Sourced from public evidenceLast checked 16 October 2026 with AI-assisted research. Editor: Akshay Dubey.MethodologyCorrections log

Everything below comes from public sources. The company has not reviewed this profile. Data last verified: 16 October 2026. Status labels are defined in the methodology. About this record.

Publisher interest: HumanAudit Inc., which publishes this site, sells governance documentation and plans NHI governance packs, including an RFP kit that does not name, rank or recommend any vendor. It builds and sells no NHI scanner and no NHI discovery, monitoring or security tool.

At a glance

Scroll sideways to see all columns.

HeadquartersMountain View, California, USA (address given in the terms of service)VerifiedSource: MintMCP: Terms of service (Official company website). The only address found on the company site is in the terms of service.
FoundedNot publicly disclosedNot publicly disclosedSource: MintMCP: Introducing MintMCP (company blog) (Official company announcement). The MintMCP name was introduced in a company blog post dated 5 February 2026. We did not find a stated founding year.
OwnershipIndependent, venture-backedVerifiedSource: MintMCP: About MintMCP (Official company website). Based on the investor list on the company About page. We found no announcement of an acquisition.
Legal entityDependable AI, Inc., a Delaware corporation doing business as MintMCPVerifiedSource: MintMCP: Terms of service (Official company website).
FoundersJiquan Ngiam, Vijay Vasudevan, Bozhi See and Phillip BensaidVerifiedSource: MintMCP: About MintMCP (Official company website). The About page lists four founders and does not state individual titles.
Total fundingNot publicly disclosedNot publicly disclosedSource: MintMCP: About MintMCP (Official company website). Investors are named by the company, but no announcement of a round or amount has been found.
InvestorsCoatue, Maven Ventures and Hustle Fund, with angel investors including Andrej Karpathy and Jeff DeanPublicly disclosedSource: MintMCP: About MintMCP (Official company website). Stated by the company. The February 2026 launch post also names WVV Capital.
ProductsMCP Gateway, LLM Gateway, Mint Guard, Agent Monitoring and Coworker Agents (AI teammates in Slack, each with its own identity)VerifiedSource: Mintive: Meet Mintive (Official company announcement).
Named customersCoursera, Stability AI, Deerfield Group and Modern Treasury (named executives quoted on the company site)Publicly disclosedSource: MintMCP: homepage (Official company website). Named with quoted executives on the company site.
AttestationsCompany states it is SOC 2 Type II audited, and offers HIPAA documentation and business associate agreementsVendor-suppliedSource: MintMCP: About MintMCP (Official company website). Company statement. HIPAA is a law, not a certification.
DeploymentHosted service; the pricing page also mentions self-hosted deploymentVendor-suppliedSource: MintMCP: Pricing (Official company website).
PricingCustom quote, per active user plus a platform fee; no public price listVerifiedSource: MintMCP: Pricing (Official company website).
NameMintive, renamed from MintMCP (announced 12 October 2026). The announcement says existing customer deployments carry over; it does not say whether the legal entity changed.VerifiedSource: Mintive: Meet Mintive (Official company announcement).
Websitewww.mintmcp.com still serves the site, branded Mintive; the announcement links product pages on mintive.com, while the blog, sign-in and documentation remain on mintmcp.comVerifiedSource: Mintive: Meet Mintive (Official company announcement).

Source audit: AI research assistants checked the facts in this profile line by line against public sources on 16 October 2026. Akshay Dubey, the editor, is accountable for publication. Each fact carries a label for the kind of evidence behind it. What could not be confirmed is listed under What we could not verify.

What changed recently

  • 12 Oct 2026: Renamed from MintMCP to Mintive, bringing its gateway, guardrail, monitoring and Coworker Agents products under one name
  • 23 Jun 2026: Partnership with Okta on Cross App Access announced
  • 5 Feb 2026: Company blog post introduces MintMCP, with MCP Gateway and Agent Monitor

Categories and identity types

Scroll sideways to see all columns.

CategoriesMCP & agent gateways (adjacent)
Identity and credential types (as the company describes them)AI agents, API keys and tokens, OAuth clients (machine to machine)

Capability evidence is in the capability matrix below.

Funding history

No funding round is recorded for this entry. For acquired companies, public companies and products of larger companies, see ownership above.

Disclosed investors: Coatue, Maven Ventures, Hustle Fund, WVV Capital; angels including Andrej Karpathy, Jeff Dean, Frederic Kerrest, Michael Grinich, Scott Belsky, Tom Willerer and Brian ShultzPublicly disclosedSource: MintMCP: About MintMCP (Official company website). Company-stated on the About page and the February 2026 launch post. Amounts and dates are not given.

The company names its investors but does not publish round sizes or dates. Aggregator pages list a seed round of about $3.8M dated December 2023. We did not find a primary source for that figure, so no funding row is shown.

What MintMCP does for agent identity

MintMCP sells a gateway that sits between AI clients and agents (the company names Claude, Cursor and ChatGPT) and the MCP servers that expose company tools and data. Administrators publish curated sets of tools, which the company calls Virtual MCPs, and assign them by role. The company states that sign-on uses SAML or OIDC single sign-on and that role membership can be driven by SCIM.

The part most relevant to non-human identity is the feature called Agent Identities. The documentation describes each autonomous agent as its own principal, with its own credential, tool access and audit trail, separate from the person who created it. Two credential types are documented: bearer keys, which can be named, given an expiry and revoked one by one, and machine-to-machine OAuth using the client credentials flow, which issues short-lived access tokens. The documentation states that client secrets can be rotated without invalidating live tokens, and that deleting an agent removes its credentials and downstream authorisations together.

On accountability, the documentation says the administrator who provisioned an agent is recorded as its creator, and that agent traffic is logged separately from human traffic.

Monitoring, guardrails and audit

Agent Monitor is described as tracking usage by person, agent, server and model, including token and spend breakdowns. Mint Guard is described as built-in protection for MCP tool calls: detection and blocking of prompt injection, and detection and logging of credentials, secrets and personal data in tool traffic. The company states that audit logs can be exported to a SIEM and streamed over OTLP.

In June 2026 the company announced a partnership with Okta on Cross App Access, under which agent and MCP connections are authorised against a customer's Okta identities and policies. The post says support is available to MintMCP customers that already run Okta, on request.

Where it sits in the market

MintMCP is a gateway company, not an NHI discovery or governance platform. It controls and records what agents do through the connections that pass through it. That places it beside other MCP and agent gateways tracked on this site, such as Runlayer, Obot AI and TrueFoundry, and next to agent identity providers such as Keycard and Arcade.dev. Under its Okta partnership, the enterprise identity provider supplies identities and policy, and the gateway enforces them at the tool level.

The MintMCP brand was introduced in February 2026, and the legal entity is Dependable AI, Inc. The founders are described by the company as early members of Google Brain who worked on TensorFlow, and on Coursera. Several customers are on record by name, including quoted executives at Coursera, Stability AI, Deerfield Group and Modern Treasury.

What public documentation does not establish

This section records what the public documentation reviewed for this profile does and does not describe, as of the verification date. Where it says a product does not do something, read that as "is not described as doing so in the sources reviewed". It is not a finding about the product, which we have not tested.

Public documentation reviewed for this profile did not establish discovery or inventory of non-human identities that exist outside the gateway, such as service accounts, cloud roles, certificates or OAuth applications in other systems. The documented scope is agents and MCP servers that are registered with MintMCP.

We did not find public documentation confirming secrets vaulting as a product, certificate management, or rotation of credentials held in third-party systems. A company blog post mentions workload identity federation for agents; the Agent Identities documentation we reviewed describes bearer keys and machine-to-machine OAuth only.

Funding amounts, founding year, customer count and prices are not publicly disclosed. The SOC 2 Type II and HIPAA statements are the company's own. We found no independent technical evaluation of the product.

Common questions

What does MintMCP do for non-human identity?
Agents often run on a person's API key, so their actions cannot be separated from that person's and cannot be revoked alone. MintMCP's documented answer is to make each agent a separate principal at the gateway, with its own credential, tool scope and log. (Our analysis of public documentation.)
Who owns MintMCP?
Independent, venture-backed.
Who leads MintMCP?
Jiquan Ngiam, Co-founder; Vijay Vasudevan, Co-founder; Bozhi See, Co-founder. See Leadership.
Does MintMCP cover AI agents?
Evidence status: Documented by vendor. Each agent is documented as its own principal with its own credential, tool access and audit trail.
Who are MintMCP's competitors?
In our analysis of product scope: Runlayer, Obot AI, TrueFoundry. See Competitors and alternatives.

Timeline

Scroll sideways to see all columns.

DateEventSource
12 Oct 2026Renamed from MintMCP to Mintive, bringing its gateway, guardrail, monitoring and Coworker Agents products under one nameMintive: Meet Mintive
23 Jun 2026Partnership with Okta on Cross App Access announcedMintMCP: Okta and MintMCP partner on Cross App Access (company blog)
5 Feb 2026Company blog post introduces MintMCP, with MCP Gateway and Agent MonitorMintMCP: Introducing MintMCP (company blog)

Leadership

Scroll sideways to see all columns.

NameRoleBackground stated in sources
Jiquan Ngiam [source]Co-founderThe About page lists Google and Coursera as prior employers. He is the named author of the company's Okta partnership post.
Vijay Vasudevan [source]Co-founderThe About page lists Google as a prior employer.
Bozhi See [source]Co-founderThe About page lists Brilliant Home and Shopkick as prior employers.
Phillip Bensaid [source]Co-founderThe About page lists Disney+ and Amazon as prior employers.

Public professional facts only.

Products

MCP Gateway

A gateway that routes AI client and agent connections to MCP servers through one controlled point. Administrators bundle tools by role, switch individual tools on or off, and manage connector credentials centrally. The company states support for SAML and OIDC single sign-on and SCIM-driven role membership. [source]

Agent Identities

Gives each autonomous agent its own identity in the gateway, separate from the person who created it. Agents authenticate with bearer keys or with OAuth client credentials that yield short-lived tokens. Credentials can be rotated and revoked per agent. Documented integrations include Claude, ChatGPT workspace agents and Cursor cloud agents. [source]

Agent Monitor

Reports usage by person, agent, server and model, with token and spend breakdowns. The launch post describes it as the runtime visibility and policy layer for agent actions. [source]

Mint Guard

Managed protection for MCP tool calls. The company describes detection and blocking of prompt injection, and detection and logging of credentials, secrets and personal data. Integration with data loss prevention tools is also listed. [source]

LLM Gateway

Listed on the pricing page as governed access to models. Public technical detail reviewed for this profile was limited. [source]

Capability matrix

This matrix records what public evidence shows. It is not a score. Documented by vendor means the company's own documentation or product pages describe the capability; it is not independent verification and says nothing about depth. Not publicly verified means we did not find public documentation, which is different from saying the product lacks the capability.

13 documented by the vendor, 0 independently reported, 8 not publicly verified.

Scroll sideways to see all columns.

CapabilityEvidence statusWhat the evidence says
NHI discovery Not publicly verifiedPublic documentation reviewed for this profile did not establish discovery of identities outside the gateway.
NHI inventory Not publicly verifiedThe product lists agents and MCP servers registered in the gateway. We did not find documentation of a wider NHI inventory.
Ownership assignment Documented by vendorDocumentation states that the administrator who provisioned an agent is recorded as its creator. A separate accountable-owner field is not described. [source]
Attribution to humans Documented by vendorUsage is tracked by person and by agent, and identity forwarding headers can pass a user identity upstream. Delegation chains from a human to an agent are not described. [source]
Access analysis Not publicly verifiedPublic documentation reviewed for this profile did not establish this capability.
Authorization policy Documented by vendorAccess is scoped by tool, dataset and action through role-based bundles and Virtual MCPs, with rules and middleware applied at the gateway. [source]
Least privilege Documented by vendorDocumentation states an agent sees only the MCPs created for it. Analysis of unused permissions is not described. [source]
Lifecycle management Documented by vendorAgents are created by administrators, and deleting an agent removes its credentials and downstream authorisations. Scope is agents inside the gateway. [source]
Credential rotation Documented by vendorClient secrets for agent OAuth clients can be rotated without invalidating live tokens. Rotation of credentials in third-party systems is not described. [source]
Secrets management Not publicly verifiedThe gateway stores and injects connector credentials centrally. We did not find public documentation presenting it as a secrets vault.
Certificate management Not publicly verifiedPublic documentation reviewed for this profile did not establish this capability.
Service accounts Not publicly verifiedPublic documentation reviewed for this profile did not establish this capability.
API keys Documented by vendorAgents can use bearer keys issued by the gateway, with optional names, expiry dates and individual revocation. [source]
OAuth apps and tokens Documented by vendorCovers OAuth brokering for MCP servers and OAuth client credentials for agents. Governance of third-party OAuth applications is not described. [source]
Workload identity Not publicly verifiedA company blog post mentions workload identity federation for agents. The Agent Identities documentation reviewed did not describe it.
Cloud identity Not publicly verifiedPublic documentation reviewed for this profile did not establish this capability.
AI agents Documented by vendorEach agent is documented as its own principal with its own credential, tool access and audit trail. [source]
Agent governance Documented by vendorRole-based tool bundles, tool-level switches and gateway rules govern what agents and AI clients can call. [source]
Monitoring and detection Documented by vendorAgent Monitor tracks usage by person, agent, server and model. Mint Guard is described as detecting prompt injection, secrets and personal data in tool calls. [source]
Remediation Documented by vendorDescribed as blocking of flagged tool calls and revocation of agent credentials. Wider remediation workflows are not described. [source]
Audit evidence Documented by vendorPer-agent audit trails and audit logs exportable to a SIEM or over OTLP are described. [source]

Customers

Named with evidence

  • Coursera: case study with the CTO on record [source]
  • Stability AI: on-record quote from the Head of Security, and case study [source]
  • Deerfield Group: on-record quote from the CTO, and case study [source]
  • Modern Treasury: on-record quote from an engineering leader, and case study [source]
  • Flux: on-record quote from the founder and CEO [source]

Company statements, not verified

  • The company states that teams at Braze and DNSFilter also use its product. We did not find a case study or quote for these names. [source]
  • The company site shows further logos, including Ro, Arlo, Earnin, Capital on Tap, Sharesies, Workleap, Caribou and Eurostar. A logo is not verified customer evidence. [source]
  • The company states that it is SOC 2 Type II audited, with continuous monitoring through Drata, and that it provides HIPAA documentation and signs business associate agreements. [source]
  • The company lists Flashfood among its customers (case study listed on the company site). [source]
  • The company lists Workstream among its customers (case study listed on the company site). [source]
  • The company lists Bared Footwear among its customers (case study listed on the company site). [source]

Positioning: claim and analysis

What the company says: MintMCP describes itself as the governance layer for AI agents: scoped data access through its MCP Gateway and runtime visibility through Agent Monitor. [source]

What NHI problem it addresses (our analysis): Agents often run on a person's API key, so their actions cannot be separated from that person's and cannot be revoked alone. MintMCP's documented answer is to make each agent a separate principal at the gateway, with its own credential, tool scope and log.

Where the evidence is strongest (our analysis): The agent identity model is documented in specific terms: two credential types, rotation without token invalidation, per-key revocation and a recorded creator. Named executives at several customers are quoted on the company site. The Okta Cross App Access partnership ties it to an enterprise identity provider.

What public information does not establish:

  • Public documentation reviewed for this profile did not establish discovery or inventory of non-human identities outside the gateway.
  • We did not find public documentation confirming secrets vaulting, certificate management or rotation of credentials held in third-party systems.
  • We did not find a primary source for any funding round, amount or date.
  • We did not find independent technical evaluation of the product. The SOC 2 Type II statement is the company's own.

Competitors and alternatives

These groupings are our analysis of overlapping product scope. They are not rankings, and order carries no meaning.

Scroll sideways to see all columns.

RelationshipCompany or approachWhy it overlaps
Direct competitorRunlayerMCP gateway and agent governance
Direct competitorObot AIMCP gateway and catalogue
Direct competitorTrueFoundryAI and MCP gateway
AdjacentKeycardidentity and access for AI agents
AdjacentArcade.devauthorised tool calling for agents
AdjacentSolo.io (agentgateway)open source agent gateway
Platform or incumbent overlapKongAPI gateway vendor with AI and MCP gateway features
Platform or incumbent overlapOktaidentity provider extending policy to agents; also a MintMCP partner
Substitute approachDirect MCP server connections with per-user API keyscommon default where no gateway is in place

Side by side comparisons

Recent news

  • 23 June 2026: MintMCP and Okta partner on Cross App Access. Agent and MCP connections through the gateway can be authorised against a customer's Okta identities and policies. The post says support is available on request to customers running Okta. [source]
  • 5 February 2026: MintMCP introduced with MCP Gateway and Agent Monitor. A company blog post introduced the MintMCP name and named Coatue, Maven Ventures, Hustle Fund and WVV Capital as backers. [source]

What we could not verify

  • Founding year, and the history of the legal entity Dependable AI, Inc. before the MintMCP name. The founders have been associated with an earlier product named Lutra AI; no primary source confirms this.
  • Individual executive titles. The About page lists founders without titles.
  • Whether workload identity federation for agents is shipping: a blog post mentions it, the Agent Identities documentation does not.
  • The contents of the company trust centre and its stated Agentic AI Foundation membership.

This list covers points where public sources were missing, unclear or could not be opened. It is not complete. If you can supply a public source, tell us.

Additional sources for the sections above

  1. MintMCP: homepage (Official company website, 2026)
  2. MintMCP: About MintMCP (Official company website, 2026)
  3. MintMCP: Introducing MintMCP (company blog) (Official company announcement, 2026-02-05)
  4. MintMCP: Agent identities documentation (Official product documentation, 2026)
  5. MintMCP: MCP gateways for AI startups (company blog) (Official company website, 2026-09-09)
  6. MintMCP: Pricing (Official company website, 2026)
  7. MintMCP: Case studies index (Official company website, 2026)
  8. MintMCP: Terms of service (Official company website, 2026-02-04)
  9. MintMCP: Okta and MintMCP partner on Cross App Access (company blog) (Official company announcement, 2026-06-23)

Other companies in this category

MintMCP (now Mintive) is listed under MCP and agent gateways. The companies below share that category in our directory. This is a category listing, not a ranking, and inclusion does not mean the products are equivalent.

Scroll sideways to see all columns.

CompanyOwnershipFunding (as stated in its profile)AI agent coverage (as stated)
Obot AIIndependent$35M seed co-led by Mayfield and Nexus Venture Partners (23 September 2025)Yes (MCP gateway and Nanobot agent framework)
TrueFoundryIndependent$19M Series A led by Intel Capital (announced 6 February 2025); $21M totalYes (Agent Gateway, TrueForge open-source agent harness)
RunlayerIndependent$42M total stated by the company: $11M seed (November 2025) and $30M Series A (June 2026) led by Felicis with Khosla Ventures participatingYes (core)
KongIndependent$345M raised to date, as of 19 November 2024Yes (MCP Registry, MCP tool ACLs, Agent Gateway for agent-to-agent traffic)
Solo.io (agentgateway)Independent$175M total per the company about page ($171.5M stated in the October 2021 release); $135M Series C at a $1B valuation led by Altimeter CapitalYes (agent-to-agent, agent-to-tool and agent-to-LLM traffic)
Traefik LabsIndependentJust over $11M in total, as reported by TechCrunch in May 2023Yes (MCP Gateway, October 2025)

Frameworks and guides that apply to this category

Full list: NHI company directory.

Sources

  1. MintMCP: About MintMCP (Official company website, 2026)
  2. MintMCP: Agent identities documentation (Official product documentation, 2026)
  3. MintMCP: Case studies index (Official company website, 2026)
  4. MintMCP: MCP gateways for AI startups (company blog) (Official company website, 2026-09-09)
  5. MintMCP: homepage (Official company website, 2026)
  6. MintMCP: Introducing MintMCP (company blog) (Official company announcement, 2026-02-05)
  7. Mintive: Meet Mintive (Official company announcement, 2026-10-12)
  8. MintMCP: Okta and MintMCP partner on Cross App Access (company blog) (Official company announcement, 2026-06-23)
  9. MintMCP: Pricing (Official company website, 2026)
  10. MintMCP: Terms of service (Official company website, 2026-02-04)

No company pays to appear on NHI Governance. Spot an error? Tell us; corrections are logged publicly.

About this record

Research summary. This profile is compiled from public sources by NHI Governance. Inclusion is not an endorsement, certification, security assessment or investment recommendation, and does not imply any partnership or affiliation with MintMCP. Names and trademarks belong to their owners. Statements marked as analysis are our opinion.

Research status
Researched from public sources; sources checked
Last verified
Next review due
2026-12-29
Editorial responsibility
Akshay Dubey (editor), HumanAudit Inc.
Company input
None recorded. Under our editorial policy, companies do not approve pages.

Something wrong or out of date? Report it. Companies have a right of reply.

Monthly change log

Know when the NHI landscape changes.

One email a month from this reference: new deals, standards status changes and corrections. No sponsored content. Unsubscribe from any email.

We first email you a link to confirm. Privacy policy.