MIND Series B Raises $72 Million to Take Autonomous DLP Into the Enterprise

MIND raised $72 million in Series B funding as businesses struggle to control sensitive data moving through AI tools and autonomous agents. The MIND Series B brings its reported total funding to $112 million. It also turns the company’s autonomous approach to data loss prevention into a much larger enterprise bet.

Crosspoint Capital Partners led the round, while existing investors YL Ventures and Paladin Capital Group also participated. MIND plans to expand in enterprise markets, develop its platform, deepen technology partnerships, and grow its channel network.

The financing lands one year after MIND announced a $30 million Series A. The company says its revenue grew more than seventeenfold during that period, while its customer count increased eightfold. Those growth figures have not been independently verified, and MIND did not disclose their starting values.

That distinction matters because the contest is no longer between MIND and traditional data loss prevention alone. MIND is entering a crowded race against larger platforms, including Microsoft Purview, Varonis, Cyera, Cyberhaven, and other data security vendors.

Its central claim is more specific than simply using AI for security. MIND says software agents can handle classification, policy creation, incident investigation, and remediation across a unified data protection system.

If that approach works at enterprise scale, security teams could replace a substantial amount of manual DLP administration. If it does not, autonomous controls could add another layer that teams must supervise, validate, and explain.

The MIND Series B Funds a Broader DLP Push

The financing gives MIND resources to move from early momentum toward larger, more demanding enterprise deployments.

MIND announced the round on September 17, 2026, through its official MIND Series B release. The Seattle-based company says it has now raised $112 million across its seed, Series A, and Series B rounds.

The latest investment follows an $11 million seed round associated with MIND’s October 2024 emergence from stealth. The company then announced its $30 million Series A in June 2025.

That funding pace reflects an investor view that data protection needs a new operating model. Employees now send information into chatbots, coding assistants, summarization tools, and automated workflows. AI agents can also retrieve and transform information without waiting for a person to move each file.

MIND describes its product as a complete data loss prevention platform. Data loss prevention, or DLP, covers systems that identify sensitive information and restrict risky handling or transmission.

The platform is designed to find information across software services, endpoints, email, on-premises file shares, generative AI tools, and agentic systems. MIND says it classifies files using both their contents and surrounding context.

Those distinctions can matter when identical text carries different risks. A number inside a public research paper presents a different problem from the same number inside a patient record.

MIND then applies policies to risky activity. Depending on the rule, the platform can block an action, ask the user for justification, or guide the user toward a safer option.

The company’s larger ambition is to automate the operational work around those controls. In July 2026, it introduced AI DLP agents for functions such as creating classifiers, producing policies, investigating incidents, and handling remediation.

MIND says these agents can accept plain-language instructions through the Model Context Protocol. MCP is a standard that lets AI applications connect with approved tools and information sources.

The Series B will support continued product work, new hiring, channel relationships, and expansion into important enterprise markets. MIND has not disclosed a valuation for the round.

It has supplied several indicators of commercial traction. The company says it reached eight-figure annual revenue less than two years after leaving stealth. It also reports dozens of customers and deployments covering hundreds of thousands of endpoints.

Those numbers provide useful scale markers, but they leave important gaps. MIND has not disclosed recurring revenue, customer retention, average contract size, or the distribution of deployments across industries.

A seventeenfold increase can represent strong demand. However, percentage growth from an early revenue base does not establish durable enterprise adoption by itself.

The new capital therefore changes more than MIND’s balance sheet. It raises the standard by which buyers and investors will judge the company’s results.

MIND must now show that its system can operate across fragmented enterprise environments. It must also demonstrate that automation reduces work without weakening oversight.

AI Agents Are Breaking the Old DLP Assumptions

The opportunity exists because enterprise AI changes both the speed and the path of sensitive data movement.

Traditional DLP programs usually watch familiar channels. They inspect email, file transfers, web uploads, cloud storage, endpoints, and removable media for regulated or confidential information.

Those systems often depend on predefined patterns and policies. Administrators build classifiers, tune detection rules, review exceptions, investigate alerts, and update controls as business processes change.

This work was difficult before generative AI arrived. It becomes harder when users paste source code, financial records, customer information, or internal plans into external assistants.

Agentic AI adds another complication. An AI agent can access several systems, retrieve documents, combine their contents, and initiate an action during one workflow.

That behavior weakens a basic assumption behind older controls. A risky transfer no longer needs to look like a person knowingly uploading a sensitive file.

Data can be exposed through a prompt, a generated response, a connector, a temporary context window, or an automated tool call. Each step can appear legitimate when inspected separately.

Gartner’s DLP research says conventional DLP struggles with generative AI risks such as encrypted traffic, limited awareness of user intent, and shadow AI. The research recommends more adaptive, risk-based controls that combine information from several security layers.

MIND is betting that the same technology creating the problem can help manage it. Its agents are designed to analyze events, generate policies, and take predefined actions without requiring an analyst at every step.

The intended mechanism starts with discovery. An organization needs to know which sensitive files exist, where copies reside, who can access them, and how applications use them.

Classification adds business meaning to that inventory. A platform must separate regulated information, intellectual property, credentials, contracts, and ordinary business data with acceptable accuracy.

Activity monitoring then provides movement and behavioral context. The system needs to identify whether information is being shared, copied, uploaded, summarized, or exposed to an AI service.

Finally, policy enforcement turns that understanding into an outcome. The platform can permit the activity, warn the user, request justification, quarantine content, or block the action.

MIND says it connects these stages inside one platform. That design seeks to avoid the gaps created when discovery, classification, monitoring, and enforcement come from separate products.

Healthcare offers a clear example. Children’s Hospital Los Angeles says it uses MIND to prevent protected health information from entering AI tools on staff endpoints.

The hospital’s challenge is not simply identifying a patient record. It must allow useful AI adoption while stopping specific information from crossing an unacceptable boundary.

The National Geographic Society describes a different scenario. It uses MIND to label sensitive documents, restrict public sharing, and observe when information interacts with AI agents.

Both cases illustrate why enterprise AI security is becoming a data control problem. Model safety cannot protect information that users or connected agents should never have supplied.

This shift also matters for knowledge-intensive teams. A searchable AI knowledge base becomes more useful as it connects with more work. Its security requirements rise at the same time.

AI systems need enough access to retrieve relevant information. Security teams must still preserve permissions, confidentiality, auditability, and restrictions on onward sharing.

MIND’s premise is that automation can reconcile those demands at machine speed. The Series B gives the company more capacity to test that premise in production.

The Real Contest Is Autonomous Control Versus Human Supervision

MIND must prove that autonomous DLP reduces operational burden without making security decisions harder to trust.

The company presents its agents as an answer to a familiar DLP problem. Security tools can create large queues of alerts, exceptions, policy requests, and classification tasks.

A tool that detects risk but requires constant tuning can transfer work rather than remove it. Security engineers still need to determine which alerts matter and which business activities are legitimate.

MIND’s AI DLP agents aim to take on that repetitive administration. One agent can produce custom classifiers, while another drafts and refines policies based on instructions and observed activity.

Other agents investigate incidents and recommend or perform remediation. The platform is intended to let human teams set strategy while software handles recurring operational tasks.

This approach creates a meaningful tradeoff. Greater automation can accelerate response and support broader coverage, but it can also magnify a mistaken classification or policy.

A false positive is not only an inconvenient alert when an agent can enforce a response. It can interrupt a business process, prevent a legitimate disclosure, or restrict information needed for urgent work.

A false negative creates the opposite problem. The system can appear comprehensive while sensitive information passes through an unrecognized format, application, or agent workflow.

Context-aware classification is supposed to reduce those errors. MIND says its multilayer system evaluates a file’s contents alongside factors surrounding its use.

However, the company has not published enough independent benchmark data to compare its classification accuracy with leading alternatives. It has also not disclosed production false-positive rates across its customer base.

The absence of those details does not invalidate the product. It means prospective buyers should treat accuracy and labor-saving claims as items for controlled evaluation.

A useful proof of concept should include messy, organization-specific information. Teams should test documents with inconsistent labels, copied content, embedded images, partial records, and changing access permissions.

Testing should also cover real AI workflows. A policy may behave differently when data passes through a browser prompt, an approved copilot, an API integration, or an autonomous agent.

Buyers need evidence about more than detection. They should examine how the system explains a decision, records agent actions, handles exceptions, and restores access after an incorrect response.

Human approval boundaries deserve particular attention. Organizations should know which actions MIND can execute automatically and which require review.

They should also determine how those boundaries change by data type, employee role, application, jurisdiction, or business process. A policy appropriate for a marketing document may be unacceptable for clinical data.

MIND has gained credentials that can support enterprise evaluations. The company says it was the first data security vendor accepted into Anthropic’s Cyber Verification Program.

That program provides vetted defensive security organizations with access suited to advanced cyber work. Acceptance indicates that an organization cleared Anthropic’s verification process, not that Anthropic independently validated every MIND product claim.

MIND also says it earned ISO/IEC 42001 certification. ISO/IEC 42001 defines requirements for establishing and improving an AI management system.

The standard covers organizational policies, risk processes, accountability, monitoring, and continual improvement. ISO notes that independent certification bodies issue certifications, while ISO itself does not certify companies.

These credentials can help procurement teams assess governance maturity. They do not replace technical testing inside a customer’s environment.

The central question is therefore measurable. Does the platform decrease investigation and policy work while maintaining acceptable detection, enforcement, and business continuity?

The MIND Series B places that question at the center of the company’s next stage. Enterprise buyers will expect evidence that extends beyond funding announcements and percentage growth.

MIND Enters a Crowded Enterprise Data Security Market

MIND’s integrated pitch must compete with established platforms, heavily funded specialists, and controls already included in enterprise software suites.

The data security market contains several overlapping categories. Traditional DLP focuses on detecting sensitive content and preventing unauthorized movement.

Data security posture management, or DSPM, maps sensitive data, permissions, exposure, and risk across cloud or software environments. Insider risk products examine behavior that can indicate misuse by employees, contractors, or compromised accounts.

AI security platforms add controls around models, prompts, agents, and connected data. Larger vendors increasingly combine pieces of all four categories.

Microsoft can connect Purview controls with the productivity applications many enterprises already use. Varonis brings a long history in data access governance, classification, behavior monitoring, and remediation.

Cyberhaven emphasizes data lineage, which tracks information as users transform and move it across applications. Cyera built its position around cloud data discovery and security posture before expanding its platform.

MIND needs to convince customers that a unified, AI-native architecture creates better operational outcomes. It must do so while competing against vendors with larger sales teams, deeper integrations, and existing enterprise contracts.

Funding across the category reinforces that pressure. Cyera announced a $540 million Series E at a $6 billion valuation in June 2025, according to Cyera financing coverage from Axios.

Cyera had raised about $1.25 billion by that point. Its scale shows how much capital investors have committed to the belief that AI adoption increases enterprise data exposure.

MIND remains much smaller by total funding. Its argument is that an integrated DLP platform with autonomous operations can win without matching every competitor’s breadth.

That focus carries advantages. A younger platform can design around SaaS, generative AI, agents, and modern endpoints without supporting decades of inherited architecture.

It can also unify workflows that older providers developed as separate modules. Buyers may value a simpler operating model if it decreases deployment time and administrative effort.

The same focus creates commercial risks. Enterprise customers often prefer security products that fit existing identity, endpoint, cloud, and compliance systems.

A technically capable standalone platform can still lose when a broader vendor bundles adequate controls into a larger agreement. Procurement simplicity can outweigh a feature advantage.

MIND’s channel expansion therefore matters as much as its product roadmap. Partners can help the company reach buyers, implement policies, and connect the platform with established security operations.

Customer concentration is another unresolved issue. MIND reports dozens of customers, but it has not disclosed how much revenue depends on its largest accounts.

The company also has not reported renewal rates or expansion patterns. Those measures would reveal whether customers broaden deployments after initial evaluations.

Independent reporting confirms the basic financing details and summarizes MIND’s product direction. However, AI data controls coverage also notes that the underlying figures behind MIND’s growth percentages remain undisclosed.

That limitation shapes the competitive analysis. Funding lets MIND hire, build integrations, and support larger customers, but it does not establish a durable market position.

Competitive pressure will also affect product priorities. MIND must expand coverage without recreating the complexity it criticizes in established DLP systems.

It needs enough configuration to meet regulatory and operational requirements. At the same time, it must preserve the simpler experience that supports its “stress free” positioning.

That balance becomes harder as deployments cross regions, industries, and business units. Each expansion introduces new data types, policies, applications, and exception paths.

The winner in enterprise DLP will not necessarily offer the longest feature list. Buyers are likely to favor the platform that consistently turns accurate context into defensible action.

What the Funding Numbers Do Not Yet Prove

MIND has reported exceptional growth, but the available evidence does not yet establish how efficiently or reliably that growth scales.

The company says revenue increased more than seventeenfold and customer count grew eightfold during the past year. It also says its platform analyzed billions of data events.

Those figures suggest rising demand, especially when combined with an eight-figure revenue claim. Yet each number comes directly from MIND and lacks the supporting detail needed for a full assessment.

The revenue comparison has no disclosed baseline. The customer figure does not separate paid production deployments from smaller accounts or early-stage contracts.

Billions of analyzed events demonstrate processing volume, but event counts do not reveal detection quality. A large volume can include routine activity with little security significance.

Endpoint coverage offers another useful but incomplete signal. MIND says it has stopped sensitive data loss across hundreds of thousands of endpoints.

The company has not disclosed how many endpoints run the complete platform, how frequently controls intervene, or how customers measure prevented loss. Those questions matter when comparing products with different reporting methods.

Automation introduces its own measurement problem. An agent can close more investigations than a human team, but speed is valuable only when the decisions are accurate.

Security leaders should look for measurable changes in false positives, investigation time, policy maintenance, exception volume, and confirmed data loss. They should also examine user disruption and the frequency of reversed decisions.

Regulated industries need an especially clear audit trail. A healthcare provider may need to explain why a specific transfer was blocked, permitted, or escalated.

Financial institutions, government agencies, and legal teams face similar requirements. A generated explanation must connect to evidence, policy, and identity context rather than provide plausible text alone.

MIND’s customers supply encouraging examples. Children’s Hospital Los Angeles says it protects patient information entering AI tools without affecting endpoint performance.

The National Geographic Society says MIND provides visibility into interactions between its information and AI agents. These cases show specific uses rather than abstract demand.

They remain customer testimonials presented by the company. Independent case studies with documented before-and-after measurements would provide stronger validation.

Enterprise adoption will also depend on implementation effort. MIND argues that its agents reduce administrative work, but complex organizations rarely begin with clean data ownership and consistent policies.

Automation can accelerate decisions after a system understands the environment. It cannot automatically resolve every conflict between unclear ownership, overlapping rules, and legitimate exceptions.

Data residency adds further complexity. A global enterprise may need different storage, processing, and enforcement practices across jurisdictions.

Agent permissions create another risk. A security agent requires enough access to inspect information and perform approved remediation.

That access makes identity controls, logging, separation of duties, and emergency shutdown procedures essential. The agent itself becomes part of the organization’s sensitive security infrastructure.

MIND must demonstrate that customers can constrain its autonomy. A platform designed to reduce manual work still needs predictable escalation when confidence is low.

The company’s certification and investor support strengthen its position, but neither substitutes for operational evidence. The next phase will test whether early success translates into repeatable deployments across diverse environments.

The financing also raises expectations around disclosure. As MIND grows, customers and investors will want clearer indicators of retention, expansion, classification performance, and response accuracy.

Those measures will determine whether autonomous DLP becomes a defensible product advantage. They will also show whether MIND is reducing complexity or relocating it.

Three Signals Will Show Whether MIND Can Scale

The next test is not another funding milestone, but evidence that MIND can expand coverage while preserving accuracy, control, and customer trust.

The first signal is measurable customer expansion. MIND needs to show that early customers add endpoints, information sources, AI applications, or business units after initial deployment.

Expansion would indicate that the product survives real operating conditions. It would also support the claim that customers gain value beyond a limited proof of concept.

Renewal data would make that signal stronger. High retention and growing contracts would suggest that autonomous DLP becomes part of routine security operations.

Weak expansion would point in the opposite direction. It could indicate that integrations, policy management, or false positives become harder as deployments broaden.

The second signal is independent technical validation. MIND should provide benchmark results or customer measurements covering classification accuracy, false positives, investigation time, and remediation outcomes.

No single benchmark can reproduce every enterprise environment. Still, transparent methods would give buyers a better basis for comparison.

Validation should include AI-specific workflows rather than only familiar file transfers. Tests should cover prompts, browser sessions, copilots, API calls, and agents retrieving information across multiple systems.

They should also examine adversarial behavior. Users or attackers may rename files, fragment information, encode content, use screenshots, or move data through approved tools.

Autonomous responses need evaluation under the same conditions. The relevant question is whether the system takes an appropriate action when evidence is incomplete or contradictory.

Independent validation would strengthen MIND’s central claim. Poor or inconsistent results would weaken the argument that autonomous management improves on carefully supervised controls.

The third signal is competitive and platform response. Microsoft, Varonis, Cyera, Cyberhaven, and other vendors are unlikely to leave autonomous DLP uncontested.

They can add agents, expand generative AI coverage, or bundle controls with existing security platforms. Some already connect data discovery, access context, behavioral monitoring, and enforcement.

MIND must maintain a meaningful difference as those features spread. That difference could come from lower administrative work, broader coverage, faster deployment, or more accurate decisions.

A feature race alone would favor larger vendors. MIND needs evidence that its architecture changes the economics or reliability of operating a DLP program.

Channel partnerships will reveal part of that outcome. Strong partners can make MIND easier to purchase, deploy, and integrate within established enterprise security stacks.

Technology integrations provide another clue. Coverage across identity, endpoint, cloud, productivity, and AI systems will determine whether the platform sees enough context to enforce useful policies.

The MIND Series B arrives when cybersecurity investors are rewarding companies positioned around AI risk. Recent market behavior also shows that businesses expect AI adoption to increase security spending.

That environment can accelerate MIND’s growth. It can also make it harder to separate durable demand from a favorable financing cycle.

For enterprise buyers, the funding is meaningful because it extends the company’s runway and supports larger deployments. It should not shorten technical or governance reviews.

Teams evaluating MIND should test their own sensitive information, real user behavior, approved AI tools, and exception workflows. They should measure both prevented exposure and operational friction.

They should also define where human approval remains mandatory. Autonomous policy work deserves the same scrutiny as any system allowed to restrict employees or modify security controls.

MIND now has more capital to argue that DLP should operate at AI speed. Its opportunity is substantial because the old boundaries around enterprise data are weakening.

The deciding evidence will come from production environments, not financing totals. Can MIND reduce manual security work while making each automated decision observable, reversible, and defensible?

That is the standard the company’s $72 million round has set. It is also the question enterprise security teams should carry into every evaluation.