Arcade Brief
B2B SaaS·AI Agent Runtime·Series A·🇺🇸
Summary
Arcade.dev is a production-ready AI agent runtime that wedges between raw model vendors and enterprise systems by offering native per-user auth, governance (SOC 2/GDPR/CCPA), and a very large pre-built tool catalog (7,500+ integrations). It competes in a split market of gateways/observability stacks and connector/auth runtimes — directly comparable to TrueFoundry, Scalekit, Composio, Merge Agent Handler and Truto — and differentiates on integration breadth and multi-user production focus. Recent momentum is real: a $12M seed (Mar 2025) followed by a $60M Series A (Jun 15, 2026) (total $72M) plus enterprise logos (LangChain, Snyk, Relevance AI) signal serious go-to-market traction, and third‑party revenue estimates imply material share (~34%) of a narrowly scoped SAM. The single biggest tension is depth versus breadth — 7,500+ integrations and published Growth pricing ($25/month) are strong signals, but integration quality, enterprise adoption beyond named logos, geographic reach and competitive response are open questions; a first meeting should validate integration quality, commercial pricing/ACV and enterprise references.
Bull case
- 7,500+ pre-built agent integrations combined with SOC 2, GDPR and CCPA compliance — strong technical and compliance foundation for production multi-user agents.
- Named enterprise customers on the site (LangChain, Snyk, Relevance AI, YUR, Sybill, Eddo Learning) that substantiate usage in production and developer-visible credibility.
- Significant capital and momentum: a $12M seed in Mar 2025 followed by a $60M Series A on Jun 15, 2026 (total $72M), indicating investor conviction and runway to scale.
- Published, transparent entry pricing (Growth: $25/month + usage) and a small, senior founding team with explicit authentication and AI engineering expertise supporting developer-led GTM.
Watch-outs
- Crowded, adjacent competitor set (TrueFoundry, Scalekit, Composio, Merge Agent Handler, Truto) offering overlapping gateway, observability or connector features — differentiation must be proven on integration depth, SLAs and enterprise workflows.
- No credible third‑party customer reviews or aggregate ratings found outside site testimonials, leaving customer sentiment and post-sales experience unverified.
- The website does not state which countries or regions Arcade sells to or operates in, creating geographic / compliance / GTM clarity risk for multinational customers.
- High integration count (7,500+) is a strength on paper but raises questions about ongoing maintenance, support model and how many integrations are enterprise‑grade versus lightweight adapters.
From their website
Stats they publish
Customers on site
6 namesPricing
3 plans: Hobby (FREE), Growth ($25/month + additional usage), Enterprise (Custom). Usage-based model with included quotas (user challenges, standard & pro tool executions) and overage rates (e.g. $0.05/user-challenge, $0.01/std exec, $0.50/pro exec). MCP servers billed ~$0.05/hr; no seat licenses.
Key features
- Arcade Runtime
Production-ready runtime that connects AI agents to systems with built-in authentication, tool integrations, and governance.
- Agent Tools Registry
Registry of 7,500+ pre-built agent integrations (Gmail, Slack, GitHub, Salesforce, Microsoft 365, etc.) for agents to perform real actions.
- Agent Authorization
Agent authorization and identity integrations that let agents act with dynamic permissions via existing identity providers.
- Central Control Plane
Centralized governance, monitoring and full audit logs for every action taken by AI agents.
- Deployments & Hosting
Flexible deployment options including cloud, VPC/on-premises, or fully air-gapped installs for data-residency and security control.
Target segments
Testimonials
“The runtime layer from Arcade makes MCP enterprise-ready. Connects to identity providers, enforces agent authorization, and enables real actions in Google, Slack, and Salesforce.”
Harrison Chase · Co-Founder and CEO“By using Arcade, we're able to authenticate users' Twitter/LinkedIn accounts without worrying about refresh tokens, broken auth, or any of the other hassle that comes with setting up auth connections.”
Brace Sproul · AI/ML Engineer“Arcade nails the sweet spot between AI, auth, and developer experience. We're using it and it's insanely useful - finally a product that lets AI agents actually do stuff.”
Randall Degges · Head of Developer & Security Relations“We built an AI sales agent that knows everything about every deal by analyzing calls, emails, and CRM data. With Arcade, that agent can now go the extra mile and effortlessly take secure actions on behalf of a rep - turning a smart assistant into a revenue-driving powerhouse.”
Gorish Aggarwal · Cofounder & CEO“With Arcade, we can bypass the complexity of Google service integration and fast-forward our AI development to building tools for our AI Assistant that teachers can actually use.”
Matt Anthes-Washburn · Co-Founder, Chief Product Officer“Integrating with Arcade’s MCP runtime has been remarkably easy, so we can focus on delivering personalized coaching experiences.”
Jacky Koh · Founder
Compliance & certifications
- Trade name
- Arcade AI, Inc.
- [email protected]
Frequently asked questions
- How does Arcade govern AI agent actions?
- Arcade enforces permissions via your OAuth/IDP. Agents act for the authenticated user; no broad service accounts. Actions stop if permissions are exceeded.
- What is an MCP runtime for AI agents?
- It's the infrastructure layer that lets agents securely connect to tools and systems; Arcade handles auth, policy, and tool execution.
- How does Arcade secure AI agent actions?
- It enforces user-specific permissions via your OAuth flows; agents act only for the authenticated user with visibility and control over tool use.
- Can I deploy Arcade on my own infrastructure?
- Yes. Deploy the MCP runtime in your cloud, a virtual private cloud, on premises, or air gapped to control data and security.
- Can I add custom policies to Arcade for AI agents?
- Yes. Use pre- and post-tool-call hooks to add policies without forking; inspect requests and responses and block or redact as needed.
- What tools can AI agents access using Arcade?
- Access pre-built MCP tools for Google Workspace, Slack, Microsoft, and Salesforce, plus custom tools via the Arcade framework and Registry.
Crunchbase
verified profile
- Arcade Raises $60M to Become the Secure Action Layer Behind Every Production AI AgentBusiness Wire · 2026-06-15
- Arcade.dev Raises $60 Million to Secure AI AgentsTechnology - WSJ.com · 2026-06-15
- Akto Announces Partnerships with Langchain, Portkey, TrueFoundry, Arcade, and LiteLLM to expand AI Agent SecurityPR Newswire · 2026-04-23
- How Antigravity’s AI Subagents Are Reshaping Engineering Teams With ArcadeGeeky Gadgets · 2026-04-08
- Arcade.dev Joins Linux Foundation's Agentic AI Foundation as Gold MemberMorningstar.com · 2025-12-09
- Arcade.dev Authors Core MCP Capability Unlocking Secure AI Agents at ScaleBusiness Wire · 2025-11-25
- Arcade.dev and Lithic Announce Partnership to Create E-Commerce AgentsFF News | Fintech Finance · 2025-08-05
- Arcade.dev and Lithic Announce Partnership to Create E-Commerce AgentsBusiness Wire · 2025-08-05
Founders & Team
6- Alex SalazarFounderCEO & Co-founderPreviously co‑founder and CEO of Stormpath (acquired by Okta); former Vice President of Product Management at Okta.
- Sam ParteeFounderCTO & Co-founderPreviously Principal AI Engineer at Redis; contributor to open-source AI projects such as LangChain and LlamaIndex.
- Nate BarbettiniFounding EngineerPreviously CTO of Cobbler; worked at Okta and Stormpath developing authentication APIs and authored a book on software development.
- Sterling DreyerFounding EngineerFounding engineer at Triton/StreamSQL and contributor at Featureform where he helped build a feature store and large-scale ML infrastructure in Golang.
- Eric GustinFounding EngineerPreviously worked at Microsoft optimizing Azure Core’s Hyperscale Networking and built an AI agent that won the 2023 Microsoft Global Hackathon.
- Demetrios BrinkmannDeveloper Go-to-Market (GTM)Founder of the MLOps Community, which grew to over 70,000 members and runs meetups and conferences for AI/MLOps practitioners.
Arcade's public team page shows Alex Salazar (CEO) and Sam Partee (CTO) as co‑founders, supported by founding engineers Nate Barbettini, Sterling Dreyer, and Eric Gustin, plus developer GTM lead Demetrios Brinkmann. Bios emphasize deep experience across authentication, AI engineering, and developer/community go‑to‑market.
- Multiple unrelated organizations named “Arcade” appear in the provided results (e.g., gamification/sales coaching Arcade, Arcade Beauty); only people explicitly tied to arcade.dev were included.
- Crunchbase and other aggregators list additional names (e.g., Mariam Naficy) that conflict with the arcade.dev team page; those aggregator-only entries were not treated as founders here.
- The arcade.dev page may list more team members elsewhere; the listed headcount reflects only the identifiable people provided in the cited team bios.
Market Size
A defensible top‑down published proxy is the global AI runtime optimization market (Technavio) at USD 3.04B (2025). A narrow bottom‑up SAM using public MCP registry counts (17,468 servers) and Arcade's publicly listed Growth ACV ($25/month → $300/year) yields ≈ USD 5.24M serviceable revenue; third‑party revenue estimates (~USD 1.8M/year) imply Arcade already captures ~34% of that narrowly scoped SAM. All figures carry material caveats about category scope and pricing assumptions.
- Technavio ↗In 2025, the AI runtime optimization market was valued at USD 3.04 billion. · Global · 31.2% CAGR
- Atlan (what-is-model-context-protocol page) ↗As of the first quarter of 2026, there were approximately 17,468 MCP servers indexed across public registries, up from 1,200 servers a year earlier. · Global
- Growjo (third‑party company estimate) ↗Approximately $1.8 million per year, with revenue per employee at $140,000. · Unknown
- TechCrunch ↗Arcade raises $12M seed funding (announced March 2025). · Unknown
- Arcade (company website)Published pricing: Growth: $25 /month +additional usage · Global
- Potential customer count (proxy)17,468 MCP servers indexed (Q1 2026) — treated as potential serviceable customers for an MCP runtime vendorAtlan — https://atlan.com/know/what-is-model-context-protocol/
- Arcade published ACV (used for SAM math)$25/month → $300/year (Growth plan); additional usage not modelledArcade published pricing (company website, pricing page)
- Arcade estimated annual revenue (used to ground SOM)≈ USD 1.8M/year (third‑party estimate)Growjo — https://growjo.com/company/Arcade.dev
- Traction and scale used to justify obtainable share7,500+ pre-built agent integrations; named production customers (LangChain, Snyk, Relevance AI, YUR, Sybill, Eddo Learning); SOC2/GDPR/CCPA; $12M seed (Mar 2025)Arcade site (known facts); TechCrunch (funding) — https://techcrunch.com/2025/03/18/arcade-raises-12m-from-perplexity-co-founders-new-fund-to-make-ai-agents-less-awful/
- TAM is a parent/adjacent category (AI runtime optimization) because no published top‑down figure for 'MCP runtime' (Model Context Protocol runtimes) was found; the TAM therefore is an upper‑level proxy and may materially overstate or mis‑align with the exact MCP runtime product niche.
- SAM uses 'MCP servers indexed' (17,468) as a proxy for distinct potential customers; public registries may list multiple servers per single organization (overcount) or omit private/enterprise deployments (undercount).
- SAM uses Arcade's publicly listed Growth plan price ($25/month) as ACV; enterprise customers almost certainly pay materially more — using Growth pricing therefore likely underestimates the true serviceable revenue pool.
- Arcade revenue cited (≈USD 1.8M) is a third‑party estimate, not company‑reported; use it as an anchoring datapoint for current share but treat as uncertain.
- SOM shown is the near‑term realised/observed revenue (third‑party estimate) expressed against the narrowly defined SAM; it is a scenario grounded on available traction and estimates, not a guaranteed future outcome.
Competitors
Top 5Market divides between gateways/observability platforms and connector/auth stacks. Arcade is a native MCP runtime with a large pre-built integration library (7,500+), SOC 2 compliance, and a production multi-user focus.
- Merge Agent Handlermerge.dev ↗Tool-calling platform with connectors, per-user auth, DLP and observability; Arcade competes with a native MCP runtime, 7,500+ integrations and SOC 2 governance.
- Composiocomposio.dev ↗Managed MCP runtime with hosted servers, sandboxed execution and 500+ toolkits; Arcade differentiates via multi-user native auth, broader 7,500+ agent integrations and enterprise customers.
- TrueFoundrytruefoundry.com ↗AI Gateway centralizing auth, governance and observability across MCPs; Arcade positions as a native MCP runtime with built-in tool catalog and direct per-user permissions.
- Scalekitscalekit.com ↗Unified agent auth and tool-integration stack with delegated identities and SSO; Arcade wins on a full MCP runtime, broader tool library and SOC 2 compliance.
- Trutotruto.one ↗Unified API engine that auto-generates tools, handles rate limits and zero data retention; Arcade counters with native multi-user authorization and extensive pre-built integrations.
Funding
- Seed Round$12 millionMarch 18, 2025Laude Ventures leadFlybridgeGaruda VenturesHanabi CapitalNeotribe VenturesAndy RachleffBusiness Wire ↗
- Series A$60 millionJune 15, 2026SYN Ventures leadWiproMorgan StanleyPYMNTS ↗
Arcade (arcade.dev) has publicly disclosed a $12M seed raise in March 2025 led by Laude Ventures and a $60M Series A in June 2026 led by SYN Ventures, bringing the company's stated total funding to $72M.
- Multiple unrelated companies named 'Arcade' appear in tracker and news results (e.g., a 2022 demo/product 'Arcade' covered by TechCrunch and several Tracxn/PitchBook profiles). The rounds listed here are those explicitly tied to arcade.dev (the MCP/runtime for AI agents) in the cited press releases and reporting.
- Crunchbase / structured-data snippets in the corpus list investor names and rounds that overlap with the items above; where press releases listed participants those participants were used for round-level investor attribution. Some trackers show additional investors (e.g., Garuda) not confirmed in the primary press release used for the seed round.
- The $72M total is the company's/press-reported total (seed + Series A). Public trackers show divergent figures for other companies named 'Arcade'; the $72M figure applies to arcade.dev per the cited press coverage.
Momentum
- FundingJune 15, 2026Arcade.dev announced a $60 million Series A led by SYN Ventures with strategic investment from Morgan Stanley and Wipro.Business Wire ↗
- PartnershipJune 15, 2026Arcade.dev announced official membership in the Linux Foundation's Agentic Artificial Intelligence Foundation (AAIF) as a Gold Member.Business Wire ↗
- MilestoneJune 15, 2026Arcade reported tool-call volume grew 25x in six months.Business Wire ↗
- MilestoneJune 15, 2026Arcade described having 8,000+ MCP tools purpose-built for agent usage.Business Wire ↗
- FundingMarch 18, 2025Arcade.dev closed a $12 million seed round.Business Wire ↗
Arcade shows accelerating public momentum: a $60M Series A on June 15, 2026 (led by SYN Ventures) accompanied by product/usage milestones reported in the same release, and an earlier $12M seed in March 2025.
- Several 2026 items derive from the company's June 15, 2026 press release (single primary source for multiple claims).
- Input contained multiple unrelated "Arcade" results (e.g., arcade.com, a gaming publisher); those items were excluded from this extract to avoid name-collision errors.
- Operational metrics and tool counts are company-reported in press materials and not independently verified in the provided sources.
- Other mentions in the provided text lacked explicit dates or clear attribution to arcade.dev and were therefore excluded.
Customer Sentiment
No credible third‑party customer reviews or aggregate ratings were found that clearly refer to Arcade (arcade.dev). Public mentions for the name "Arcade" in the supplied sources refer to unrelated consumer arcades, games, or other products with the same name, so customer sentiment for arcade.dev cannot be assessed from these records.
Not enough public customer reviews to judge sentiment — common for early‑stage or B2B developer platform companies.
All supplied third‑party excerpts concern physical arcades, arcade game reviews, consumer apps, or other unrelated products named “Arcade”; none explicitly reference arcade.dev or its site.
Some platforms in the snippets (G2, Capterra, GetApp, Trustpilot) host reviews for products named "Arcade," but the provided text does not confirm those reviews are for Arcade (arcade.dev), so no ratings or themes are reported.
AI Visibility
We ask ChatGPT WITH web search about the company and show its answers plus the sources it cited — the real AI-search experience. The key signal is share of voice: does the company come up when AI is asked about its space, not just when asked by name.
Models asked ChatGPT · web search
Accuracy of what AI said: Accurate — Direct answer correctly describes Arcade.dev's agent-runtime product and reported $12M seed; founder and location claims align with available listings.
Arcade.dev is a San Francisco-based software development company specializing in providing infrastructure that enables AI agents to securely perform real-world actions within enterprise systems. Founded in February 2024 by Alex Salazar, a former executive at Okta, and Sam Partee, formerly of Redis, Arcade.dev focuses on bridging the gap between AI capabilities and practical enterprise applications. (startupintros.com)
Core Offerings: Arcade.dev's platform addresses three primary challenges in deploying AI agents:
- Security and Authorization: The platform integrates seamlessly with existing identity providers, ensuring AI agents can act on behalf of users with appropriate permissions. (businesswire.com)
- High-Accuracy Tool Execution: Arcade.dev offers a comprehensive catalog of agent-optimized tools and a framework for building custom integrations, enhancing the reliability of AI agents in executing tasks across various enterprise systems. (businesswire.com)
- Centralized Governance: The platform provides a unified control plane for all Model Context Protocol (MCP) tools, offering visibility, version control, and audit capabilities essential for scaling AI development safely. (businesswire.com)
Funding and Growth: In March 2025, Arcade.dev secured $12 million in seed funding led by Laude Ventures, with participation from Flybridge Ventures, Hanabi Capital, Neotribe, and Andy Rachleff. This investment supports the company's mission to enhance the security and functionality of AI agents in enterprise environments. (businesswire.com)
As of June 2026, Arcade.dev employs between 11 and 50 individuals. (linkedin.com) The company has experienced significant growth, with a reported 550% increase in employee count over the past year. (growjo.com)
Notable Achievements:
- Enterprise Tech 30 Recognition: In March 2026, Arcade.dev was named to the Enterprise Tech 30 list, highlighting its position as a leading private company in the enterprise technology sector. (arcade.dev)
- Linux Foundation Membership: In December 2025, Arcade.dev joined the Linux Foundation's Agentic Artificial Intelligence Foundation as a Gold Member, underscoring its commitment to advancing the MCP ecosystem and promoting secure, scalable AI agent deployment. (businesswire.com)
- Partnership with Lithic: In August 2025, Arcade.dev partnered with Lithic to develop AI e-commerce applications capable of making secure, authenticated payments, marking a significant advancement in agentic commerce. (businesswire.com)
Through these initiatives, Arcade.dev continues to play a pivotal role in enabling AI agents to perform secure, real-world actions, thereby transforming AI applications from conversational interfaces into trusted automation tools.
Here is a concise list of notable companies and products in the AI agent runtime space:
- OpenAI: Offers the Assistants API and Agents SDK, enabling developers to build custom AI agents with secure tool integrations. (webisoft.com)
- Anthropic: Provides Claude AI models, integrated into platforms like ServiceNow's Build Agent for creating applications and workflows through natural language. (axios.com)
- MCPWorks: An open-source AI agent runtime that allows for the creation of persistent agents with scheduling, webhooks, and encrypted state, running continuously with reduced token usage. (mcpworks.io)
- Northflank: A full-stack cloud platform supporting AI agents, APIs, databases, and background workers, offering microVM-based isolation and on-demand GPUs. (northflank.com)
- Clayva: Provides an agent runtime platform where AI agents autonomously improve products, featuring multi-model agent intelligence and real-time collaboration. (clayva.ai)
- Inworld AI: Offers the Inworld Agent Runtime for building real-time voice and chat agents, with integrated metrics and experiments to optimize user outcomes. (inworld.ai)
- RunForge: A deployment platform for AI agents, handling managed browsers, credential management, human approvals, crash recovery, and customer-facing connection portals. (runforge.sh)
- RuntimeAI: Provides security and compliance infrastructure for autonomous AI, offering identity, policy enforcement, and runtime governance at enterprise scale. (runtimeai.io)
These companies and products are at the forefront of developing and providing AI agent runtime solutions.
As of June 2026, the AI agent runtime space has seen significant advancements, with several companies and products emerging as leaders. Here are some notable ones:
- OpenAI's GPT-5 Runtime: OpenAI has continued to innovate with its GPT series, and the GPT-5 Runtime offers enhanced capabilities for deploying AI agents in various applications. [Source: OpenAI's official website]
- Anthropic's Claude Runtime: Anthropic has developed the Claude Runtime, focusing on safety and interpretability in AI agent deployment. [Source: Anthropic's official website]
- Google's Gemini Runtime: Google's Gemini Runtime provides a robust platform for AI agent execution, integrating seamlessly with Google's ecosystem. [Source: Google's AI blog]
- Microsoft's Azure AI Agent Services: Microsoft's Azure platform offers comprehensive services for deploying and managing AI agents, emphasizing scalability and security. [Source: Microsoft's Azure documentation]
- Meta's LLaMA Runtime: Meta has introduced the LLaMA Runtime, designed for efficient and scalable AI agent operations within its platforms. [Source: Meta's AI research publications]
- Hugging Face's Transformers Agent: Hugging Face provides the Transformers Agent, an open-source solution for deploying AI agents with a focus on community collaboration. [Source: Hugging Face's official website]
- Cohere's AI Agent Platform: Cohere offers a platform tailored for AI agent deployment, emphasizing ease of use and integration. [Source: Cohere's product page]
- AI21 Labs' Jurassic-X Runtime: AI21 Labs has developed the Jurassic-X Runtime, catering to large-scale AI agent applications with advanced language understanding. [Source: AI21 Labs' official website]
- Stability AI's StableAgent: Stability AI provides StableAgent, focusing on stable and reliable AI agent deployment for various use cases. [Source: Stability AI's announcements]
- EleutherAI's GPT-NeoX Runtime: EleutherAI offers the GPT-NeoX Runtime, an open-source alternative for AI agent deployment, emphasizing transparency and community involvement. [Source: EleutherAI's GitHub repository]
These companies and products represent a diverse range of solutions in the AI agent runtime space, catering to different needs and priorities.
Arcade.dev is present in direct queries (AI returns a coherent product description and funding details) but is not surfaced unprompted in category or recommendation lists. The direct description appears trustworthy and matches public reporting on product focus and seed funding, though the company is not yet prominent in broad vendor roundups.
- AI answers can vary run-to-run; visibility observed here may differ on another query.
- Search/indexing coverage for early-stage startups is uneven, so absence from category lists does not prove lack of market activity.
- Vendor roundups often prioritize large platform vendors, so small specialist runtimes like Arcade.dev may be omitted despite accurate direct listings.
Risks & Red Flags
No material public risk signals found. An absence of public concerns isn’t a clean bill of health — early or private companies may simply not generate coverage.
An absence of public reports is not a clean bill of health: early-stage or private companies may have non-public disputes, incidents, or vendor/customer issues that do not appear in press or public filings.
Standard diligence is still recommended (review SOC 2 attestation and related compliance artifacts, ask for customer references, request any material legal/regulatory history) because no material third‑party concerns were publicly identified.