Wand AI
Also known as: Wand OS, Wand Synthesis AI, Agent Control Panel
Agentic labor infrastructure and operating system for the hybrid workforce, letting governments and global enterprises create, execute, orchestrate, and govern AI agents that work alongside humans, from a single process up to entire divisions.
Wand AI is a Palo Alto company, founded in 2022 by Rotem Alaluf with talent from DeepMind, Google Brain, and Microsoft Research, that describes itself as the world's first agentic labor infrastructure provider built for governments and global enterprises. Its platform, Wand OS, is the system through which organizations operate a unified human and agent workforce: teams add agents, define their roles, objectives, and guardrails, connect the tools and data they need, and evolve their skills over time, coordinating multi agent, multi human workflows from a single command center. A patent portfolio covers the operating mechanics of the agentic workforce. The company reports use by more than one hundred thousand knowledge workers and hundreds of medium and large enterprises, with enterprise proof points including a strategic Franklin Templeton deployment for agentic AI in asset management announced in late 2025 and a large scale partnership with Presight, the Abu Dhabi analytics company majority owned by G42.
The product suite spans four layers matched to AI maturity: an Agent Control Panel to configure, monitor, and govern every agent with live performance data, budget restrictions, and compliance controls; a Collaboration Platform, a shared chat workspace where humans prompt and coach agents in real time; Process Automation, where agents run complex workflows end to end; and Division Automation, handing an entire function to AI. Governance is central, with guardrails on policies, objectives, access, and autonomy, decision tracking, and agent accountability, and agents escalate edge cases to humans based on confidence levels. Technically, Wand positions itself as a multi agent cognitive layer sitting on top of AI models rather than building models itself, routing queries to the best combination of specialized agents, and it lets users target specific data sources without moving the data, with intelligent data discovery across thousands of emails, documents, and data sheets. A 2026 platform update added industry specific agents, including risk mitigation for insurers and memo analysis for venture capital and private equity.
Wand is described as SOC2 ready with flexible deployment on premises, in a private cloud, or hosted, and is listed on the Microsoft commercial marketplace. It is a strong fit for large enterprises and public sector organizations that want a governed hybrid workforce grounded in their own data with flexible deployment, and a weaker fit for small teams seeking a low cost self serve tool, since pricing is enterprise and quoted through sales.
Vendor details
Canonical URL
https://wand.ai
Category
Multi-agent platform
Subcategory
Agentic workforce operating system
Funding status
Independent, headquartered in Palo Alto, California, founded in 2022 by Rotem Alaluf, and seed stage per Crunchbase, backed by world renowned entrepreneurs and leading US venture capital firms with team talent from DeepMind, Google Brain, and Microsoft Research. Wand reports more than one hundred thousand knowledge workers and hundreds of medium and large enterprise customers, with named enterprise relationships including Franklin Templeton and a strategic partnership with Presight, and holds a patent portfolio over the operating mechanics of the agentic workforce.
Company status
independent
Use cases & customers
Primary use cases
Target customers
Deployment options
Integrations
Wand positions itself as a multi agent cognitive layer that sits on top of AI models rather than building them, routing each query to the best combination of specialized agents. Agents work across systems, tools, and departments, connect to the tools and data they need, and can target specific enterprise data sources without moving the data. The platform is listed on the Microsoft commercial marketplace, and deployment options span hosted, private cloud, and on premises.
In practice
An asset manager wants agentic AI in its investment process without losing control. Wand's workforce and agent management technologies let Franklin Templeton deploy agents at scale under strict oversight, compliance, and risk control.
An enterprise worries about agents acting outside policy. Wand's control panel enforces budgets, compliance controls, and guardrails on access and autonomy, tracks every decision, and has agents escalate edge cases to humans based on confidence.
A knowledge worker needs answers buried across the organization. Wand's intelligent data discovery reaches across thousands of emails, documents, and data sheets, and its agent network assembles the right specialists to answer in minutes.
Sources & related URLs
Related / legacy domains
Research sources
Capability coverage
10.0 / 14 capabilities · 71%
| Integrations & Tool CallingAgents work across systems, tools, and departments, connect to the tools and data they need, and target specific data sources without moving the data, wand.ai and a-teaminsight coverage retrieved 2026-07-07 | Full |
|---|---|
| Workflow OrchestrationMulti agent, multi human workflows run from a single command center, agents execute complex workflows end to end through Process and Division Automation, and the cognitive layer determines which combination of agents answers each query, wand.ai product page and blog retrieved 2026-07-07 | Full |
| Knowledge Grounding & RAGIntelligent data discovery gives instant access to information across thousands of emails, documents, and data sheets, and users can target specific data sources without moving data, genai.works listing and a-teaminsight retrieved 2026-07-07 | Partial |
| Human Oversight & GuardrailsGuardrails cover policies, objectives, access, and autonomy, with budget restrictions, compliance controls, role based controls, decision tracking, and agents that escalate edge cases to humans based on confidence levels, wand.ai product page and blog retrieved 2026-07-07 | Full |
| Security, Identity & GovernanceDescribed as SOC2 ready with built in governance and security features, compliance controls, role based controls, and agent accountability, though certification is stated as ready rather than completed, wand.ai and Crunchbase profile retrieved 2026-07-07 | Full |
| Observability & AuditabilityBuilt in dashboards, decision tracking, and agent accountability at every stage, with the Agent Control Panel monitoring every agent on live performance data and cost, wand.ai and blog retrieved 2026-07-07 | Full |
| Memory & State PersistenceAgents learn, adapt, and evolve with the business, retrain automatically, and refine their own intelligence over time, though a first class persistent memory store is not documented, wand.ai and Crunchbase profile retrieved 2026-07-07 | Partial |
| Deployment & Data ResidencyFlexible options for on premise, private cloud, or hosted deployment, with data targeted in place rather than moved, wand.ai homepage retrieved 2026-07-07 | Full |
| Prebuilt Agents, Templates & PacksShips dozens of uniquely skilled agents plus industry specific agents such as insurer risk mitigation and memo analysis for private equity, while emphasizing creating agents in minutes from detected skill gaps, prweb release and a-teaminsight retrieved 2026-07-07 | Partial |
| Triggers & Channel CoverageThe command center triggers process automations and agents sense operational needs to drive processes independently, though broad multichannel trigger coverage is not documented, wand.ai product page and Crunchbase retrieved 2026-07-07 | Partial |
| Model Flexibility & RoutingDescribed as a multi agent cognitive layer that sits on top of AI models rather than building them, acting as a brain that routes each query to the best combination of agents and generative AI, a-teaminsight coverage retrieved 2026-07-07 | Full |
| APIs, SDKs & MCP ExtensibilityThe platform integrates with existing enterprise systems and is listed on the Microsoft commercial marketplace, but a public API, SDK, or MCP surface is not documented, wand.ai and Microsoft marketplace listing retrieved 2026-07-07 | Partial |
| Testing, Debugging & OptimizationThe agentic network enables self criticism to reduce hallucinations, agents validate their own outputs, and the control panel tracks live performance, though a dedicated evaluation harness is not documented, prweb release and blog retrieved 2026-07-07 | Partial |
| Browser & Computer UseNo browser or computer use capability is described, wand.ai retrieved 2026-07-07 | Unable to verify |
Pricing
Not public; enterprise platform quoted through sales, scoped from single process automation up to entire divisions
enterprise agreement, believed scoped to agents, processes automated, and deployment model
What is public
No list prices, units, or entry point are published. The four product layers, deployment options spanning hosted, private cloud, and on premises, and the scale at your pace positioning are public, but pricing is not.
Billing mechanics
Presumed enterprise agreement negotiated with sales, with scope tied to which product layers are deployed, from Agent Control Panel and Collaboration Platform through Process and Division Automation, and to the deployment model chosen. Listed on the Microsoft commercial marketplace, which may support cloud commitment procurement.
Cost watchouts
Expanding from process automation to division automation, or from hosted to on premises deployment, may materially change contract size beyond an initial quote.
Variable cost rationale
Deployments are explicitly designed to scale at the customer's pace from a single process to full divisions, so cost presumably grows with the number of agents, processes, and departments automated, though pricing is negotiated as an enterprise agreement rather than metered on usage.
Additional watchouts
Confirm how pricing scales as automation expands from single processes toward division level deployment, and whether on premises or private cloud deployment carries different commercial terms than hosted.
Sales call required
Yes — required for paid access
Free / trial
No public free tier; contact and demo led
Key ambiguities
Whether pricing is seat based for the collaboration layer, workload based for the automation layers, or a blended enterprise agreement.
Missing data
No public anchor for entry price, billing unit, or contract structure.
Related vendors
- CrewAI — Open-source multi-agent orchestration framework for building…
- Relevance AI — Platform for building and deploying a workforce of AI agents with a…
- Agent Zero — Open-source multi-agent framework with its own Linux-VM agent,…
- Akka — Enterprise agentic AI platform that builds and runs multi agent…
- Databricks Mosaic AI — Databricks Mosaic AI platform for building, deploying, and…
- Dataiku — Enterprise AI and machine learning platform with an agentic layer…