Agnidipta Sarkar, Chief Evangelist, ColorTokens is a cybersecurity and digital resilience chief with greater than three many years of expertise spanning cyber protection, threat administration, enterprise continuity, privateness, and Zero Belief. At ColorTokens, he works with boards, C-suite executives, and safety leaders to strengthen breach readiness and join cybersecurity applications with enterprise priorities, whereas additionally contributing to worldwide requirements and trade initiatives via organizations together with ISO, the Cloud Safety Alliance, NIST, and ISA. Earlier than becoming a member of ColorTokens, Sarkar served as Group CISO at Biocon and held senior data safety and threat management roles at organizations together with DXC Know-how, Hewlett Packard Enterprise, and HP.
ColorTokens is a cybersecurity firm centered on microsegmentation, Zero Belief, and serving to enterprises turn into extra resilient when attackers breach conventional perimeter defenses. Its flagship Xshield Enterprise Microsegmentation Platform is designed to forestall attackers and malware from transferring laterally throughout a company by creating granular safety boundaries round workloads and property spanning knowledge facilities, cloud infrastructure, endpoints, Kubernetes environments, operational expertise (OT), and Web of Issues (IoT) units. The platform additionally incorporates AI-assisted workflows for locating environments, creating segmentation insurance policies, and accelerating coverage deployment, with the broader aim of decreasing a company’s assault floor and limiting the potential affect, or “blast radius,” of a profitable breach.
You will have spent greater than three many years transferring from hands-on networking and safety roles at HCL, Wipro, HP, HPE, and DXC to serving as Group Chief Data Safety Officer at Biocon and contributing to worldwide safety requirements. How did these frontline experiences form your conviction that organizations should put together to comprise breaches reasonably than assume they will forestall each intrusion?
In my early days at HCL and Wipro, cybersecurity felt just like the Wild West, thrilling, chaotic, and filled with alternative. I dove in, experimenting, failing, studying, and succeeding at all the pieces expertise might throw at me: firewalls, IDS, MFA, encryption, audits, governance. However the actual schooling got here once I watched my first safety incident unfold. Out of the blue, all these finest practices crumbled beneath stress. That was my wake-up name.
At HP, I received a front-row seat to chaos, first as a fly on the wall, then as an operational chief, watching chaos unfold up shut. Each incident drove residence the identical lesson: IT by no means stands nonetheless, and no funding can assure security. After I lastly took the reins as Group CISO, I knew my job was to construct a playbook for dealing with breaches, not just for my safety operations, however for the entire group. One which was repeatable, predictable, and will maintain the enterprise working when the storm hit, constantly enhancing the safety operations over time.
On paper, the instruments seemed excellent. In actuality, IT Service Administration uncovered cracks you by no means noticed coming. Asset administration, patching, configuration, change, threat—all snarled with human error. That’s the place digital weak point hides, and when the alarms go off, it’s a nightmare to untangle. There isn’t a script for the conflict room when a breach hits. Chaos is the rule, not the exception. That’s the reason I’ve spent my profession constructing construction within the eye of the storm and serving to leaders reduce via the noise and handle threat, not get swept away by it.
Wanting again, I’m grateful for each lesson—watching from the sidelines, getting my arms soiled within the thick of assaults, and at last main from the entrance. That’s why I champion breach readiness as an artwork, not only a science. You’ll be able to map out assault patterns all day, however what issues is what occurs within the conflict room, when leaders have to carry the road. That readability solely comes from residing via it.
That’s the reason I constructed a breach readiness framework to anticipate, face up to, and evolve the skills to face the subsequent cyberattack. The framework helps organizations transfer with the storm, not in opposition to it. The ultimate aim is to comprehend Koun Ryusui, drifting like clouds, flowing like water, in order that, when the subsequent unprecedented assault hits, as an alternative of chaos, enterprises train structured flexibility and resilience, leveraging their expertise investments to emerge stronger and extra assured.
AI helps attackers automate reconnaissance, vulnerability exploitation, and lateral motion, probably compressing actions that when took weeks into hours or minutes. Which components of the standard incident-response mannequin turn into ineffective when assaults start working at machine velocity?
When assaults transfer at machine velocity, the defenders who survive are those who can analyze and act quicker than the adversary. Sure, AI does give attackers new instruments, nevertheless it empowers defenders too. In my expertise, there are two methods to deal with these unprecedented, AI-powered assaults.
First, it’s essential to redesign your digital panorama. Construct breach-ready zones and microsegments that separate your crown jewels from the remaining. Consider it like a maze: some paths are open, others are blocked, making it powerful for unauthorized identities to maneuver freely. That’s how you retain AI-powered assaults at bay by slowing them down, as a result of the maze retains altering with modifications in digital techniques.
Second, you want templates, playbooks, and clear roles prepared earlier than the alarm ever sounds. In order that when that does, each human and machine is aware of what must be executed and in what precedence. The trick is to purchase time and sluggish the AI down, drive it to work tougher, and deny it simple motion. That’s how you retain your defenses within the struggle when each millisecond counts.
Ask any sailor who has survived a submarine breach. When water comes speeding in, you wouldn’t have time to guess its subsequent transfer. It is advisable have bolstered the best doorways forward of time, realizing what it’s essential to maintain operational and the place the stress will hit. When the breach occurs, all you are able to do is quarantine the flooded compartment and maintain the remainder of the vessel working. That’s the way you shield your minimal viable enterprise. Within the digital world, which means your minimal viable digital enterprise.
That’s what breach readiness is all about. Ought to an assault ever get via, you quarantine the affected space, set off your BCP, and maintain the unaffected core enterprise working. No shutdown, no disaster. Simply an incident emergency.
You advocate discussing breach readiness in enterprise and monetary phrases reasonably than treating it solely as a technical difficulty. Which metrics ought to boards use to find out whether or not the group might proceed working via a critical cyberattack?
Breaches don’t simply hit your techniques; they hit all the pieces and everybody that is dependent upon them. Ask the affected person turned away from a hospital, the traveler stranded at an airport, or the automobile supplier left in debt for months. That’s the actual price of a cyberattack. It isn’t if, however when. That’s the reason boards should take a look at breach readiness as a enterprise and monetary crucial.
There are two metrics each board ought to observe.
First, set up the quantum of fabric affect the board is prepared to simply accept in pursuit of digital and AI ambitions. That units the bar for the way a lot of your digital enterprise should keep operational throughout a breach. I name these the Most Acceptable Materials Impression (MAMI) and the Minimal Viable Digital Enterprise (MVDE).
Many confuse MVDE with enterprise continuity, however they don’t seem to be the identical. When you say lower than 10% affect is suitable, then 90% of your small business should maintain working even throughout a breach. Most enterprise continuity plans solely get you again to 30% at finest, leaving 60% of your digital enterprise uncovered. That paradox retains leaders up at night time.
When you set your MVDE at 70%, you possibly can guarantee stakeholders that you’re investing in defenses that maintain the enterprise working, even when the worst occurs, constructing a maze that’s powerful for human or AI attackers to breach, and the muscle to quarantine assaults at machine velocity. That is the place microsegmentation performs a foundational lego brick. Now you can plug most of your cybersecurity investments like EDR, Firewalls, SASE, Id (human and non-human), Entry, Authorization, and OT cybersecurity into one seamlessly related sign airplane that your SOC can function throughout energetic breaches to comprise cyberattacks and isolate the MVDE.
Observe the MAMI and MVDE each quarter, for each new digital and AI initiative, and your buyers and stakeholders will solely search proof of how effectively the resilience is being managed.
“Blast radius” is more and more used as a measure of cyber resilience. How can a company calculate its potential blast radius earlier than an assault happens, and what would represent an appropriate stage of publicity?
Blast radius issues. Considerably.
However it isn’t the one metric that counts. I at all times advocate assessing your present blast radius as step one to constructing a zero-trust, breach-ready enterprise. All you want is a Breach Readiness Impression Evaluation (BRIA), a non-intrusive, API-driven dipstick evaluation leveraging your present EDR. For OT and mainframes, you would possibly want brokers or home equipment.
Not all blast radius is dangerous. Generally, it’s essential for utility efficiency. The satan is within the particulars. Take a cache service that hastens lookups for product information and person periods. It wants system-to-system connectivity to run an e-commerce platform with a number of microservices. However that very same blast radius turns into an exploit if unauthorized customers or techniques can entry it.
You can’t take a look at blast radius in isolation. Mix it with different safety patterns like conduct anomalies, entry misuse, privilege creep, authorization overruns, and authentication failures to actually perceive your breach publicity. And bear in mind, what is suitable in a single trade is likely to be a deal-breaker in one other.
How does breach readiness differ from conventional incident response, catastrophe restoration, enterprise continuity, and zero-trust applications, and the place ought to duty for coordinating these disciplines reside?
These should not separate subjects; they’re intertwined disciplines that should be coordinated round a single enterprise end result.
I outline breach readiness as an enterprise resilience self-discipline: constantly getting ready, architecting, exercising, and governing the group to anticipate assaults, constrain blast radius, preserve the Minimal Viable Digital Enterprise throughout the group’s pre-defined Most Acceptable Materials Impression, and restore capabilities whereas the assault is being contained. Breach readiness is the weaponization of zero belief structure and lays out what should occur earlier than, throughout, and after a cyberattack. Breach readiness shifts the query from ‘How do we keep attackers out?’ to ‘How do we make sure an incident cannot cause more damage than we are prepared to tolerate, and how much of the business can keep running while we respond?’ Breach readiness is about survivability whereas the incident continues to be taking place.
The conceptual leap is important. Disparate enterprise practices of backup and restoration, incident response, catastrophe restoration, and enterprise continuity all play necessary roles in being breach prepared, as a result of they handle the disruption. They start when techniques are attacked. All three applications of incident response, catastrophe restoration, and enterprise continuity focus upon “How quickly can we restore what has been disrupted?”. Breach Readiness is targeted on “How much of it can remain available in the first place?”.
Breach readiness breaks down silos to make sure the enterprise penalties of a compromise are managed, and stakeholders are assured by specializing in how you can maintain a lot of the enterprise “unaffected” and invoke a BC/DR for the half that’s affected. Accountability and possession for being breach-ready lie with government management. The CEO/Board owns the chance urge for food and acceptable enterprise end result. The COO or an equal enterprise resilience government ought to coordinate breach readiness throughout the group, with the CISO proudly owning the cybersecurity dimension.
For enterprises that intend to realize a state of Kuon Ryushi, full organizational participation makes use of expertise and AI to handle the results of a cyberattack is essential.
ColorTokens is utilizing AI to assist safety groups analyze environments and generate microsegmentation insurance policies extra rapidly. The place ought to organizations belief AI to automate defensive choices, and which containment actions ought to proceed to require human approval?
Completely. We’re doing precisely that.
However let’s be clear: ‘using AI to help security teams’ is simply the tip of the iceberg. At ColorTokens, we consider in utilizing AI responsibly to make enterprises actually breach-ready. There’s a a lot greater story than simply ‘AI makes microsegmentation easier.’
Latest incidents with Anthropic, OpenAI, AISI, and even the health club reserving case in Australia are failures of structure, not governance. When autonomous AI escapes its sandbox, it isn’t at all times malicious or unreliable. It’s about how the structure was constructed. People should clearly outline the working envelope, and AI should keep inside it. That’s the place governance is available in.
For instance, you would possibly permit AI to isolate any endpoint flagged as compromised with confidence above a sure threshold, until it’s an OT security system, area controller, fee system, or production-control asset. That’s bounded autonomy. However it’s simpler stated than executed since you want the structure to be context-specific and clearly laid out utilizing material consultants.
For my part and expertise, corporations should use AI to carry out discovery, correlation, dependency mapping, attack-path evaluation, blast-radius evaluation, coverage suggestions, coverage synthesis, coverage simulation, low-risk coverage optimization, pre-authorized containment, and steady verification and depart the dedication of enterprise criticality, acceptable disruption, security boundaries, crown-jewel definitions, security constraints (in OT), most acceptable affect, autonomy thresholds, exception approvals, main manufacturing isolation, and irreversible actions to people.
Backside line – AI needs to be allowed sufficient freedom to finish its process, however beneath specific situations enforced by infrastructure outdoors its sphere of management. AI ought to suggest actions, should await people to approve, after which implement modifications at machine velocity. The target isn’t to make AI autonomous. It’s to make defensive autonomy bounded.
At ColorTokens, we use AI to make the defender’s management loop quick sufficient to maintain up with attackers. Centralized coverage decisioning, a number of enforcement mechanisms, wealthy telemetry, agentless safety for techniques that may’t run brokers, cloud and container assist, and AI-assisted discovery and coverage synthesis—all are a part of AI-assisted breach readiness.
Microsegmentation has existed as an idea for years, however organizations regularly affiliate it with complicated deployments and inflexible insurance policies. What has modified technologically that makes it extra sensible throughout cloud infrastructure, Kubernetes environments, endpoints, legacy techniques, and operational expertise?
A number of issues.
What began as a flowery technique to join and management networks, guaranteeing devoted connectivity and bandwidth, has advanced right into a foundational cybersecurity functionality. Immediately, microsegmentation can rework massive enterprises into breach-ready organizations, rapidly and confidently. Know-how shifts have helped us transfer previous the outdated complications: mapping dependencies, redesigning community boundaries, configuring VLANs and firewalls, writing guidelines by hand, and worrying about breaking enterprise site visitors. Segmentation was sluggish, costly, and restricted to the info middle. Not anymore.
First, id is now the principle conduit for cyberattacks. That’s the reason enhanced id governance is the primary precept of zero belief. In trendy cloud environments, IP addresses are meaningless for safety. Fashionable microsegmentation makes use of id, entry, workload, utility, service, and context, not simply community location. The safety boundary follows id, entry, and the applying, not the underlying infrastructure alone.
Second, agentless microsegmentation now goes past home equipment. It extends safety to legacy techniques, IoT units, and OT that can’t run brokers. Fashionable microsegmentation integrates with EDR, increasing protection and slashing deployment time from months to hours, as a result of EDR already has the telemetry microsegmentation wants.
Lastly, AI is eradicating the final huge bottleneck: context-specific breach-ready coverage engineering. AI can analyze a number of knowledge factors for context, map dependencies and assault paths, advocate zoning and microsegmentation insurance policies, and simulate their affect earlier than enforcement. Fashionable microsegmentation enforces coverage instantly at endpoints and workloads utilizing native OS controls. You now not want to revamp the community; microsegmentation is now seamless, and coverage follows the workload wherever it goes.
The aim is not only to dam each lateral motion try. However to research sufficient indicators at machine velocity to make sure that when an attacker will get in, the compromised workload doesn’t turn into a freeway to all the pieces else. Fashionable microsegmentation is now not nearly dividing networks however about controlling and monitoring the blast radius.
What proof ought to executives demand when evaluating the monetary case for microsegmentation, notably concerning ransomware downtime, cyber-insurance premiums, regulatory publicity, and the price of protecting essential providers operational throughout an assault?
Monetary dedication of breach affect and correlating a microsegmentation funding that may guarantee unaffected operations throughout a breach requires executives to first notice that cyberattacks will succeed, no matter the funding in cybersecurity. In 2025 and 2026, organizations that succumbed to cyberattacks and confronted unexpected downtime embody many financially robust organizations like JLR, Nike, and Stryker. Subsequently, it isn’t a matter of if they’d be breached, however when.
When you settle for this actuality, investing in foundational breach readiness turns into pressing and a matter of enterprise survival. To speculate properly, allow us to perceive the 2 board-level indicators in additional element.
The primary is a measure of survivability. The Most Acceptable Materials Impression (MAMI) that the highest administration and governing physique are prepared to simply accept for digital transformation or AI adoption ambitions. It’d begin as a income quantity, however should increase to monetary viability, buyer belief, status, or model over iterations.
The second is an indicator of aggressive edge. The Minimal Viable Digital Enterprise (MVDE) that should stay operational even when probably the most unprecedented cyberattacks happen. All enterprise instances for brand spanking new initiatives should contemplate each components. The MVDE can start from 70% and maintain growing based mostly upon the success of the underlying id and microsegmentation program.
Here’s what executives ought to search for when evaluating microsegmentation: proof that it will possibly tackle ransomware downtime, decrease cyber-insurance premiums, scale back regulatory publicity, and maintain essential providers working throughout an assault.
- Does the microsegmentation platform permit a single platform throughout datacenters, customers, OT techniques, and cloud?
- What share of our enterprise is structured into zones and microsegments based mostly upon materials affect to enterprise versus the earlier overview?
- How swiftly can the microsegmentation resolution be operational with zones, microsegments, and managed conduits, designed to maintain the MVDE operational?
- Can the microsegmentation resolution combine with EDR and use it as an agent to cut back the agent footprint?
- Does the microsegmentation platform use any conversational synthetic intelligence to find out context-specific guidelines and insurance policies?
- Do we’ve the flexibility to watch modifications within the conduct of legitimate customers in functions and limit errant customers by id?
- Can the imply time to detect and reply by quarantining malicious conduct scale with new digital and AI adoption in enterprise?
- Can the microsegmentation platform simulate and mannequin cyber protection eventualities to progressively scale back breach publicity of operations?
Whereas these would tackle most use instances, compliance will want particular person statements of applicability for every regulation, which may clarify what a part of the regulation could be met by the ColorTokens platform instantly, what wants course of enveloping the expertise, and what wants extra expertise and processes.
Many organizations conduct penetration assessments and incident-response workouts, but these might not reveal how far an attacker might transfer after gaining entry. How ought to corporations check containment and survivability beneath sensible AI-accelerated assault situations?
If organizations have to embrace the essence of Koun Ryusui in constructing breach readiness, they need to combine microsegmentation into present cybersecurity operations and evolve it over time. To do this, microsegmentation wants sign integration with different instruments within the Safety Operations Heart, just like the SIEM, the EDR, the Subsequent Gen Firewalls, and many others. This could make sure that AI-powered microsegmentation can orchestrate containment when wanted, based mostly on indicators of assault and behavioral anomalies.
What you want then should not particular person assessments, however workouts that assist evolve your breach readiness.
There are three issues to train in case your microsegmentation and survivability are sensible. First is the present state of Breach Readiness. This may be carried out earlier than you start your ColorTokens deployment, after you will have enforced your insurance policies, and upon each change to find out whether or not your change has altered the design of your MAZE, which accommodates material-impact-based breach-ready zones, microsegments, and managed conduits.
The second is the velocity at which you’ll quarantine cyberattacks when an precise breach occurs. The important thing parameters to check are detection accuracy and the velocity of MVDE containment and isolation utilizing AI. And that may be executed via penetration assessments, red-teaming workouts, and breach assault simulation assessments. These assessments needs to be executed with out notification to the Safety Operations Heart to verify their response to altering breach parameters.
The third is how your group would react in an precise breach situation. Earlier than exercising, guarantee your “be breach ready playbooks” are communicated to all related inside stakeholders, technical assist third events, and upkeep suppliers, particularly asset house owners and OT techniques integrators. Then conduct simulated workouts earlier than conducting a red-teaming evaluation to supply a sensible expertise.
As enterprises deploy extra autonomous AI brokers with entry to functions, knowledge, credentials, and infrastructure, how will breach readiness have to evolve to comprise not solely compromised human accounts and units, but in addition compromised or misbehaving AI brokers?
Breach readiness should evolve from containing compromised human accounts and units to treating autonomous AI brokers as a definite, high-velocity class of non-human id (NHI) that may act with legit credentials at machine velocity. The problem sounds important on paper as a result of NHI are anticipated to outnumber human identities at the very least 250 instances by present estimates.
Brokers break conventional controls centered on customers, endpoints, and static service accounts. They maintain stay entry to functions, knowledge, credentials, and infrastructure; purpose and chain actions constantly; and could be compromised by way of immediate injection, instrument poisoning, reminiscence manipulation, supply-chain points, or easy misalignment/drift. And after they misbehave or are hijacked, the blast radius expands quicker than human-paced response can match. Breach response can not await full certainty or multi-team escalation, and therefore hardening and automatic pre-approved fundamental responses will likely be obligatory.
As extra autonomous AI brokers get deployed, breach readiness wants to make use of Id as the first management airplane for brokers, improve Zero Belief to “Agent Trust”, govern guardrails, reminiscence safety, and governance hygiene, graduate to pre-approved friction and automatic containment, and combine Microsegmentation and architectural isolation for containment, together with Runtime visibility, behavioral baselining, and sequence detection.
Breach readiness must shift from “can we recover after encryption?” to “can we contain an autonomous actor operating with valid credentials before it cascades?” Indicators of breach readiness ought to embody time-to-understand (which agent, what actions, what knowledge/techniques had been touched, and whether or not exercise is ongoing) and time-to-friction for brokers particularly. The ColorTokens platform can quickly uncover dependencies, generate and refine coverage, and implement containment, particularly when augmented by AI for coverage automation and turns into central to protecting tempo. On this mannequin, we prioritize isolation, containment, and least privilege over excellent prevention.
ColorTokens’ microsegmentation platform extends id, steady verification, microsegmentation, behavioral monitoring, and pre-authorized containment to the agentic layer, which can maintain the blast radius manageable and protect operational continuity. The defensive signature required is similar one demanded by machine-speed ransomware: excessive velocity of understanding and friction, grounded in assume-breach structure reasonably than hope that brokers will at all times keep aligned.
Thanks for the nice interview, readers who want to study extra ought to go to ColorTokens.

