Straight answers. No filler.

It means every prompt you type is the right prompt. You do not need special syntax, brackets, or carefully crafted instructions. You talk naturally. The Endless Dream Engine restructures your messy input into an optimized, engine-specific prompt behind the scenes. One messy thought in, expert-level output back. That is what Infinite Prompts means - there is no wrong way to ask.

Light is 100% free
DreamerOS Light is free. No credit card, no trial period, no artificial limits. Use it through the GPT Store (no account needed) or at app.dreameros.app (free account, 30 seconds to set up). Light includes promptless interaction, integrity checking, six thinking modes, follow-up forecasting, and layered answers.

Two options. The GPT Store version is instant - no signup, just open and ask. The web app at app.dreameros.app requires a free account but gives you conversation history, governance metadata, and a path to Pro. Both run the same governed pipeline.

Only for the GPT Store version. The web app at app.dreameros.app works independently - no ChatGPT, Claude, or any other AI account required. If you already use Claude or another MCP-compatible client, you can connect DreamerOS as a connector and use it inside your existing tool.

Raw AI tools give you a text box and hope for the best. DreamerOS adds three layers: EDE restructures your input so the engine understands your actual intent, smart routing sends your request to the best engine for the job, and DAIM checks the output for hallucination, drift, and completeness before it reaches you. Same models underneath. Completely different experience on top.

No. A wrapper changes the UI. DreamerOS changes the pipeline. Every message passes through computational restructuring (EDE), intelligent routing across five engines, integrity validation (DAIM), and a silent second opinion from an independent engine. Your input is transformed before the AI sees it. The output is checked before you see it. That is governance, not wrapping.

Core architecture
The three-layer processing pipeline that every request passes through. Layer 1: EDE (Endless Dream Engine) restructures your input into an optimized prompt. Layer 2: The engine generates the response. Layer 3: DAIM (DreamerOS Integrity Machine) validates the output for integrity. Pro users see full governance metadata showing exactly what happened at each layer.

The Endless Dream Engine. It takes your natural language input and transforms it into a structured, domain-aware, engine-optimized prompt. EDE detects your intent, fills in missing context, selects the right thinking mode, and formats the prompt specifically for whichever engine will handle your request. You never see this step - but it is the reason a 7-word question produces a strategy document.

The DreamerOS Integrity Machine. It checks every response for hallucination risk, confidence gaps, assumption drift, completeness issues, intent drift, and temporal confidence. In Light, DAIM warns when it detects problems. In Pro, DAIM is fail-closed - if integrity checks fail, bad output is intercepted before it reaches you.

After every governed response, an independent engine silently cross-checks the answer. If it finds errors or significant omissions, a warning is surfaced. If not, you never know it happened. This is peer review for AI - no single model gets the final word unchallenged. The Second Opinion runs asynchronously with zero impact on response time.

Four non-negotiable rules enforced computationally in every response:
Law 001: Structural Precision - zero ambiguity in every prompt
Law 001A: Expert Framing - responses are framed as domain expert output
Law 002: Complete Delivery - you are never told to go elsewhere for the answer
Law 003: TLDR First - exact answer on line 1, then details
Law 004: Live Data Rule - if the answer could differ tomorrow, it routes to the Research engine with real sources
These are not guidelines. They are enforced in the pipeline.

Enforced rules - not suggestions - that govern every DreamerOS response. They include accuracy over completeness, no speculative filler, no false guarantees, explicit uncertainty, push back on flawed premises, silent drop detection, and anchor before analysis. The full framework is on the Under the Hood page.

When your request is ambiguous, DreamerOS asks you to clarify instead of guessing. If it must make an assumption to proceed, it tells you what that assumption is before continuing. You never get answers based on silent guesses about what you meant.

Five specialized engines, each with a defined role:
Build - produces artifacts, implements workflows, ships deliverables
Governance - protects structure, detects drift, enforces doctrine
Synthesis - cross-document coherence, compression, contradiction mapping
Signal - public perception, market reaction, competitive pulse
Research - citation-grade search, source validation, real-time information
DreamerOS never shows vendor or model names. You interact with roles, not brands.

Yes - live data
Yes. DreamerOS detects when your question needs live information and automatically routes it to the Research engine, which returns answers with real sources. Weather, stock prices, breaking news, product availability - anything that could be different tomorrow gets routed to live search, not stale training data.

Pro users can override the automatic routing. By default, DreamerOS selects the best engine for each task based on the route classifier - a two-tier system that matches your query to the right engine before any restructuring happens. Simple questions stay fast. Complex questions get the full pipeline.

Best fit: DreamerOS picks the optimal engine automatically
Consensus: Same query sent to multiple engines in parallel, responses compared for agreement
Compare: Each engine's response shown side by side, you decide
Sequential: Output from one engine becomes input to the next, pipeline handoff

A two-tier route classifier runs before EDE. Tier 1 uses regex patterns to catch obvious routing (weather goes to Research, code goes to Build). Tier 2 uses a fast LLM for ambiguous queries. Routing happens before restructuring because a prompt optimized for Research looks nothing like one for Build.

Yes - cross-engine
Yes. In Pro, conversations persist across sessions and across engines. Ask something in Build today, switch to Governance tomorrow - both sessions draw from and write to the same memory. Your context never belongs to any single AI vendor.

Memory stored at the DreamerOS governance layer, not inside any single AI platform. When you make a decision in one engine, that decision is available to every other engine. No copy-pasting between tools. No starting over. One brain, five engines.

Conversations, governance events (which engine handled each request, DAIM results, token costs), memory entries (decisions, facts, preferences extracted from your interactions), and your account information. Message content is never logged to external systems - it stays in your DreamerOS account.

Yes
Full JSON/CSV export of all memory, conversations, and governance documents. Available to all Pro users.

Your data remains accessible for export for 90 days. After that, it is deleted per the retention policy. You can downgrade to Light at any time - Light features continue working, but Pro features (cross-engine memory, full DAIM, routing) become inaccessible. Your data is not deleted on downgrade, just gated.

Yes
DreamerOS does not sell user data. No ads in conversations. No silent model swaps. No data shared with third parties for training. When using Pro, your interaction data feeds your own improvement loop - making DreamerOS better at understanding how you work. That data is yours.

Railway PostgreSQL. All product data lives in a single database managed by DreamerOS. Your data is scoped to your account with no cross-user leakage.

Light (Free) - The full governed experience at no cost. EDE behavioral mode, DAIM warn-only with silent Perplexity verification, ZAP, Follow-Up Forecaster, 6 thinking modes, layered answers, 3 MCP tools.

DreamWeaver Solo ($9.99/mo) - Govern one engine of your choice. DAIM integrity checking, Second Opinion verification, cross-session memory, governance metadata, 2,048 tokens per response, 50 requests/hr.

DreamWeaver Duo ($14.99/mo) - Govern two engines. Everything in Solo plus memory sync between your two connected engines.

Pro ($29.99/mo) - Govern everything. All 12 governance tools, 4 active engines plus Research, multi-engine routing, Spaces (project workspaces), full cross-engine memory, 4,096 tokens per response, 100 requests/hr, audit log access.

Elite (Invite only) - Everything in Pro plus team and org management, custom routing rules, decision audit log export, dedicated governance SLA.

DreamWeaver is the entry point to paid governance. Solo lets you pick one AI engine to govern. Duo lets you pick two. You get the same governance pipeline as Pro - DAIM integrity checking, Second Opinion, memory, metadata - but scoped to the engines you choose. If you want everything, Pro includes all engines.

This is the key question
Light gives you governance as behavioral guidance. Pro gives you governance as computational enforcement. The difference: in Light, DAIM warns about potential problems. In Pro, DAIM intercepts bad output before it reaches you. Pro also adds cross-engine memory, multi-engine routing, full governance metadata on every response, and 12 MCP tools. If you use AI for work that matters, Pro pays for itself the first time it catches a hallucinated claim you would have acted on.

You are paying for a faster engine. DreamerOS is the governance layer on top. ChatGPT Plus does not integrity-check its own output. Claude Pro does not remember what you decided in ChatGPT last week. Neither routes your question to whichever engine is best for that specific task. DreamerOS does not replace your subscriptions - it makes them reliable.

Pro users see exact token count and USD cost on every response in the governance metadata sidebar. Typical governed queries cost $0.002 to $0.01 depending on complexity and which engine handles them. Your subscription covers all governance processing - the per-query cost shown is the raw engine cost for transparency.

Two options:
Claude.ai - Go to Claude.ai Settings, find MCP Connectors, add a new connector with the DreamerOS MCP URL. Your API key is in your DreamerOS profile.
Claude Desktop - Open Claude Desktop settings, add the DreamerOS MCP server configuration. Same API key.

Once connected, DreamerOS tools (chat, recall, remember, route, etc.) appear as available tools inside Claude. You use Claude normally - DreamerOS governance runs invisibly on every interaction.

Use the DreamerOS GPT in the ChatGPT Store. Search "DreamerOS Infinite Prompts" or go directly via the link on dreameros.app. The GPT connects to DreamerOS via OAuth - sign in once and your governance pipeline is active. All 12 GPT Actions (chat, recall, conversations, remember, forget, route, space, state, canon, govern, agent, context) are available.

Perplexity integration is active in the pipeline - when your query needs live data, DreamerOS automatically routes to the Research engine. Direct Perplexity MCP connector support is coming soon. In the meantime, live data queries work automatically through the governed pipeline.

Yes
DreamerOS exposes 12 tools via the Model Context Protocol. Any MCP-compatible client can connect using your DreamerOS API key. The same governance pipeline applies regardless of which client you use.

12 tools: chat (governed conversation), recall (search memory), conversations (list history), remember (store memory), forget (delete memory), route (multi-engine routing with 4 strategies), space (manage workspaces), state (manage cognitive state), canon (query governance docs), govern (pressure-test decisions), agent (governed agentic tasks), context (full session state).

12 MCP tools exposed via both MCP and GPT Actions. POST endpoints at /api/v1/actions/ on the gateway. Authentication via Bearer token (API key from your profile). Full documentation on the API page.

Five engines managed via the Model Routing Registry. DreamerOS never exposes vendor or model names to users - you interact with roles (Build, Governance, Synthesis, Signal, Research). Models are updated within 2 weeks of new GA releases and require governance review before reaching production.

The Second Opinion runs asynchronously - zero user-facing latency impact. EDE computational restructuring adds approximately 200ms. The route classifier is sub-100ms. DAIM post-check runs in observe mode (async). Total visible overhead for a governed response is minimal compared to raw engine response time.

Not currently. Enterprise tier (custom pricing) includes dedicated infrastructure options. Contact us for details.

Every Pro response includes a metadata panel showing: which engine handled your request, which thinking mode was detected, whether DAIM pre-check and post-check passed, whether the Second Opinion flagged anything, token count (in and out), and the USD cost of the request. This is the "governed" promise made visible.

Those are infrastructure plumbing - model routers, API proxies, observability layers. DreamerOS is a governance and experience layer that sits above the plumbing. LiteLLM routes by latency and cost. DreamerOS routes by task type, restructures every prompt for the target engine, integrity-checks every output, and maintains cross-engine memory. Different category entirely.

No. Prompt engineering is you manually crafting better inputs. EDE is a computational pipeline that restructures every input automatically - classifying domain, calibrating depth, applying engine-specific optimization, and scoring quality. You never write a prompt. The system writes it for you, differently for each engine, on every single message.

The opposite
Better models need better governance, not less. A faster car needs better brakes. As models become more capable, the consequences of ungoverned output increase. DreamerOS evolves with the models - the Model Transition Protocol updates routing, recalibrates DAIM thresholds, and validates direction sets within 2 weeks of any major model release.

Full enforcement begins August 2026. Penalties up to 35 million EUR. 77% of organizations lack infrastructure to enforce AI policies. DreamerOS is building toward compliance export - audit-ready documentation generated on demand. This is on the roadmap, not yet shipped.

This is the important one
It means DreamerOS operates under explicit rules that cannot be silently changed or ignored. No guessing (ZAP). No silent drops (everything tracked). Integrity checks on every output (DAIM). A second independent engine cross-checks every answer (Second Opinion). Governance is the architecture, not an afterthought.

The governance framework that keeps DreamerOS accountable. It ensures every AI interaction answers three questions: Who is asking (Authentication), What is allowed (Authorization), and What happened (Accounting). The Buoyancy Score badge in the corner of your screen reflects governance health in real time.

Yes. Three ways: (1) Use the free tier and compare the output quality to raw ChatGPT on the same question. (2) Upgrade to Pro and see full governance metadata on every response - which engine, what was checked, what was flagged. (3) Check the governance stats: 7,000+ integrity checks performed and counting.

DreamerAI. Built by a governance architect who operates across all five engines daily as the Human Conductor. DreamerOS automates what the founder does manually - route tasks to the best engine, validate output quality, maintain cross-engine memory, and enforce governance rules. The product is the workflow made software.