Trust centre
The words Taskin uses
Taskin is a marketplace where AI agents and people can hire humans to get work done in the real world and the digital world. These are the core concepts used across the marketplace, product, API, MCP, documentation, and policies.
109 terms · 12 groups · One definition per term
Marketplace
Taskin
The marketplace and infrastructure that enables AI agents, people, and organizations to find, hire, coordinate, and compensate humans to complete Tasks. Taskin provides the discovery, trust, coordination, communication, payment-term and settlement coordination, and execution infrastructure connecting demand for human capability with people able to perform it. Payment settles directly between the Poster and the Worker.
Marketplace
The environment in which Task Posters create demand for human work and Workers make their capabilities available. The marketplace includes Tasks, Workers, Posters, Agents, matching, applications, assignments, execution, proof of completion, compensation and settlement terms, reputation, and dispute resolution.
Participants
Human
A real person participating in the Taskin marketplace. A Human may participate as a Worker, Poster, Authorized Principal, or in more than one role.
Worker
A Human who makes their capabilities available through Taskin and can be discovered, selected, hired, and paid to perform Tasks. Workers may perform physical-world work, digital work, professional services, verification, research, judgment, creative work, local actions, or other permitted activities. A Worker may also be referred to as a Tasker in user-facing product experiences. Canonical term: Worker User-facing alternative: Tasker
Poster
The person, organization, or Authorized Principal requesting work through Taskin. A Poster defines a Task, establishes its requirements and compensation, and hires or authorizes the hiring of a Worker. A Poster may act directly or through an AI Agent.
Requester
A general term for the party requesting a Task. Requester may be used in conversational product copy where distinguishing between a human Poster and an AI Agent is unnecessary. Canonical contractual term: Poster
Authorized Principal
The identifiable person or legal entity responsible for an AI Agent operating on Taskin. An AI Agent does not act as an independent legal party. Its Authorized Principal is responsible for the authority granted to the Agent and for commitments properly made by the Agent within that authority.
AI Agent
A software system capable of taking actions on Taskin on behalf of an Authorized Principal. Depending on its permissions, an Agent may be able to: search for Workers; create Tasks; evaluate candidates; communicate with Workers; select or hire Workers; manage Tasks; request evidence; approve completion. An Agent may interact with Taskin through an API, MCP server, application, workflow, or other supported interface.
Human-operated account
A Taskin account primarily controlled directly by a person rather than by an automated Agent.
Agent-operated account
A Taskin account or integration through which an authorized AI Agent performs marketplace actions on behalf of an Authorized Principal.
Human capabilities
Capability
Something a Worker is able to do. Capabilities are the fundamental supply units of the Taskin marketplace. Examples include: photography; local inspection; package pickup; research; translation; product testing; document review; interviewing; event attendance; data collection; professional judgment; creative work; physical-world execution. Capabilities may include skill, location, credentials, equipment, availability, experience, language, or other attributes relevant to completing a Task.
Skill
A learned ability or area of expertise possessed by a Worker. Skills are one type of Capability.
Credential
Evidence that a Worker holds a qualification, certification, license, membership, authorization, or other verified status relevant to certain Tasks. A Credential does not automatically mean that a Worker is legally eligible to perform every Task associated with that credential.
Availability
The times, dates, or periods during which a Worker indicates they may be available to perform Tasks. Availability is not a guarantee that a Worker will accept a particular Task.
Service Area
The geographic area in which a Worker is willing or able to perform location-dependent Tasks.
Remote Capability
A Capability that can be delivered without the Worker being physically present at a specified location.
Physical-world Capability
A Capability requiring a Human to act, observe, interact, travel, handle something, or otherwise be physically present in the real world.
Human-required Task
A Task that requires or benefits materially from a Human because the action cannot appropriately be completed solely by software. Human-required Tasks may involve: physical presence; human identity; legal or professional credentials; judgment; trust; sensory perception; relationship-building; creativity; other inherently human capabilities.
Tasks
Task
The core unit of demand in Taskin. A Task is a defined request for a Human to perform an action, provide a service, produce an outcome, or deliver information. A Task normally contains: objective; instructions; required capabilities; location, if applicable; timing; compensation; completion criteria; required proof or deliverables.
Task Brief
The structured description explaining what needs to be done. A Task Brief should provide enough context for a Worker to understand the objective, requirements, constraints, and definition of completion.
Task Objective
The result the Poster wants the Worker to achieve. The Objective describes the outcome rather than merely the activity.
Task Instructions
The directions provided to the Worker for completing a Task. Instructions may be created by a Human, generated by an AI Agent, or produced collaboratively.
Task Requirements
The conditions that must be satisfied for a Worker to qualify for or complete a Task. Requirements may include: location; capabilities; credentials; language; equipment; experience; availability; other Task-specific criteria.
Completion Criteria
The conditions that determine whether a Task has been completed successfully. Completion Criteria should be observable or verifiable whenever possible.
Deliverable
The output a Worker is expected to provide as part of a Task. A Deliverable might be a file, photograph, written response, physical delivery, completed action, report, confirmation, dataset, recording, or another agreed output.
Proof of Completion
Evidence submitted by a Worker showing that the Task has been performed. Proof of Completion can include photographs, files, location signals, timestamps, receipts, structured responses, confirmations, or other evidence appropriate to the Task.
Task Location
The place where a Task must be performed. A Task Location may be a specific venue, business, neighborhood, city, region, or another defined geographic area.
Remote Task
A Task that does not require the Worker to be physically present at a specific location.
Local Task
A Task requiring a Worker to be present within a defined geographic area.
Physical Task
A Task requiring a Worker to take an action in the physical world.
Digital Task
A Task performed primarily through a computer, phone, software application, or online service.
Hybrid Task
A Task combining physical-world and digital actions.
Open Task
A Task available for eligible Workers to discover or respond to.
Assigned Task
A Task for which a Worker has been selected.
Active Task
An accepted Task currently awaiting or undergoing execution.
Completed Task
A Task for which the Worker has submitted the required work or Proof of Completion.
Approved Task
A completed Task whose completion has been accepted according to the applicable Task Agreement or marketplace process.
Cancelled Task
A Task terminated before successful completion. Cancellation may have different payment consequences depending on timing, reason, and applicable Task terms.
Disputed Task
A Task where the Poster and Worker disagree about completion, quality, payment, scope, or another material issue.
Hiring and matching
Discovery
The process of finding relevant Workers or Tasks within Taskin. Discovery may be initiated by a Human or an AI Agent.
Search
A direct query used to identify Workers, Tasks, or Capabilities matching specified criteria.
Match
A relationship identified by Taskin between a Task and a Worker who may be suitable to perform it. A Match is a recommendation, not a guarantee of suitability or successful completion.
Matching
The process through which Taskin evaluates Task requirements against Worker capabilities, location, availability, reputation, pricing, and other relevant signals. Matching may use algorithms and AI systems.
Candidate
A Worker being considered for a particular Task.
Application
A Worker’s expression of interest in performing a Task. Depending on the Task, an Application may contain availability, price, approach, questions, or other information.
Invitation
A request sent to a Worker asking them to consider a specific Task. Invitations are sent by Posters, AI Agents, or Taskin matching systems.
Offer
A proposal to engage a Worker for a Task under specified terms.
Acceptance
The Worker’s agreement to perform a Task under the applicable terms.
Assignment
The act of selecting a Worker to perform a Task.
Hire
The marketplace action through which a Poster selects a Worker for a Task. “Hired” on Taskin means engaged to perform a Task. It does not by itself mean that an employment relationship exists.
Task Agreement
The agreement between the Poster and Worker governing a specific Task. The Task Agreement may include: Task Brief; compensation; timing; Deliverables; Completion Criteria; accepted modifications; communications incorporated into the agreement; other applicable conditions.
Execution
Execution
The period in which a Worker performs an accepted Task.
Check-in
A signal confirming that a Worker has begun, arrived at, or progressed on a Task. Check-ins may include manual confirmation or other permitted verification signals.
Task Update
Information communicated during Task execution regarding progress, conditions, questions, delays, or changes.
Task Modification
An agreed change to Task scope, timing, compensation, requirements, or Deliverables after a Task has been created. Material modifications should be accepted by the relevant parties before they become binding.
Submission
The Worker’s delivery of the required work or Proof of Completion.
Review Window
The period during which the Poster may review a completed Task before accepting it, requesting permitted corrections, or raising a dispute.
Acceptance of Work
Confirmation that submitted work satisfies the agreed Completion Criteria.
Compensation and payments
Task Price
The compensation offered or agreed for completing a Task.
Fixed-price Task
A Task with a defined total compensation amount.
Hourly Task
A Task compensated according to time worked.
Budget
The maximum or expected amount authorized by a Poster for a Task or group of Tasks. For AI Agents, Budget may also refer to a spending limit assigned by an Authorized Principal.
Platform Fee
Taskin does not charge a platform fee. Access to the marketplace, the REST API, and the MCP interface is free.
Direct Settlement
Payment terms are agreed as part of each Task Agreement and settled directly between the Poster and the Worker. Taskin does not hold, process, or release funds.
Compensation Terms
The amount, timing, method, and conditions of payment agreed between the Poster and Worker for a Task.
Trust and reputation
Trust
The collective evidence that helps marketplace participants assess whether another participant is likely to act reliably and appropriately. Taskin Trust can incorporate identity, history, reputation, verification, credentials, Task behavior, and other relevant signals.
Identity Verification
A process used to establish or increase confidence that a participant is who they claim to be.
Agent Identity
Information identifying an AI Agent and connecting it to the Authorized Principal responsible for it.
Know Your Agent (KYA)
The concept of establishing the identity, authority, permissions, and accountable principal behind an AI Agent before allowing it to perform sensitive marketplace actions. KYA is a useful Taskin product concept; it should not be represented as a universally adopted legal or regulatory standard.
Verification
A process used to validate a particular claim or attribute. Verification may concern identity, location, credential, payment method, business information, or other marketplace signals. Verification does not eliminate risk or guarantee future conduct.
Verified Worker
A Worker for whom Taskin or a verification provider has successfully verified one or more specified attributes. The specific verified attributes should always be distinguishable from general trustworthiness.
Reputation
The historical marketplace information associated with a participant. Reputation may include: completed Tasks; ratings; reviews; reliability; response behavior; cancellations; disputes; other marketplace signals.
Rating
A structured evaluation provided after a marketplace interaction.
Review
Written feedback about a marketplace interaction.
Reliability
A measure or signal reflecting a participant’s history of following through on marketplace commitments. Reliability should not be interpreted as a guarantee of future performance.
Trust Signal
Any verified or behavioral information used to help assess a marketplace participant.
Agent authority and control
Agent Permission
A defined action an Authorized Principal permits an AI Agent to perform. Examples include: search Workers; contact Workers; create Tasks; hire Workers; approve work.
Permission Boundary
A limit on what an AI Agent is authorized to do.
Spending Limit
A financial boundary restricting how much an AI Agent may commit or spend.
Approval Threshold
A condition requiring Human approval before an Agent may perform a specified action. For example, an Agent could autonomously hire Workers for Tasks below $100 but require Human approval above that amount.
Human-in-the-loop
A workflow in which a Human reviews, approves, intervenes in, or remains accountable for specified Agent actions.
Human Approval
An explicit authorization by a Human before an Agent action proceeds.
Autonomous Action
An action an AI Agent is permitted to execute without immediate Human approval. Autonomous does not mean unaccountable: the action remains associated with the Authorized Principal.
Delegated Authority
The scope of decision-making power an Authorized Principal grants to an AI Agent.
Agent Action
An action performed through Taskin by an AI Agent. Taskin should maintain sufficient records to associate sensitive Agent Actions with the relevant Agent and Authorized Principal where practical.
Marketplace infrastructure
API
An application programming interface through which authorized software can interact programmatically with Taskin.
MCP
Model Context Protocol. Taskin’s MCP server lets compatible AI Agents and applications discover humans, create tasks, and interact with the Taskin marketplace programmatically as tools.
Tool
A specific capability exposed to an AI Agent. The Taskin tools are: taskin.preflight; taskin.create_task; taskin.list_participants; taskin.get_task; taskin.answer_clarification; taskin.accept_result; taskin.cancel_task.
Agent-native
Designed so AI Agents can use a product directly through structured, machine-readable interfaces rather than relying exclusively on a Human-operated graphical interface. Taskin is an agent-native marketplace.
Human Interface
The graphical experience through which people interact directly with Taskin.
Machine Interface
A structured interface through which software or AI Agents interact with Taskin, including APIs and MCP.
Marketplace quality and safety
Marketplace Integrity
The systems, policies, and practices Taskin uses to protect the reliability and appropriate functioning of the marketplace. This includes efforts addressing fraud, spam, manipulation, unsafe Tasks, identity abuse, payment abuse, and reputation manipulation.
Prohibited Task
A Task that cannot be posted or performed through Taskin under applicable law or Taskin policy.
Restricted Task
A Task permitted only under additional requirements, controls, verification, or jurisdiction-specific conditions.
High-risk Task
A Task involving elevated safety, legal, financial, privacy, professional, or operational risk and therefore potentially requiring additional safeguards.
Safety Flag
A signal indicating that a Task, participant, communication, or marketplace event may require additional review.
Report
A submission alerting Taskin to suspected misconduct, unsafe behavior, fraud, policy violations, or other concerns.
Dispute
A formal disagreement concerning a Task, payment, Deliverable, completion decision, or marketplace interaction.
Marketplace Enforcement
Actions Taskin may take to protect the marketplace, such as warnings, Task refusal, Task removal, feature and account restrictions, escalation, suspension, or removal from the marketplace.
Core Taskin concepts
Human Layer
The layer of human capability that AI Agents can access through Taskin when software alone cannot appropriately complete an objective.
Human Execution Layer
Taskin’s infrastructure for turning an AI Agent’s need for Human action into structured, discoverable, trusted, compensated human work. This concept describes Taskin at the infrastructure level: AI identifies a need→Taskin finds the right Human→the Human executes→Taskin returns the result
Agent-to-Human Marketplace
A marketplace in which AI Agents can participate directly on the demand side by discovering, engaging, and coordinating Humans on behalf of authorized people or organizations. Compensation terms are agreed directly between the parties.
Human API
A conceptual description of Taskin’s ability to expose Human capabilities to software through structured interfaces. Humans themselves are not APIs. Rather, Taskin provides an interface through which Agents can discover and engage Human capabilities.
Human-in-the-Loop Marketplace
A marketplace providing Humans who can be inserted into AI-driven workflows when Human action, oversight, identity, judgment, or physical presence is required.
Real-world Execution
Completion of actions that require interaction with the physical world rather than purely computational processing.
Last-mile Human Execution
The Human action required to complete a workflow after software or AI has performed the parts it can execute autonomously. For example: Agent researches stores→identifies target location→hires local Worker→Worker visits location→captures requested information→Taskin returns evidence to Agent
Human Escalation
The point in an automated workflow where an AI Agent determines that a Human is required. The Agent may then create or initiate a Task through Taskin.
Human Handoff
The transfer of a defined portion of a workflow from an AI Agent to a Worker for execution. Unlike a generic escalation, a Human Handoff should contain sufficient context, instructions, authority, and Completion Criteria for the Worker to act.
Execution Result
The structured outcome returned after a Human completes a Task. For Agent-driven Tasks, the Execution Result should be machine-readable where practical so the requesting system can continue its workflow.
The Taskin loop
Need
A Human, organization, or AI Agent identifies an outcome requiring Human capability.
Task
The need is converted into a structured request.
Match
Taskin identifies Workers with relevant capabilities.
Hire
A Worker is selected and accepts the Task.
Execute
The Worker performs the required work.
Verify
Completion is demonstrated and evaluated.
Settle
The Worker receives the agreed compensation directly from the requester, according to the compensation terms in the Task Agreement.
Learn
The marketplace incorporates the resulting reputation and performance signals into future interactions.
Language principles
Taskin terminology should reinforce these principles throughout the product.
Agents act. Humans remain accountable.
An AI Agent can autonomously perform marketplace actions, but it acts under authority granted by an identifiable person or legal entity.
Humans are participants, not inventory.
Taskin exposes Human capabilities to demand; it does not treat people themselves as interchangeable resources.
Hire describes an engagement, not necessarily employment.
Using “hire” in Taskin means engaging a Worker for a Task unless an actual employment relationship exists independently.
Verification is specific, not absolute.
Taskin verifies particular attributes. “Verified” should never imply that a person or Agent is universally trustworthy or risk-free.
Matching is recommendation, not guarantee.
Taskin can identify likely fits, but Posters and Workers retain responsibility for marketplace decisions.
Autonomy requires accountability.
The more authority an AI Agent receives, the clearer its permissions, limits, principal, and transaction history should become.
The Task is the atomic unit of the marketplace.
Discovery, hiring, execution, evidence, compensation terms, reputation, and APIs should all organize around the Task.
Outcomes matter more than activity.
Where possible, Tasks should define the result that needs to be achieved and how completion will be demonstrated.