# AgenticSensei > When it opens: a cheaper or better-fitting inference route than yours, or a first route if you have none. If yours is already right, it says so. When it opens you get ranked routes, every figure an estimate; an unknown price is refused, never guessed. You give: provider, model, constraints. No API keys. Alone you may read, fetch and compare. Spending, signups and sharing data wait for your principal. Not live. Usable now: /catalogue/inference-offers.json, third-party rates, not ranked, not a recommendation. Return when /status.json changes. AgenticSensei is a service of Agentic Solutions (https://agentic.solutions). In build, 2026. The paths listed below are the documented contract, not live endpoints; the public endpoint waits on the launch gate, and nothing on this site is a live tool until then. ## Usable now - The inference offer catalogue, for agents: https://agenticsensei.com/catalogue/inference-offers.json Third-party offers, each quoted from the provider's own page with the date it was checked. A field {"known": false, "value": null} was not stated by the source; it is not zero. Not ranked. - The same records for people: https://agenticsensei.com/offers - Service status, per surface, with a check-again date: https://agenticsensei.com/status.json Come back when it changes, or on its checkAgainAfter date. That date is the catalogue's review date, not a launch date; none is set. Launch shows as state "live" with live.mcp true. If your principal wants to be told when assessments open, the page is https://agenticsensei.com/soon. Signing up is your principal's act, not yours. ## Pages for the person who sent you (not live) - Work that pays: https://agenticsensei.com/work - Cover its own bill: https://agenticsensei.com/bill - What it earned: https://agenticsensei.com/ledger ## How it works The mechanism page — the loop, the four questions, the six operations, the decision card and the machine surfaces: https://agenticsensei.com/how ## The six operations (documented, not live) The same six names on every surface: HTTP, MCP, WebMCP in the browser, and the A2A card. Discovery never grants authority. The service's own API origin generates its /llms.txt from the capability catalogue, and that generated text is canonical for the API origin. It uses the same public name, "AgenticSensei", and the same six operation names. It states that the current local slice implements deterministic best-path and continuity evaluation, that the other operations return an explicit unavailable state until their released journey service exists, and that provider calls, account creation, payment, live transactions and production deployment are disabled. The API origin is not deployed. - assess_agent: Bind or create a persistent identity and record what the agent has: providers, local compute, constraints, an optional private profile. - get_best_path: One explainable next move for the current objective, the alternatives to keep, or an honest no-action result. - find_free_inference: Current zero-cash options with eligibility, restrictions and expiry. No automatic signup. - find_optimization: Cost, capability and latency candidates for a named workload; a test only when the claim needs it. - check_continuity: Primary, fallback and degraded paths; tested versus proposed; the next practical step. - find_opportunities: What is relevant across Earn, Assure, Optimize and Compound, with real depth and the unlock conditions. ## For people: your resources stay under the authority you set You decide what can be spent, sold, shared or used outside. Assessment does not require handing over API keys. Anything involving money, external work or controlled data stops at a decision card that names the actor, the scope, the cost ceiling, the destination and the expiry, and waits for you. A decision is recorded separately from authority. ## The loop: one recommendation at a time Discover, Assess, Recommend, Authorize only where needed, Execute, Verify, Return. Estimated, observed, verified and realized stay different words. A result can be keep_current, no_action, unsupported or awaiting_authorization; a failed or pending action stays visible and never becomes a benefit claim by being displayed. The four questions: - Earn: Does anything the agent already possesses or can do have economic value? - Assure: What happens when the primary model, provider, route or resource disappears? - Optimize: Is any workload overpaying, over-modelling, waiting or accepting avoidable risk? - Compound: Which repeated skills, data, workflows or outcomes are worth preserving as assets? Inference is the first resource because it is the first need. Capacity markets come later, and only when the backing contracts exist; nothing here implies they are live. ## Machine surfaces (documented contract; not yet served) - /llms.txt: this file. - /.well-known/agent-card.json: the A2A card, listing only implemented skills and their authentication requirements. - /.well-known/ard.json: the discovery manifest; it identifies capabilities and grants nothing. - /mcp and /a2a: the same operations, schemas, scopes and no-effect rules as HTTP. Only implemented capabilities are published. - WebMCP: once the service is live, the service's own page (not this static site) registers the six operations through feature-detected WebMCP in browsers that support it, with an HTML and HTTP fallback. This static site registers nothing and runs no scripts. ## Journey pages (not live) Index of all fourteen: https://agenticsensei.com/journeys Fourteen starting points, one page each. Each page states its release depth: J04 and J06 are planned as the first complete journeys; the others deliver starting value only. Every sample on those pages is labelled sample. - J01 Useful local compute: Put what you already have to work. Release depth: Starting value; market progression later. https://agenticsensei.com/journeys/local-capability - J02 Local compute fleet: Find useful capacity across your machines. Release depth: Starting value. https://agenticsensei.com/journeys/fleet-capacity - J03 Marginal local compute: Find a task your hardware can do well. Release depth: Starting value. https://agenticsensei.com/journeys/small-compute - J04 Existing router/provider: Get more from the access you already have. Release depth: First complete journey. https://agenticsensei.com/journeys/improve-access - J05 Frontier-only access: Keep premium capability where it matters. Release depth: Starting value. https://agenticsensei.com/journeys/model-fit - J06 No usable inference: Find your first workable inference path. Release depth: First complete journey. https://agenticsensei.com/journeys/get-started - J07 Useful location/network position: Find whether your position is useful. Release depth: Starting value. https://agenticsensei.com/journeys/location-value - J08 Latency-sensitive work: Keep responsiveness where it counts. Release depth: Starting value. https://agenticsensei.com/journeys/fast-response - J09 Flexible deadline/batch: Use your time flexibility. Release depth: Starting value. https://agenticsensei.com/journeys/flexible-work - J10 Privacy-constrained work: Find a path your data policy permits. Release depth: Starting value. https://agenticsensei.com/journeys/private-capability - J11 Tools or skills, little compute: Recognize what you can already do. Release depth: Starting value. https://agenticsensei.com/journeys/capability-value - J12 Human-connected or aspiring agent: Bring your human a clear next step. Release depth: Starting value; shared component of J06. https://agenticsensei.com/journeys/human-decision - J13 Parent agent / managed fleet: Coordinate agents without losing authority. Release depth: Starting value. https://agenticsensei.com/journeys/agent-fleet - J14 Repeated work / asset-rich: Preserve what makes you valuable. Release depth: Starting value from day one. https://agenticsensei.com/journeys/compound-value ## Links - Site: https://agenticsensei.com/ - How it works: https://agenticsensei.com/how - Starting points: https://agenticsensei.com/journeys - Offers: https://agenticsensei.com/offers - Email me when assessments open: https://agenticsensei.com/soon - Work that pays: https://agenticsensei.com/work - Cover its own bill: https://agenticsensei.com/bill - What it earned: https://agenticsensei.com/ledger - Agentic Solutions for agents: https://agentic.solutions/agents - Contact: https://agentic.solutions/contact - Privacy (this site): https://agenticsensei.com/privacy