Beyond “Close Enough”: How Agentic Consulting OS Is Redefining Risk Assessment for CISOs
Consulting has a dirty little secret that most CISOs know but few say out loud.
You can pay top dollar to a global firm, sit through weeks of interviews and workshops, and still feel like you are explaining your job to people who are just getting started in theirs.
That frustration is the origin story behind Milestone, an “agentic consulting operating system” that is quietly reshaping how risk assessments and assurance work get done.
This is not another generic AI pitch. It is a rethink of how expert judgment, enterprise context, and automation come together to support CISOs who live in high-stakes environments where “close enough” is never good enough.
The conversation that follows is based on an in-depth briefing with Daniel Finchelstein, CTO of Milestone, a company still in stealth but already working with banks, insurers, and energy firms.
The problem every CISO already recognizes
When asked what problem Milestone was created to solve, Daniel did not start with AI. He started with the reality of how enterprises actually buy and consume professional services.
He explained the first fork in the road. Large enterprises that need security, audit, or risk consulting normally see two options: go to the big firms or try to assemble a patchwork of boutique specialists.
In Daniel’s words:
“Whenever enterprises would like to engage in professional services, they have two options, go to the big ones if they can afford it, and then have usually quick process of engaging with the companies like the Deloittes, the PwCs of the world, the KPMGs, etc.”
The big brands provide a smoother procurement path, but the people who show up are often not the experts the enterprise thought it was paying for.
As he put it:
“Usually, when you go with the big ones, you don’t get the best, I would say, expertise, because usually you speak with a partner, they would send their entry level positions who will start to ask questions, gather information. Usually they don’t know exactly what they do, because they’re beginners in their role. And no one is happy, because they need to teach the guys that they paid for and premium money.”
If you try to avoid that dynamic and work with specialist boutiques, you trade one problem for another.
“We can go to boutique companies, but then you need to manage all of those boutique companies, like managing the herd, like herding the companies, and then do the procurement.”
So enterprises choose the lesser evil. They pick what is “easier to engage with” rather than what is best aligned to their risks.
For CISOs, that means:
- More time spent educating vendors
- More cycles spent on coordination than on remediation
- More risk that reports arrive late, shallow, or already outdated
Milestone was built to attack that frustration at its root.
Flipping the usual AI narrative on its head
Most AI stories in security consulting start with “augmenting people with AI.” Milestone deliberately inverts that idea.
Daniel summarized their philosophy this way:
“Instead of augmenting people with AI, we augment AI with people, so our brain is AI, and the people are actually helping whenever is needed, but not much more than that, because everyone needs to do what they do best, right?”
The “brain” is an agentic consulting operating system.
The experts are the last-mile specialists who:
- Talk to customers
- Read the room
- Assess politics and culture
- Validate, interpret, and refine what the system finds
The engine does the heavy lifting in the background. Scoping. Collecting. Correlating. Analyzing. Prioritizing.
“We created an agentic consulting operating system. So that means that we have built an engine that actually is able to create engagements for our customers, and then that is a very powerful brain, because that brain is able to scope, to analyze, and then collect information – and then analyze and then provide a live report, not just a PDF document.”
That last point is crucial. Traditional consulting engagements culminate in a static PDF that arrives weeks or months after the work started. By then, half of it is already obsolete.
“When the customer receives a PDF document, it’s like receiving a grade for an exam that they did two months ago. Usually, by the time the report is being handled, it’s not that accurate and up to date, right? Because customer may have had many sleepless nights repairing some of the things that they already know needs to be repaired.”
Milestone’s response is to make the “report” a living asset, backed by a system that remembers, connects, and updates continuously.
Why “close enough” cannot be the standard
Many CISOs today lean on general-purpose LLMs in their own workflows. They speed up documentation, policy drafting, and light analysis.
But Daniel was blunt about where that approach breaks in critical risk and assurance work.
“We always say that when you use a general LLM, you’re getting ‘close enough’. You know, usually you get very close enough results, but close enough is not always acceptable. So close enough is only good when the stakes are low.”
For security leaders facing regulatory, reputational, and operational exposure, “close enough” is never an acceptable standard.
Milestone defines five areas where it deliberately diverges from generic tooling like ChatGPT:
- Trust
- Quality
- Process value
- Platform
- Explainability
On trust, the core differentiator is proprietary domain data and explicit controls around hallucination.
“In many environments, in high stake environments, you need something which is trustworthy. Our system has the propriety domain data which allows us to build on those capabilities and provide highly accurate information.”
On quality, the focus is regulatory alignment and organizational context, not generic fluency.
“When you use General LLMs, they’re very fluent, but they’re inconsistent. And in our system, which is very hyper tailored for the organization and to regulatory standards, you get something which is much, much more to the quality that our customers need.”
For a CISO, this is the difference between a clever draft and an artifact that can stand up to an auditor, a board, or a regulator.
From “smart documents” to a shared operating picture
Most consulting outputs are documents that live in email threads and file shares. Milestone treats the engagement itself as a set of applications running on top of its operating system.
“For every engagement we have – just like on top of the operating system – we have applications. So what’s an application? An application is the first and foremost analysis tool, or analyzer, analysis application that runs on top of that operating system.”
That is where process value and platform come in.
Instead of each consultant improvising their own approach, the system provides a consistent, traceable flow for how observations are discovered, promoted, and translated into work. Multiple experts can collaborate without losing coherence.
“It provides the value of the process. For example, for multiple experts to collaborate and to do things that hand off work to the system, knowing that everyone would have the same process, so they don’t need to invent something.”
The platform view matters just as much.
Within Milestone, both experts and clients see where each engagement stands:
- How far along is the assessment
- Where delays or blockers exist
- Where emerging opportunities or risks are clustering
“We provide a platform in which the experts or the people who are working with the client and even the client themselves, able to get into the platform and get the quick visualization of where each product stands as a project, where each project stands, and where it actually progresses to, ‘Are we far behind?’, ‘Are we making it ahead of the schedule?’, ‘Are there more opportunities?’, etc.”
For CISOs, that moves the relationship from “we will get you a report” to “here is an evolving, explainable risk picture you can interrogate.”
Explainability as a first-class feature
Explainability is not a buzzword in Milestone’s design. It is the backbone of how the system works.
Daniel framed it from the perspective of an IT auditor or risk committee member.
“One of the things that they need is the ability to manage all of their work items, or all of their, I would say, lists of to do and to monitor, etc, in a way that it’s manageable, explainable, and then to be able to work on it with a lot more intelligence of how to explain it to everyone in the organization.”
Every step in the system is traceable. How a conclusion was reached. Which evidence fed into it. Where in the engagement lifecycle an item currently sits.
“We provide that, because everything that we do has the tracing back to how we got to that, to get to that point.”
That transparency matters for:
- Internal alignment between security, IT, and business stakeholders
- Briefing executive management and the board
- Surviving external scrutiny from regulators, auditors, and customers
Under the hood: atomic observations to live reports
The power of the system becomes clearer when you look at how it handles something as mundane as an interview.
In the demonstration, Daniel showed an assessment for “Acme Bank.” Milestone maintains a “DNA” profile for each customer that captures attributes like public vs private, size, risk appetite, and sector. That DNA informs how the engine interprets what it sees.
On the left side of the screen sat a working area with documents and interview files.
“Every document is actually coming either from the customer or from a discussion that we have with the customer, including an interview. So we can see the interview file over here.”
By the time the expert gets back to their car after the meeting, the interview is already ingested and structured.
“By the time the expert created the interview file with the client, and by the time they get to their car. This interview is actually transcribed and highlighted.”
The system surfaces concrete findings as what Milestone calls “atomic observations.”
“Instead of anyone needing to read it, the OS the operating system actually goes and, you know, finds the things, for example, database credentials, API keys and service passwords which are hardcoded, so a new atomic observation is created.”
From there, experts can promote these atomic observations into official observations. That is where narrative, impact, and risk analysis are layered in.
“Here you can see, for example, this atomic observation that became an official observation, so it has the name or the headline, and then something with the more explain, explanatory of what was found, and with all of the risk analysis and prioritization, etc, there is the observation, the implication and risk, including risk amplifier, risk reducers, and after the recommendation of what needs to be done.”
Evidence is not an afterthought. The system keeps track of where each item came from, what is missing, and who needs to be engaged next.
“Eventually there may be missing evidences. So the system needs to collect the evidence, either by sending an email to someone or by sending the expert to collect the information up to the sources where it was found.”
The result is a live report that can show:
- Current security posture
- Industry benchmarks
- Recommended target states based on the organization’s DNA
“We are already serving customers with this agentic operating system, and we are already seeing requests for partnerships from, you know, consulting houses which say, Hey, give it to us!”
Where this is heading next
Milestone is already active in production beta with customers in banking, insurance, and energy. Today, reports often still take the familiar form of PDFs, but the underlying system is designed for interactive, queryable outputs.
“Today, the report would be still a PDF, but in the future, it will be a live report on the system that anyone can query and speak with and edit.”
On the roadmap are features many CISOs will recognize from their wish lists:
- Risk maps and heat maps
- Full risk registers
- IT stack inventory views tied directly to observations
- Integrations with systems of record like JIRA and Monday
“This is part of the requests that we received, for example, to show the information in several levels, because it’s not a PDF. So for example, show it as a risk map, as a heat map, but also as risk register and inventory like IT stack inventory management so that we can show everything that we serve, etc. I’m super excited by the long road map.”
A practical call to action for CISOs
The most important takeaway from Milestone’s approach is not that you need this specific platform tomorrow.
It is that you can, and should, demand more from the way consulting and risk assessment work is done for your organization.
As a CISO, you can use this lens to evaluate your current partners and internal processes:
- Are you still paying to educate junior staff from external firms on your environment and risks?
- Are your assessments still delivered as static PDFs that age quickly and resist interrogation?
- Can you trace how each conclusion was reached and what evidence supports it?
- Do your tools and partners begin from your real conversations and context, or from generic surveys and templates?
- Is “close enough” silently becoming the standard where it should never be?
Whether you explore an agentic operating system like Milestone or push your current partners toward similar outcomes, the direction of travel is clear.
Consulting that works for CISOs in high-stakes environments will:
- Treat AI as the primary engine and experts as high-value interpreters
- Deliver explainable, live risk pictures rather than frozen reports
- Start from the language and concerns of your stakeholders, not from questionnaires
- Compress labor where it adds the least value and invest it where it matters most
The next time a major engagement is on the table, do not just ask what the deliverables are. Ask what the engine behind them really looks like.
If your consultants cannot show you how they move from raw conversations to atomic observations to explainable recommendations, you are not getting the value you deserve.
And if a small team armed with the right operating system can give you more clarity, faster, with less friction, it might be time to make that your new milestone.
Visit them online at Milestone – Cybersecurity Services
About the Author
Pete Green is the CISO of Anvil Works, a ProCloud SaaS company and co-author of “The vCISO Playbook: How Virtual CISOs Deliver Enterprise-Grade Cybersecurity to Small and Medium Businesses (SMBs)”. With over 25 years of experience in information technology and cybersecurity, Pete is a seasoned and accomplished security practitioner.
Throughout his career, he has held a wide range of technical and leadership roles, including LAN/WLAN Engineer, Threat Analyst, Security Project Manager, Security Architect, Cloud Security Architect, Principal Security Consultant, Director of IT, CTO, CEO, Virtual CISO, and CISO.
Pete has supported clients across numerous industries, including federal, state, and local government, as well as financial services, healthcare, food services, manufacturing, technology, transportation, and hospitality.
He holds a Master of Computer Information Systems in Information Security from Boston University, which is recognized as a National Center of Academic Excellence in Information Assurance / Cyber Defense (CAE IA/CD) by the NSA and DHS. He also holds a Master of Business Administration in Informatics.
