SOC Personnel & Workforce: When Tier 1 Is Automated, Where Do New Analysts Start?
AI is absorbing the Tier 1 queue, which is where most analysts learned the job. Here is where new analysts can start now, and what a SOC has to build to grow its own seniors.
Where the first rung went
The last post in this series ended on a question about growth: if AI takes the Tier 1 queue, where does a new analyst learn the job? For about twenty years the answer was simple. You got hired into the queue, you closed thousands of alerts, and somewhere around alert ten thousand you started to know what normal looked like. Most of the senior analysts working today came up that way.
The big picture: The queue is being automated first, and the industry is now finding out how much training was buried inside it. In ISC2's May 2026 survey of practitioners who use AI at work, most said the need for entry-level roles has already shrunk. In the same survey, about half said AI is also creating new kinds of entry-level work. Both answers can be right. The old first job is going away faster than anyone is designing the new one.
Why it matters: Every senior analyst you will want to hire in 2030 is somebody's junior analyst today. If the first rung disappears without a replacement, the shortage of experienced people we described in Talent Shortage vs. Skills Gap gets worse on a delay, and nobody's dashboard shows it until the senior requisition sits open for a year.
By the numbers:
- 56% of the 856 practitioners in ISC2's Rethinking AI's Impact on Cybersecurity Roles survey said AI has reduced the need for entry-level positions over the past year. 53% said it is creating new types of entry-level roles.
- 32% of organizations reporting AI-driven role changes cut SOC and security analyst positions, the largest reduction of any role, in the 2026 SANS | GIAC workforce research. Only 4% said entry-level roles are hard to fill.
- 9% to 32% is how far unclear career paths rose as a hiring obstacle in a single year in the same SANS research.
- 65% versus 5.9% is the growth in senior-titled versus junior-titled cybersecurity postings across G7 countries from October 2025 to March 2026, per the Cisco-led AI Workforce Consortium.
- 49% of security leaders in the consortium's survey said hands-on experience with AI agents is the hardest competency to find in entry-level candidates.
- 16% is the relative employment decline for workers aged 22 to 25 in the most AI-exposed occupations by October 2025, according to Stanford's Canaries in the Coal Mine research. Experienced workers in the same occupations held steady.
What Tier 1 actually taught
Nobody designed the Tier 1 queue as a training program. It was a staffing answer to alert volume. It trained people anyway, because the work forced a junior analyst to do a few things thousands of times.
A library of normal. Closing benign alerts is how an analyst learns that the finance team's month-end batch jobs look strange and are fine, or that one service account touches the HR database every night. That library is the institutional knowledge we covered in When Outsourcing Becomes a Liability, and the queue is where most people built theirs.
Escalation judgment. Every alert ended in a small decision: close it, dig further, or hand it up. A junior analyst got feedback on that call from the senior analyst who caught the escalation, usually within the same shift.
Writing it down. Case notes are how a manager checks reasoning. ISC2's 2025 hiring research found documentation and alert and event management among the core tasks hiring managers expect from entry-level staff.
Tool and data fluency. A year in the queue meant a year of querying the SIEM, pivoting in the EDR console and reading sign-in logs, which is the same fluency a detection engineer needs later.
None of that requires the alert volume itself. The volume was just how the reps got delivered, and once the volume goes to a machine, the reps have to come from somewhere else.
Aviation ran into the same problem earlier. In 2013 the FAA issued SAFO 13002, a safety alert telling airlines that continuous use of autoflight could erode the manual flying skills pilots need when automation fails, and that flight data showed more manual handling errors. Its fix was to schedule manual flying on purpose, in training and on the line, and the FAA followed up in 2017 with SAFO 17007 on maintaining that proficiency. Nobody suggested removing the autopilot. The airlines had to put back, deliberately, the practice that automation had taken away by accident.
What the data says is happening to entry-level work
Most of the evidence points the same way, with some caveats worth keeping.
Inside security, practitioners report the change directly. In the ISC2 survey, 56% said AI reduced the need for entry-level roles over the past year, 19% saw little or no impact, 12% said the need increased, and 13% weren't sure. SANS found SOC and security analyst roles at the top of the reductions list, followed by threat intelligence analysts at 26% and incident responders at 22%. Those are the three jobs where people have traditionally learned the craft.
Posting data shows hiring tilting toward experience. Senior-titled roles grew more than ten times faster than junior-titled roles in the AI Workforce Consortium's G7 data. CyberSN, which we cited in the first post in this series, put only 9.5% of 2026 US postings at entry level.
Outside security, the Stanford team found the same age pattern across AI-exposed occupations using ADP payroll data. Their February 2026 follow-up found the decline for young workers became visible in 2024, grew to about 16% by October 2025, and hadn't reversed. They tested whether interest rates explained it and found little support for that, though they note AI isn't always the only driver.
Two counterweights belong next to those numbers. The Bureau of Labor Statistics still projects 21% growth for information security analysts from 2025 to 2035, about 14,100 openings a year. And BLS notes that many analysts arrive from IT, often after working as network or systems administrators. The front door was never the only way in.
Where new analysts can start now
If the queue isn't the first job, something else has to be. Eric Hulse, our VP of Engineering, described where displaced middle-of-the-SOC work goes in The Middle of the Shift Map Is Becoming Something You Call: analysts move left into detection engineering, right into response, or into building and supervising the agents that now do first-pass triage. Those same three directions work as starting seats, as long as each one comes with the reps the queue used to provide.
This table is our own working framework for entry-level seats in a SOC where AI handles first-pass triage.
Starting seat | What a new analyst does | What it teaches | How it goes wrong |
|---|---|---|---|
Investigation reviewer | Reads AI-completed investigations, reruns the evidence on a sample, agrees or overrides, and writes down why | Escalation judgment and the library of normal, built from finished cases instead of raw alerts | Becomes rubber-stamping when nobody measures overrides or checks the reviewer's calls |
Detection support | Tests proposed rules against real data, tracks which rules produce benign closures, drafts tuning changes for review | How evidence maps to data sources and fields, and what noise costs downstream | Turns into ticket admin with no ownership of a single rule |
Response scoping | Works blast radius on confirmed cases: other accounts, other hosts, other sessions | How an investigation becomes a remediation plan, under a senior responder | Junior takes containment actions without a second set of eyes |
Agent operations | Maintains the investigation steps and context that agents rely on, flags where agent output and evidence disagree | How the automation reasons, where it fails, and what context it lacks | Becomes prompt-tweaking with no grounding in investigation method |
IT and identity side door | Hires from help desk, identity administration or systems administration into any seat above | Environment knowledge on day one, the thing hardest to teach | Gets skipped because the requisition asks for SOC experience the candidate doesn't have |
Two notes on that table. The reviewer seat is the most natural replacement for Tier 1, and it's also the easiest one to waste, because reviewing machine output can look identical whether someone is learning or just clicking approve. And the side door may be the cheapest fix available. ISC2's 2025 hiring research found that 90% of hiring managers would consider a candidate with only prior IT work experience, and a majority said training an entry-level hire to work independently takes four to nine months, according to Infosecurity Magazine.
A career path that still produces seniors
A starting seat is not a career path. SANS found unclear progression jumped from 9% to 32% as a hiring obstacle in one year, and it was already the third-ranked reason people leave. The path has to be written down, and it has to say what proof gets someone to the next stage. Throughput can't be that proof anymore, because the machine now does the throughput.
This is how we'd sketch the first three years. The stages and time ranges are our interpretation, and a smaller team will compress them.
Stage | Typical timing | Core work | Ready to move on when |
|---|---|---|---|
Foundation | 0 to 6 months | Investigation review on a sample, paired with a senior; one rotation through detection support | Their overrides of AI verdicts hold up on senior review, and their case notes explain the evidence |
Ownership | 6 to 18 months | Owns a small set of detections end to end; leads scoping on lower-severity confirmed cases | A rule they own gets measurably quieter without losing true positives |
Specialization | 18 to 36 months | Chooses a lane: hunting, detection engineering, response or agent operations | They can encode an investigation method someone else, or an agent, can run and audit |
Four practices make that path work.
Put manual reps back on the schedule. The aviation lesson applies directly. Give junior analysts a weekly block where they work cases from raw telemetry without the AI summary first, then compare their conclusion with the machine's. That is how they learn to catch the 89% problem: in the ISC2 survey, 89% of practitioners had seen AI recommendations lead to incorrect outcomes at their organization.
Make review a graded skill. Track how often each reviewer overrides an AI verdict and how often those overrides are right. Our CTO Dean De Beer has argued that an override rate of zero is a warning sign, and for a junior analyst it also means they've stopped learning. A good reviewer disagrees sometimes and can show why.
Budget the mentoring. Senior time used to be the scarce input in the old model, and it still is. If a senior analyst's week is fully booked with escalations, no development happens. Put pairing hours in their job description the same way you put on-call hours there.
Rotate before you specialize. SANS recommends structured mentorships and on-the-job rotations for building entry-level talent that works alongside AI. A rotation through detection support and response scoping in the first year gives a new analyst enough context to choose a lane with some idea of what it involves.
Where automation helps, and where it adds pressure
The same tools that remove the queue can make the new seats better training grounds than the queue ever was. A junior analyst in a traditional SOC spent most of their day gathering data. An AI-completed investigation hands them the full timeline on every case, so they see more finished reasoning in a month than a Tier 1 analyst used to see in a year. When every step and its evidence is visible in the investigation record, a junior can review it the way they'd review a senior's case notes. In the tiered model post, Dean described the analyst becoming an "investigation director," managing investigations and refining the technology for their environment.
The pressure is real too. In the ISC2 survey, 65% of practitioners spent more time over the past year deciding when to trust or act on AI recommendations, and 63% spent more time validating AI output. Practitioners were split almost evenly on whether AI has reduced hands-on learning at work, 37% to 36%. Those who reported more stress from AI were more likely to say they are often expected to act on AI output without understanding how it was produced, 34% against 23%. Drop a junior analyst into that environment without a method or a mentor, and they end up signing off on work they can't evaluate.
The same survey offers a useful anchor: 62% said AI has not reduced the need for foundational cybersecurity skills. The fundamentals still have to be learned somewhere. Every SOC now decides, usually without noticing, whether to build that learning into the new seats or assume it happens on its own the way it did in the queue. We made the broader case for reinvesting recovered time in From Alert Queue to Outcome.
What industry experts are saying
SANS CEO James Lyne put the long-term cost plainly when the 2026 workforce research came out at RSAC. If organizations treat the lower end of the profession as something AI replaces and stop developing foundational practitioners, he said, they "won't have seniors and experts later."
Jeff Pollard, vice president and principal analyst at Forrester, told GovInfoSecurity that technology organizations are "largely abandoning the concept of hiring entry-level personnel." In the same piece, SANS's Rob T. Lee argued the field should train people the way medicine and law do, through something closer to a residency or clerkship, and ISC2 CISO Jon France said entry-level work is more likely to change than to disappear outright.
Kam Karaji, director of cybersecurity and risk management at the NFL, warned in Infosecurity Magazine that "we're at risk of squeezing out the entry point." He expects a split between organizations that keep strong internal development pipelines and those that don't. ISC2 CEO Scott Beale, in the same piece, advised using AI to remove repetitive work while keeping the learning and prioritizing mentoring.
Command Zero CEO Dov Yoran has asked the hiring-side version of the question in this series before: if you eliminate your Tier 1 footprint, where do your future Tier 2 and Tier 3 analysts come from? Nobody we've read has a complete answer yet. The organizations making progress are the ones designing the first job on purpose.
The limits of the argument
Between the lines: The ISC2 figures come from 856 professionals who already use AI in their work, so they describe how AI-using practitioners see the change. That group may notice the shift earlier than the profession as a whole. The 56% figure is perception, and nobody has published a count of Tier 1 seats actually eliminated. The SANS role-reduction numbers apply only to organizations that reported AI-driven role changes, and SANS itself found only 16% of organizations reporting real headcount cuts overall.
Other sources carry their own limits. The Stanford research covers AI-exposed occupations in general and doesn't isolate security roles, and its authors say AI isn't always the only cause. The AI Workforce Consortium is led by Cisco, which sells security products and training, and its blog summary doesn't give a survey sample. SANS sells training, ISC2 sells certifications, and Forrester sells research. The aviation comparison is an analogy: no regulator has studied skill erosion in SOC analysts the way the FAA and NASA have studied it in pilots. Our starting-seat and career-stage tables are a working framework built from this evidence, and we haven't tested them across many SOCs.
Command Zero has a stake here. We build an AI SOC platform that takes on investigation work, including the first-pass triage that used to fill the Tier 1 queue. We think the glass-box investigation record makes a better classroom than the alert queue was, and we'd say that. A platform doesn't write a career path, fund a mentor's time or decide to hire a junior analyst. Those decisions belong to people running the SOC.
The bottom line: Tier 1 trained the profession by accident, and AI is removing it on purpose. The fix is to design the first job on purpose too. Before your next headcount plan, answer three questions: which seat a new analyst starts in, what proof moves them to the next stage, and whose calendar holds the hours to teach them. If you can't answer the third, you don't have a pipeline yet. The next post in this series looks at the skills side of that shift: upskilling analysts to direct AI agents instead of competing with them.
Terms in this post
Terms defined in Command Zero's glossary are linked below.
- SOC (Security Operations Center)
- SOC analyst
- SOC Tier 1
- SOC Tier 2
- Alert triage
- Detection engineering
- Threat hunting
- SOC skills gap
- Encoded expertise
Is AI replacing Tier 1 SOC analysts?
AI is absorbing much of the work Tier 1 analysts did, especially first-pass alert triage and data gathering. In ISC2's 2026 survey, 56% of practitioners said AI reduced the need for entry-level roles over the past year, and 53% said it is creating new types of entry-level roles. The job is changing faster than it is disappearing, but the traditional alert-queue role is shrinking.
Where do new SOC analysts start if Tier 1 is automated?
Common starting points include reviewing AI-completed investigations, supporting detection engineering, scoping confirmed incidents under a senior responder, and maintaining the investigation methods that AI agents use. Many analysts also enter from IT, help desk or identity administration roles, which bring environment knowledge with them.
What skills did Tier 1 work teach?
Tier 1 work built a mental library of normal activity in an environment, escalation judgment, documentation habits, and fluency with SIEM, EDR and log data. Those skills still matter. Teams now need to provide the practice deliberately, since the alert volume that used to supply it is handled by automation.
What does a SOC analyst career path look like in an AI SOC?
A practical path moves from supervised investigation review and rotations, to owning detections and scoping lower-severity cases, to specializing in threat hunting, detection engineering, incident response or agent operations. Progression should depend on evidence such as accurate overrides of AI verdicts and rules that get quieter without missing real threats.
Why does entry-level hiring matter if AI handles the routine work?
Senior analysts start as junior analysts. If organizations stop hiring and developing entry-level staff, the shortage of experienced practitioners grows over time. The 2026 SANS | GIAC workforce research found expert and senior roles are already the hardest to fill, while only 4% of organizations said entry-level roles are hard to fill.
How can a SOC keep junior analysts learning when AI does triage?
Schedule regular practice on raw telemetry before showing the AI summary, grade reviewers on the accuracy of their overrides, budget senior mentoring time, and rotate new analysts through detection and response work before they specialize. These practices rebuild the repetitions that the alert queue used to provide.
What new entry-level security roles is AI creating?
Survey respondents point to roles that validate AI output, operate and maintain AI-assisted workflows, and support AI security. Among organizations adding roles, SANS found AI/ML security specialist, AI security engineer and AI governance analyst most often cited, though most of those roles still ask for some experience.



