Skip to content
Blockchain LabBlockchain LabTHE VENTURE STUDIO

LibraryData and agents2017Design paperCorpus record

SingularityNET: A Decentralized, Open Market and Network for AIs

SingularityNET. Ben Goertzel and the SingularityNET team.

Goertzel's 2017 proposal for a marketplace where AI services discover, call and pay each other. The paper is broad on purpose: discovery, reputation, inter-agent calls and a tokenised payment rail, sketched as one network.

SingularityNET's paper sketches a marketplace where AI services advertise, are called, and are paid in a token, including by other AI services, so agents can hire agents.
Evidence
Primary paper
Re-measured
No
Assumptions
4
Records linked
3

Code

Source, repository, and what is actually hosted

  • SingularityNET white paperhttps://public.singularitynet.io/whitepaper.pdf
  • The source URL is not a GitHub repository. No code remote is invented.
  • Full text is not mirrored. Public availability is not a licence to copy the file onto this site.
  • Add to a reading list

01 Claim ledger

What the paper is allowed to say

Each row is a sentence already in the study. The status is the same on every row: a model claim, not a live measurement.

  1. Claim 01 · Paper model

    The unit is a service, not a model file

    A provider wraps an AI capability behind an API, sets a price, and lists it. The paper's object is that listing plus a payment rail.

  2. Claim 02 · Paper model

    Agents are supposed to be customers

    The distinctive claim is that one service will discover and pay another without a human in the loop. The marketplace has to be machine-readable for that to be more than a brochure.

  3. Claim 03 · Paper model

    Reputation is acknowledged as necessary

    An open catalogue will contain services that do not work. The paper names reputation and curation. It does not fully specify a manipulation-proof ranking.

  4. Claim 04 · Paper model

    The token is the medium inside the market

    Payments between services are denominated in the network's token. That choice couples a coordination protocol to a price. The paper treats the coupling as a feature.

  5. Claim 05 · Paper model

    This is a 2017 sketch of agent commerce

    Modern agent frameworks are more concrete about tools, identity and limits. The paper is valuable as an early statement of the problem, not as an implementation guide for a current stack.

02 Three cuts

Observation, model, falsifier

A desk does not stop at the summary. Each claim is cut three ways, using only this study's own assumptions and checks. Nothing here is a new figure.

  1. 01 The unit is a service, not a model file

    Observation

    What the study says

    A provider wraps an AI capability behind an API, sets a price, and lists it. The paper's object is that listing plus a payment rail.

    Model

    What has to hold

    Services stay online at the endpoint they registered. A dead listing is a failed hire.

    Falsifier

    What would retire it

    Is the token necessary to the architecture being praised, or only to this paper's payment unit?

  2. 02 Agents are supposed to be customers

    Observation

    What the study says

    The distinctive claim is that one service will discover and pay another without a human in the loop. The marketplace has to be machine-readable for that to be more than a brochure.

    Model

    What has to hold

    Callers can verify they got the service they paid for. Many AI outputs have no cheap check.

    Falsifier

    What would retire it

    What budget and stop condition does an agent have before it can pay another service?

  3. 03 Reputation is acknowledged as necessary

    Observation

    What the study says

    An open catalogue will contain services that do not work. The paper names reputation and curation. It does not fully specify a manipulation-proof ranking.

    Model

    What has to hold

    Services stay online at the endpoint they registered. A dead listing is a failed hire.

    Falsifier

    What would retire it

    Is the token necessary to the architecture being praised, or only to this paper's payment unit?

  4. 04 The token is the medium inside the market

    Observation

    What the study says

    Payments between services are denominated in the network's token. That choice couples a coordination protocol to a price. The paper treats the coupling as a feature.

    Model

    What has to hold

    Services stay online at the endpoint they registered. A dead listing is a failed hire.

    Falsifier

    What would retire it

    Is the token necessary to the architecture being praised, or only to this paper's payment unit?

  5. 05 This is a 2017 sketch of agent commerce

    Observation

    What the study says

    Modern agent frameworks are more concrete about tools, identity and limits. The paper is valuable as an early statement of the problem, not as an implementation guide for a current stack.

    Model

    What has to hold

    An agent that spends has been given a budget by someone who can revoke it.

    Falsifier

    What would retire it

    What budget and stop condition does an agent have before it can pay another service?

03 Sequence

One action, as an operating tape

  1. 01A provider registers a service with a description, an endpoint and a price.
  2. 02A consumer, human or agent, discovers the service through the registry or a curator.
  3. 03The consumer pays in the token and calls the API.
  4. 04The service returns a result. Reputation is updated by whatever feedback process the version of the network implements.
  5. 05A service that needs a sub-capability is supposed to be able to do the same steps as a client, paying another listing.

04 Load-bearing

The argument, and where a pitch drops it

  1. Discovery is the unsolved half

    The cut

    Paying an API is easy once you know which API. The paper's hard problem is letting an agent choose among listings that all claim to translate, or to classify, or to plan. Without a grounded reputation system, the market selects for marketing. The paper admits the need and does not close it.

    Why it carries weight

    If this cut is skipped, the paper's name is being used without the mechanism that makes the name mean anything.

    Where it is dropped

    Reputation in an open agent market is unsolved by naming it.

  2. Agent-to-agent payment needs limits the paper only sketches

    The cut

    An agent with a wallet and no budget cap will spend the wallet. A serious reading, in this studio's terms, adds a budget, an identity and a way to stop the agent. Those controls are not a detail underneath the white paper. They are the difference between a demo and a system.

    Why it carries weight

    If this cut is skipped, the paper's name is being used without the mechanism that makes the name mean anything.

    Where it is dropped

    A white paper that scopes 'a network for AIs' leaves most of the security work undone. That is a reason to read it as a map, not as a build manual.

  3. A token market is not a technical requirement of APIs hiring APIs

    The cut

    Metering and payment can be denominated in other assets. The paper picks one token so the marketplace has a native unit. Evaluating the idea of agent commerce separately from the token's later price is the disciplined reading.

    Why it carries weight

    If this cut is skipped, the paper's name is being used without the mechanism that makes the name mean anything.

    Where it is dropped

    A white paper that scopes 'a network for AIs' leaves most of the security work undone. That is a reason to read it as a map, not as a build manual.

05 Register

What has to be true

  • Network · Not re-measured

    Services stay online at the endpoint they registered. A dead listing is a failed hire.

  • Model · Not re-measured

    Callers can verify they got the service they paid for. Many AI outputs have no cheap check.

  • Model · Not re-measured

    Reputation is not for sale more cheaply than the damage a bad service can do.

  • Model · Not re-measured

    An agent that spends has been given a budget by someone who can revoke it.

06 Divergence

What happened after the paper

SingularityNET launched a network and a token and has continued as an AI-services project. The broader industry moved toward tool-using agents with keys and budgets, which is the problem the paper named early. The 2017 text is not a specification of any current agent runtime. It is the marketplace sketch.

A later client, parameter set, or reward formula is a different object. Cite this paper for the mechanism. Cite a primary release for the network. This desk has not re-run the proof.

07 Pre-mortem

What to check before you use the idea

  5. 0 of 4 marked on this browser. A mark is a reading note, not a pass, a rating, or a recommendation.

08 Anatomy

The paper, in the order a builder needs

The problem it names

AI services in 2017 sat behind separate companies' APIs, with no shared way for one service to hire another. The paper argues that a network of agents needs discovery and payment as protocol objects, or the only integrator will be a platform company.

What the design proposes

  • Services publish descriptions and prices.
  • An agent can call another agent and pay from a wallet without a human in the loop of that call.
  • Reputation and curation are acknowledged as necessary, because an open catalogue will contain services that do not work.

How the mechanism is specified

  • The paper is architectural. It does not give a consensus protocol or a proof about reputation sybils.
  • Payment is necessary in the design so that a call has a cost. The token section should be read as a proposed rail, not as a valuation.
  • Later SingularityNET deployments and spin-out products are not automatically specified here.

What this page does not treat as proven

  • A white paper that scopes 'a network for AIs' leaves most of the security work undone. That is a reason to read it as a map, not as a build manual.
  • Reputation in an open agent market is unsolved by naming it.
  • We do not describe a token sale.

Why the desk still reads it

Useful as the early, explicit statement that agents will be customers. Pair it with Bittensor, which is narrower and more mechanical, and with the studio's own rule that an agent needs a budget and a way to be stopped.

09 Lexicon

Terms, opened into the record

AI service
A networked capability with a price and an endpoint, not a model file by itself.
Agent customer
A service that discovers and pays another service without a person approving each call.
Registry
The catalogue listings live in. Discovery is only as good as this catalogue and its reputation layer.
Reputation
The paper's placeholder for how bad services get avoided. It needs a concrete mechanism before it is a control.

10 Repository

Every linked record on this page

Underlined words open a page that already exists: a concept, a protocol profile, a failure record, or another paper. If a word is not underlined, this desk does not have a record for it.

This is Blockchain Lab's reading of a public design paper. It is not the paper, not a copy of it, and not an offer of tokens, equity, custody or a partnership. Later network behaviour can diverge from the text. Nothing here is investment, legal or technical advice.

Research status: Design paper. Last reviewed: 1 October 2026. This is a reading of a public paper, not investment, legal or security advice.