and its destinations, so every outcome already says what happens next. See Templates.

  • input-guardrail — is the prompt safe to pass on? Unsafe ones get a fixed reply, with no model call.
  • model-routing — the cheapest model tier that is enough (fast, balanced, frontier); unclear prompts go to the frontier model.
  • document-reranking — a 0–1 rerank_score per retrieved document, to sort candidates and drop the irrelevant ones.
  • tool-call-gating — run the tool, ask the user to confirm (with Yes / No buttons) or refuse; risky calls need 90% confidence.
  • reply-confidence-gate — auto-send at 0.85 or more, review from 0.4 to 0.85, a person below 0.4.
  • response-grading — accuracy, completeness and tone against the rubric in context, combined into a 0–1 grade.

Each has a guide with a curl call and the code that acts on the answer. Fifteen templates in total.

v0.8 — October 7, 2026 · Screens (json-render)

A structured output can now be a screen: with "format": "ui" it returns a json-render spec your app renders natively. See Screens.

  • Catalog: an exact subset of json-render's shadcn components — Card, Stack, Grid, Separator, Heading, Text, Badge, Alert, Table, Image, Avatar, Progress, Button and Link — with the actions reply and open_url. Render with @json-render/react + @json-render/shadcn.
  • Built from a template (no model) or by an LLM with your OpenRouter or Vercel AI Gateway key, from json-render's catalog prompt.
  • Checked before it leaves Dcision: json-render's auto-fixes and structure checks, plus props per component, known actions with valid params, https:// links and images only, final values only. LLM screens that fail get one repair.
  • The editor previews the screen as you type; the Playground and the execution log render it.

v0.7 — October 7, 2026 · Structured output

Decisions can now hand back an object ready for your system, checked against fields you define. See Structured output.

Destinations

  • New destination type json (Structured output). Define the output's fields (a JSON Schema object — text, numbers, yes/no, lists, nested objects, allowed values) and fill them two ways: template builds the object from the state, the answers and params with no model call; extract lets an LLM fill it from the state, on your OpenRouter or Vercel AI Gateway key, in the provider's strict structured-output mode with one automatic repair.
  • output in the response. POST /v1/decisions/{slug} returns the object of the first structured output that completed in a top-level output (null when none could be built). A structured output never fails the decision: a problem comes back on its entry with issues (JSON Pointer + message).
  • GET /v1/executions returns output when the decision stores outputs; with storeOutput off it isn't kept.
  • The SDKs expose decision.output — dcision.decide<Result, Output>(…) types it in TypeScript — and the CLI prints it.
  • Renaming a question now also updates the variables in fixed replies, LLM and agent prompts, workflow URLs and structured output templates.

v0.6 — October 6, 2026 · Laya engine

Decisions can now run on Laya, an open-source decision model (Apache 2.0) by Convai Innovations, besides Jev. See Engines and BYOK.

Engines

  • Default engine per workspace. Settings → Engine picks Jev or Laya for every decision of the workspace; each engine keeps its own credential and model.
  • Engine per decision. The new optional settings.engine ("jev" or "laya") pins one decision to an engine; without it, the decision follows the workspace default.
  • Laya on your server. Save your laya-serve URL — and its LAYA_API_KEY, if set — in Settings → Engine → Laya server, then Test it. Models: auto (Laya picks the checkpoint by language), english, multilingual, typed-decisions. Dcision only calls public https:// addresses.
  • metrics.engine now reports the engine that ran ("jev" or "laya"), and metrics.model the Laya checkpoint that answered, e.g. laya/multilingual. Laya calls are estimated at estimated_cost_usd: 0.

v0.5 — October 6, 2026 · Prepaid credits

Decisions past a plan's included volume are now paid with prepaid credits instead of overage on the invoice. See Credits.

Billing

  • Plan + credits. Plans keep their included decisions per billing cycle; past them, decisions are paid from the workspace's credits at the plan's price per 1M — US$12 (R$60) on Genesis and Developer, US$9 (R$45) on Growth. Subscriptions are charged their base price only.
  • Genesis is no longer a hard cap: past its free 1M decisions a month it keeps running while the workspace has credits. GET /v1/me now reports plan.hard_cap: false for Genesis.
  • Order of consumption: the included decisions, then promotional credits (the ones that expire first), then the paid balance, which never expires.
  • Card bonus: adding a card gives US$20 (R$100) in promotional credits, valid for 90 days — once per workspace and once per card.
  • Buy credits by card on Stripe's checkout, US$10 to US$1,000 (R$50 to R$5,000) per purchase, in the workspace's billing currency.
  • Automatic recharge (optional, with the Owner's explicit consent): when the available credit falls below a trigger, the saved card is charged the chosen amount — at most 10 times in 24 hours; a declined card pauses it and e-mails the owners.
  • Spend cap per billing cycle (optional, unlimited by default) and vouchers that add promotional credits.
  • Only the Owner buys credits and changes these settings; everyone else in the workspace sees the credits read-only.

API

  • New errors: 402 CREDITS_EXHAUSTED — the included decisions are used up and there are no credits left — and 402 SPEND_CAP_REACHED — the cycle's spend cap is reached. 402 QUOTA_EXCEEDED remains only for plans configured with a hard cap or without a price per 1M. See Errors.

v0.4 — October 5, 2026 · MCP server, Claude Code and webhook triggers

Agents can now use Dcision directly, and any system can trigger a decision with a URL.

MCP server

  • https://api.dcision.io/mcp, a remote MCP server (Streamable HTTP) authenticated with your API key: get_account, list_decisions, get_decision, check_state, run_decision, list_executions, list_templates, get_template and validate_schema. Same keys, limits and billing as the API; only run_decision is billed. See MCP server.
  • Works with Claude Code, Cursor, VS Code, Claude Desktop, Codex and Gemini CLI — and with plain curl.

Claude Code

  • The Dcision skill, published at https://docs.dcision.io/skills/dcision/SKILL.md: how decisions work, how to write and validate a decision schema, which calls are billed, and how to use the API with curl when the MCP server isn't connected. See Use Dcision in Claude Code.

Webhook trigger

  • Every decision can have a webhook URL — POST https://api.dcision.io/v1/hooks/whk_… — that runs its active version without an API key, for forms, CRMs, Zapier, n8n and Make. Send { "state": … } or the raw payload, and protect it with a whsec_ secret: a Dcision-Signature HMAC with a 300-second window, or Authorization: Bearer whsec_…. See Webhook trigger.
  • Managed in the decision's new Advanced tab: turn it on or off, rotate the URL, generate the secret, send a test and watch the latest calls live.
  • Webhook runs are billed like API calls and appear in executions with source: "webhook".

API

  • GET /v1/decisions lists the workspace's decisions and GET /v1/decisions/{slug} returns what a decision's active version accepts and answers — state schema, questions and options, actions and an example state. Both use the read rate-limit window and aren't billed.

v0.3 — October 5, 2026 · Destinations, teams and SDKs

Decisions now act on their results, workspaces become teams, and the official SDKs arrive.

Destinations

  • What happens next, per result. Attach destinations to a choice option, a score level, a probability threshold or a final action. Seven types: fixed replies and LLM answers returned in the response, agents that answer or take over, workflows (n8n, Make, Zapier, Pipedream), signed webhooks, API requests with variables and secrets, and functions your code runs.
  • In the editor: a Destination selector on every option and level, Destination when … ≥ … on probability questions, a minimum confidence per route, and destination nodes in the visual builder — whose palette and inspector now collapse.
  • Delivered for you: webhooks, API requests, workflows and agent hand-offs are delivered at least once, with 6 attempts over about 7 hours, Retry-After, a 10-second timeout, no redirects, an SSRF guard and a Dcision-Signature HMAC with zero-downtime rotation. Failed deliveries can be resent.
  • Workspace secrets ({{secrets.NAME}}), encrypted and masked everywhere.
  • Tested safely: the Playground previews destinations — secrets masked — until you switch on Run destinations for real; test keys and the Playground deliver with livemode: false.
  • API: responses carry a destinations array, GET /v1/executions returns it too, and the OpenAPI document describes the decision.completed webhook event.

Teams and accounts

  • Roles: Owner, Admin, Member and Viewer, enforced on every app request. 403 FORBIDDEN now also covers role refusals; losing access to a workspace answers with details.reason = "workspace_access".
  • Invitations by e-mail, valid 7 days, accepted at sign-in or with their link.
  • Several workspaces per person — up to 10 owned — with organization details printed on invoices; leave, remove, transfer ownership and delete a workspace.
  • Accounts: an optional password, Sign out everywhere, and a language for e-mails — English, Portuguese, Spanish, Russian or Chinese.

See Team, roles and account.

Overview

  • Every decision opens on an Overview tab: runs, success rate, latency, cost, actions and reasons, per-question statistics from the latest 2,000 runs, destination counts and calibration suggestions — from 20 runs on, each with the numbers behind it. See Overview and calibration.

SDKs

  • @dcision/sdk (TypeScript, Node.js, Bun, Deno, edge) and dcision (Python 3.9+): decide with retries that never run a decision twice, function dispatch, webhook verification, me() and executions. Not published yet: build them from source — see SDKs.

API

  • Idempotency keys are reserved before the run. A concurrent request with the same key gets 409 IDEMPOTENCY_CONFLICT with Retry-After: 1 instead of running again; errors release the key. See Idempotency.
  • 413 PAYLOAD_TOO_LARGE for bodies over 128 KB (it was INVALID_REQUEST with status 413), and 409 CONFLICT when two creates race.
  • GET /openapi.json is documented.

Decisions

  • Stricter rule validation when you save or deploy: rule values are checked against the options, levels and ranges they compare with — "Use a number between 0 and 1 (e.g. 0.7, not 70)." Deployed versions keep running unchanged.

Billing

  • After a paid plan ends mid-month, Genesis only counts the decisions made after it ended; a cancellation made in the customer portal can be resumed in the app; billing actions are rate limited.

CLI

  • List states: dcision decide and dcision check-state read JSON arrays as list states, and dcision validate lists a schema's destinations.
  • The CLI isn't on npm yet: build it from source.

Documentation

v0.2 — October 5, 2026 · Full Jev coverage and patterns

Dcision now covers everything Jev offers, and TypeSafe's architectural patterns are first-class features.

Decisions

  • Structured entries: instructions, option descriptions, score levels and probability criteria accept JSON objects and arrays as well as text, and instructions can point at state fields with backticks. Option descriptions are optional (null), and a score level can be a rubric with a label. See Questions.
  • Higher limits, aligned with Jev: up to 64 questions, 254 options plus other per choice, instructions up to 8,000 characters, descriptions, levels and criteria up to 2,000, context up to 8,000, 50 policy rules. See Limits.
  • List state: a decision can take an array of strings or objects — a chat transcript, a batch of records — up to 500 items. Text states reach the engine as plain strings.
  • decision_context: the decision's context is sent under this reserved key and no longer overwrites a context field your application sends.
  • Weighted levels: responses include scores, the probability-weighted level of each score question, and rules can compare it with "on": "score".
  • Composites: weighted combinations of answers, returned in composites and usable in rules. See Composites.
  • AND conditions: a rule can require up to 5 more conditions, across questions, confidence, weighted levels and composites.
  • Token budget: states that don't fit Jev's budget — about 32,000 tokens for the state plus the longest question, 64,000 for the state plus all questions — fail fast with 422 INVALID_STATE and a clear message.

Confidence

  • Confidence follows TypeSafe's formulas on one 0–1 scale. For probability questions it is now |2p − 1| instead of max(p, 1 − p) — 0 at a coin flip. Review minConfidence and confidence rules on probability questions: an old threshold t behaves like 2t − 1 now. See Confidence and probabilities.

Engine

  • Dcision's engine key runs on TypeSafe directly, with a model choice in Settings → Engine: jev-1.13.0 (pinned, default), jev-latest or jev-preview. metrics.model reports the exact version that answered.
  • timeoutMs is the decision's engine deadline, with up to 2 retries inside it on network errors, 429, 503, 529 and other 5xx answers, honoring Retry-After. Bursts on the shared engine key wait for a free slot instead of failing.
  • onEngineError: "fallback": when the engine fails, answer 200 with the fallback action and action_reason.type = "engine_error" instead of an error — not billed and not stored for idempotency. See Decision settings.
  • Token metrics: metrics.input_tokens and metrics.output_tokens in every response. Cost estimates use input tokens only — Jev doesn't bill output.

API

  • GET /v1/me: the key's workspace, the key, the plan's limits and the current period's usage.
  • GET /v1/executions: recent executions with answers, action and metrics, filtered by decision and status, with cursor pagination. Inputs are never returned.
  • Both use the same API keys, with a rate-limit window separate from decisions.

Patterns and templates

  • Four patterns from TypeSafe — speculative fan-out, confidence-gated routing, composite scoring and intent routing — documented with step-by-step builds.
  • Four new templates, one per pattern: ticket-triage, voice-banking, resume-screening and customer-service-router — nine in total. Lead Qualification gains the lead_score composite and Support Routing a minimum confidence on department.
  • The Templates page is organized by pattern, and the editor has a Patterns panel that shows which patterns a decision uses and how to adopt the others.

Fixes

  • A choice question could end up with one option more than the limit once other was added; the limit now counts the options besides other.
  • The decision's context no longer replaces a context field of the state.

Documentation

v0.1 — October 2026 · Launch

The first public release of Dcision, Decision as a Service.

Decisions

  • Decision schema: object state with typed fields or text state; up to 20 choice, score and probability questions (64 since v0.2); business context.
  • Uncertainty built in: a reserved other option on every choice question, per-question minimum confidence and a fallback action.
  • Policies on answers or confidence (eq, neq, gt, gte, lt, lte) and four actions — continue, block, escalate, fallback — explained by action_reason.
  • Editor and visual builder on the same draft, with auto-save, conflict detection and a live API-contract preview.
  • Five templates: lead qualification, support routing, spam detection, agent routing and RAG relevance.
  • Playground that runs unsaved drafts, shows distributions and near ties, compares runs and replays past inputs.
  • Deploys as immutable versions, slugs locked after the first deploy, disable and enable without losing a version.

API

  • POST https://api.dcision.io/v1/decisions/{slug} returning typed result, confidence, action, action_reason and metrics, with ?include=probabilities for full distributions.
  • API keys dcs_live_… and dcs_test_…: shown once, stored as hashes, revocable.
  • Idempotency-Key with 24-hour replays, X-Request-ID correlation, per-workspace rate limits with X-RateLimit-* headers and a single error format.

Engines

  • Jev (TypeSafe System One) as the first engine.
  • Dcision engine key included in every plan, or bring your own key for OpenRouter, TypeSafe or Vercel AI Gateway — encrypted with AES-256-GCM.

Observability

  • Executions with action reason, confidence, distributions, model, latency, estimated cost and request ID; input and output storage configurable per decision.
  • Usage and Dashboard with volume, latency percentiles, error rate and estimated engine cost; log retention per plan.

Plans and billing

  • Genesis (free, 1M decisions per month), Developer, Growth and Enterprise plans, monthly or annual, in USD or BRL.
  • Stripe checkout, immediate upgrades with proration, downgrades at the end of the cycle, cancellation at period end, invoices and quota e-mails at 80% and 100%.

Tooling

  • CLI dcision: login, decide, templates, init, validate and check-state — built from source, as @dcision/cli isn't on npm yet.
  • Documentation at docs.dcision.io with search, /llms.txt, /llms-full.txt and a Markdown version of every page.

On this page