Episode 26 of The Security Podcast of Silicon Valley came out on December 1, 2022, with Jon McLachlan hosting. The guest was Andrew Spangler, head of security and compliance at Harness, a software delivery platform (Podbean, Apple Podcasts, Spotify, LinkedIn, Harness). The episode is called “Andrew Spangler, Head of Security and Compliance at Harness, on Building Trust in Security.”
Trust runs through the whole talk. Andrew builds it with clients in a few hours, with a new team during an incident, and with candidates in an interview. The other thread is automation: hand the repeat work to scripts and pipelines, and spend people on the next problem.
This page follows the conversation in order: a section for each part of it, twelve figures, eight clips from the tape, a post from Andrew’s LinkedIn, the full transcript, and five tools: a quiz, a scorecard for a Log4Shell-sized week, a calculator for repeat work, a timeline, and a to-do list. Where the decodes of the tape disagree on a word, the transcript puts its best guess in brackets, and the page doesn’t quote or retell those words.
From a help desk to Harness
Jon opened by reading Andrew’s path from LinkedIn: education technology at a school district, systems administration, security consulting at Rapid7, security analysis at Puget Sound Energy, NCC Group, Datadog, and Harness. LinkedIn dates the first of those roles to August 2009 and the move to Harness to April 2021 (LinkedIn).
Andrew credited the people: good mentors at every stop, and colleagues generous with what they knew.
“Lots of good people along the way”
Andrew Spangler, at 1:25
Can I break it? Can I fix it?
Andrew’s first formal job in IT was a mobile help desk for a few elementary schools, and security caught Andrew’s attention there. Part of the pull is the view behind the scenes of infrastructure and applications. Andrew named the urge that many people in the field share: understand how something works, then take it apart.
“Why does it work this way? And can I break it? Can I fix it?”
Andrew Spangler, at 4:13
Two hundred new tokens
The next job was systems administration. One of the big tasks there was rebuilding the company’s multi-factor authentication after the RSA compromise.
Wikipedia tells that story. On March 17, 2011, RSA announced that it had been the victim of “an extremely sophisticated cyber attack.” The attackers sent phishing emails to two small groups of RSA employees, with an Excel file that exploited a flaw in Adobe Flash, then used the Poison Ivy remote access tool to take over machines and reach servers on RSA’s network. On June 6, 2011, RSA offered token replacements or free security monitoring to its more than 30,000 SecurID customers (Wikipedia).
Jon asked whether the company swapped RSA out. It didn’t: the old RSA tokens came out, and new RSA tokens went in. The vendor’s message, in Andrew’s paraphrase: the root credentials were compromised, they would be rotated, and new fobs and hardware were on the way. Andrew’s job was to decommission the old setup, then spin up, configure and deploy the new keys.
“probably right around 200 tokens that I was able to walk through the process of reprovisioning.”
Andrew Spangler, at 5:21
The work earned more trust.
“my boss started to trust me with a little more firewall administration”
Andrew Spangler, at 5:28
The magic word
Jon caught the word in that story.
“it all comes back to trust”
Jon McLachlan, at 5:55
Jon called trust the cornerstone of security: something between people, and between people and what they expect from systems.
A couple of hours to build trust
A friend put Andrew in touch with someone who had recently started at Rapid7, then a pre-IPO company Andrew didn’t know. The job was security consulting in professional services: helping customers build vulnerability management programs, plus advisory work. Wikipedia notes that Rapid7, a Boston-based security company, acquired the Metasploit Project on October 21, 2009 (Wikipedia).
The engagements lasted one or two weeks, and the consultant came in from outside to advise on sensitive processes. By Andrew’s estimate, that left a couple of hours to build rapport and trust.
Jon asked for a favorite technique. Andrew called it what helps a person feel comfortable: finding common ground, such as video games, music, travel, camping or hiking in the desert. Andrew also worried, as a consultant, that the client knew far more about their own systems, and that the consultant would only be in the way. So Andrew spent a little time up front on casual conversation, to learn what interested each person, in the server room or the meeting room.
“People are the interesting part of any job that I’ve held.”
Andrew Spangler, at 8:34
Different paths, stronger teams
Jon answered the story of Andrew’s route into tech with a point about paths.
“I think that everyone takes a different path”
Jon McLachlan, at 11:08
Those differences, Jon said, make life interesting and make teams stronger than the sum of their individual parts. Andrew agreed, and called it one of the best things about security: the field has been a good avenue for people with all kinds of backgrounds. Some of the best professionals Andrew has worked with came from the liberal arts or had no formal training, alongside plenty of excellence from STEM.
“I love the diversity in backgrounds and the different perspectives”
Andrew Spangler, at 12:09
For Jon, that’s where creativity comes from: differences, more than similarities. Similarities help build trust early. Seeing the world a little differently is where new ideas start. Jon recommended Seeing What Others Don’t, Gary Klein’s 2013 book of case studies on how people gain insights (Wikipedia). Andrew added it to a Goodreads list. The page for episode 16 covers the same book.
Tools for people who build tools
Jon asked what Harness does. Andrew described a software delivery platform: CI and CD, plus modules that work with the platform or on their own, such as feature flags, and security testing orchestration for the software pipelines.
Two things drew Andrew in. One was the leadership team. Harness’s founder, Jyoti Bansal, had founded AppDynamics, which Cisco bought for $3.7 billion in 2017 (Wikipedia, Wikipedia). The other was the space. For years, Andrew had suspected that much of the work Andrew did would someday be automated, and now platforms like Harness do it. Andrew likes industries that
“build tools for people that build tools”
Andrew Spangler, at 14:46
Hand the drudgery to scripts
Jon saw automation as the next step, with layoffs in the news and teams asked to do more with less. Some people fear automation for their jobs. Jon’s view is the opposite: script everything Jon does in a day, and move on to the next layer of problems.
Andrew had felt that fear too: would the work Andrew was good at matter in five or ten years? Time in the seat, and at cloud-native companies like Datadog and Harness, showed Andrew that automation frees people up for more interesting problems.
“It’s less drudgery, it’s less spreadsheets, it’s less manual deployments”
Andrew Spangler, at 16:16
“Let’s let the robots do the drudgery, and we can focus on what comes next.”
Andrew Spangler, at 16:29
Jon pictured a future as a grumpy old man telling a young person about the old days.
“when I was your age, we had this thing called spreadsheets.”
Jon McLachlan, at 16:48
Jon’s own rule: anything done more than once is a sign that it could be automated.
A minor crisis that pulled a team together
Jon asked about Andrew’s best day at Harness. The first came within about 60 days. Andrew had inherited a team of two engineers at a globally distributed, remote-first company. A report came in that a software engineer’s laptop had been compromised by ransomware. Nothing reached the cloud environments or production, but the team wanted to stop any further spread, so they treated it as an incident.
Wikipedia defines ransomware as malware that holds a victim’s data hostage until a ransom is paid, through encryption, theft with a threat to publish, or both (Wikipedia).
Working the incident built confidence on the team, and Andrew tied it back to trust.
“going through sort of a minor crisis together really has a way of gelling a team.”
Andrew Spangler, at 19:05
Priority zero
The second best day was about hiring. The people Andrew works with, and works for, have always been what Andrew calls “priority zero.” Hard workers who give everything can still fall short without a good team around them. Building the security team out from two people has been rewarding, Andrew said, with hires who are competent, work hard and catch the vision.
The best worst day
The third was what Andrew called the best worst day: Log4Shell. Wikipedia describes it as CVE-2021-44228, a flaw in the popular Java logging library Log4j that allowed arbitrary code execution. It was privately reported to the Apache Software Foundation on November 24, 2021, got its CVE identifier on December 10, 2021, and received a CVSS score of 10, the highest available (Wikipedia).
At Harness, the response reached well past security. Engineering, customer success and legal all joined in, calm and collected but urgent, over an incident that kept changing for several weeks.
“We’re all in one piece, and now we’re on the other side.”
Andrew Spangler, at 20:36
A professional pessimist
Jon noticed that two of Andrew’s three best days were incidents. Both of them, Jon said, look for the good that will come out of whatever is going on.
“as a security professional, you’re kind of a professional pessimist”
Andrew Spangler, at 21:19
Security people ask the tough questions, so it gets easier to see what’s wrong. Andrew’s counterweight is perspective: look at the moment, back at the past and ahead to the future, and see that even when things feel chaotic, it’s going to be okay.
“This is not the worst thing that’s happened.”
Andrew Spangler, at 21:59
Jon agreed about perspective. Software that flies airplanes or doses chemotherapy puts lives on the line, and most CI/CD systems don’t.
Keeping the lights on
That brought Andrew back to Puget Sound Energy. Wikipedia describes it as a Washington utility that provides electric power and natural gas to the Puget Sound region, with more than 1.24 million electric customers (Wikipedia). Andrew’s role included vulnerability management for operational technology. It took Andrew into the field, to generation facilities and substations.
Wikipedia defines operational technology as hardware and software that detects or causes a change through direct monitoring and control of industrial equipment, sometimes called IT in the non-carpeted areas (Wikipedia). Andrew worked with professional engineers on the security of components of the bulk electric system. NERC, the regulator for North American grid reliability, describes its mission as reducing risks to the reliability and security of the grid (Wikipedia).
What stood out was time. Phones, keyboards and desktop computers have a good lifespan of three to five years.
“In the OT space, these systems are designed for 50 to 70 years.”
Andrew Spangler, at 25:14
Andrew described walking into a substation and hearing that the equipment went in in 1971, still runs, and shouldn’t be touched. Andrew came away with deep respect for the people who keep the lights on and keep the drips in hospitals moving.
Jon tries to keep a phone for six years. Andrew teased Jon about the single camera on the back of the old one, and Jon reported an upgrade.
Twenty whys
Jon named a growth mindset. Andrew traced it to curiosity, the kind that asks why about an answer, then keeps asking.
“there’s 20 whys afterwards”
Andrew Spangler, at 26:53
Andrew finds that same shared curiosity in the people Andrew connects with most in the industry. Jon often says that in a world without computers, Jon would probably have been a philosopher. Jon’s advice: ask why over and over, down to the core values in play, then compare them with your own. From a different view of the world, some problems stop looking like problems.
“Sometimes they look like features.”
Jon McLachlan, at 28:19
When people move on
Jon asked about the worst day. Andrew’s answer was people leaving the team for their next adventure. Even when the move is good for them, it’s bittersweet. And on a small team at a fast startup, 40 or 50 hours of work a week suddenly need a new home.
The saving grace, Andrew said, is the team: good people with different backgrounds, curiosity and an eagerness to learn.
Jon started as an engineer and moved into leadership. One of the most rewarding parts of that shift is helping people take their careers to the next level, at the same company or a different one.
“my loyalty is connected to people.”
Jon McLachlan, at 31:24
When growth can’t happen inside the company, Jon still helps, with options from Jon’s network. Watching people grow into leaders is the reward. Andrew agreed. Seeing people succeed, and helping with mentorship and insight along the way, has been the best part of Andrew’s last five or six years.
Tell me about a win
Jon asked for Andrew’s favorite interview question. Andrew asks candidates to describe a win from their career. An interview starts like any first meeting with a stranger, with some ambiguity and trepidation. A candidate might expect a quiz, or a trivia session with a search engine open on the side. Andrew wants to know what the candidate thinks a win is.
“I feel like that gives candidates the opportunity to brag a little bit”
Andrew Spangler, at 33:22
They get to talk about something they’re proud of and comfortable discussing. It tells Andrew more about a person than most trivia questions do. It also shows their values, and where a new hire could land a quick win during onboarding.
Pipelines that fix themselves
Jon asked what tool Andrew wishes someone would build. The first answer was personal: an easy button for nitro cold brew at home, next to the ice and water on the fridge. Wikipedia describes nitro cold brew as cold brew coffee with nitrogen gas added, introduced in the early 2010s (Wikipedia). If Samsung is listening, Andrew would buy one, and Jon called that a first sale.
The professional answer was self-healing tools, or AI remediation, like Dependabot on GitHub: the tool queues up a pull request, and a person reviews and approves it. Andrew wants that tied to the work on software bills of materials and vulnerability exchange. Wikipedia describes a software bill of materials as an inventory of the components used to build a piece of software, open source and proprietary (Wikipedia).
“the robots can do this work and free us up to be more focused on what comes next.”
Andrew Spangler, at 36:13
Jon asked whether the secure software development lifecycle could be fully automated someday. Andrew would love to see it, and said Harness was running in that direction.
“I think there’s a future where we’re at least close to that.”
Andrew Spangler, at 36:41
Give security a shot
For advice to a younger self, Andrew spoke to people outside the field. There are many jobs in security, and the name alone can sound intimidating. The industry carries historical baggage, and Andrew sees healing at many companies.
One of the best parts of the job, for Andrew, is talking with people who are considering a career in security, or a move into it from another role. That, Andrew said, is where the industry gets stronger, and where much of its creativity comes from. Andrew’s last words of advice: give security a shot, and you might be surprised how much you enjoy it.
“give security a try.”
Andrew Spangler, at 38:06
What happened after the tape
Andrew’s role since. LinkedIn lists Andrew as senior director and head of security and compliance at Harness since August 2022, after a director role in the same job from April 2021 (LinkedIn).
Talking with people new to security. Andrew’s advice was to give security a shot, and to reach out. LinkedIn lists Andrew as a mentor in the Cyversity Mentorship Program’s Winter 2023 cohort, “Sharing experiences, tips, recommendations, and advice for the next generation of security professionals” (LinkedIn).
Compliance as a baseline. In May 2025, Andrew shared a conversation on Security & GRC Decoded, a podcast hosted by Raj Krishnamurthy of ComplianceCow. In the host’s summary, Andrew argued that compliance becomes a forcing function for a good security baseline (LinkedIn). In November 2025, Andrew praised the Harness GRC team for bringing software delivery tools, CI/CD and AI into governance work, in a whitepaper the team published (LinkedIn).
Standards for AI agents. On March 31, 2026, Andrew announced joining the consortium behind AIUC-1. The standard’s site calls it “The technical certification for agent security,” maintained by the Artificial Intelligence Underwriting Company, with six risk domains, from data and privacy to society (AIUC-1).
Andrew on joining the AIUC-1 consortium to shape AI best practices.
Andrew Spangler · March 31, 2026
The same themes on this show. Episode 17 had Sean Cassidy, then the first head of security at Asana, on the day Log4Shell turned engineering into allies. Episode 36 had Feross Aboukhadijeh of Socket on dependency updates and the open source supply chain. Episode 37 opened The Founder’s Guide to Compliance.
Did you catch it? A seven-question quiz
Seven questions from the episode. Each answer links the moment in the transcript or its section.
Quiz · seven questions
Did you catch it?
Pick an answer to see whether you caught it.
-
What did Andrew rebuild as a systems administrator?
b. "probably right around 200 tokens that I was able to walk through the process of reprovisioning." Transcript, 5:21.
-
On a one- or two-week consulting engagement, how long did Andrew say you have to build rapport?
a. A couple of hours, by Andrew's estimate, to build rapport and trust with a client. See A couple of hours to build trust.
-
How does Andrew get a short engagement off to a good start?
c. Common interests, such as video games, music or travel, and a little time up front to learn what interests each person. See A couple of hours to build trust.
-
What incident in Andrew's first 60 days at Harness pulled the new team together?
b. Nothing reached the cloud or production, and the team treated it as an incident. See A minor crisis that pulled a team together.
-
Which incident did Andrew call the best worst day?
a. Engineering, customer success and legal joined the response, over several weeks. See The best worst day.
-
How long are OT systems designed to last, by Andrew's account?
c. "In the OT space, these systems are designed for 50 to 70 years." Transcript, 25:14.
-
What is Andrew's favorite interview question?
b. "I feel like that gives candidates the opportunity to brag a little bit" Transcript, 33:22.
Ready for a Log4Shell week?
Eight questions on whether your team could have its own best worst day, built from the episode. Each tip links the section behind it.
Scorecard · eight questions
Ready for a Log4Shell week?
Answer for the services your customers depend on most. If two answers fit, pick the lower one.
-
If a library like Log4j had a critical flaw today, could you list every place it runs?
- No
- For some services, after a search
- Yes, from a bill of materials for our software, kept current
A bill of materials for software lists the components behind it. See Pipelines that fix themselves.
-
Who brings engineering, customer success and legal together in an incident?
- Nobody named
- Whoever is around
- A named incident lead, with a runbook
"We're all in one piece, and now we're on the other side." See The best worst day.
-
Can a fixed dependency reach production through your pipeline the same day?
- No
- With manual steps
- Yes, through the normal pipeline
"It's less drudgery, it's less spreadsheets, it's less manual deployments" See Hand the drudgery to scripts.
-
Do bots open pull requests for vulnerable dependencies, with a person approving each one?
- No
- In some repositories
- In every repository
"the robots can do this work and free us up to be more focused on what comes next." See Pipelines that fix themselves.
-
If an engineer's laptop reports ransomware, is there a plan?
- No
- An informal one
- Yes, written and practiced
"going through sort of a minor crisis together really has a way of gelling a team." See A minor crisis that pulled a team together.
-
Could you tell customers quickly whether they're affected?
- No
- After a few days
- Yes, the same day
Customer success was part of the response at Harness. See The best worst day.
-
Does the team look back at each incident together?
- No
- Sometimes
- After every incident
"This is not the worst thing that's happened." See A professional pessimist.
-
If one person leaves, does their part of the response have a new owner?
- No
- Partly
- Yes, written down
On a small team, one departure can leave 40 or 50 hours of work a week. See When people move on.
A Log4Shell-sized week would start with a search. Begin with an inventory of your dependencies and a named incident lead.
The basics are there. Add dependency pull requests, a written ransomware plan and a same-day path to production.
Keep practicing, and keep the inventory and the owners current as the team changes.
How much drudgery can you hand off?
The calculator counts the hours your team gets back when scripts take over repeat work, as in Hand the drudgery to scripts.
Calculator
How much drudgery can you hand off?
Four numbers: the people on the team, the hours each spends a week on repeat work such as spreadsheets and manual deployments, the share a script or pipeline could take, and the working weeks in your year.
A team of 3 spending 8.0 hours each a week on repeat work gets back about 12.0 hours a week when scripts take 50% of it. That is about 576 hours over the working weeks, or 14.4 40-hour weeks of work.
The defaults are placeholders, so swap in your own counts. The math assumes the scripts need no upkeep, so treat the result as a ceiling. Andrew's measure is what the time goes to next: harder problems and strategy.
From a help desk to the AIUC-1 consortium
Andrew’s path, the field around it, and this show. Filter by topic.
Timeline
From a help desk to the AIUC-1 consortium
- AppDynamics
Jyoti Bansal, later a co-founder of Harness, founds AppDynamics (Wikipedia, on this page).
- A school help desk
Andrew starts in IT as an education technology specialist (LinkedIn, on this page).
- Rapid7 buys Metasploit
The Boston-based security company acquires the Metasploit Project (Wikipedia).
- Systems administration
Andrew's next role, where the RSA rotation would land (LinkedIn, on this page).
- The RSA breach
RSA announces an attack that put SecurID information at risk (Wikipedia, on this page).
- New tokens on offer
RSA offers token replacements or free monitoring to its more than 30,000 SecurID customers (Wikipedia).
- Rapid7
Andrew joins Rapid7 as a security consultant (LinkedIn, on this page).
- Puget Sound Energy
Andrew becomes a senior security analyst at the utility (LinkedIn, on this page).
- NIST's energy sector work
Andrew contributes to NIST practice guides through the National Cybersecurity Center of Excellence, into November 2017 (LinkedIn).
- NCC Group
Andrew joins the security consulting firm (LinkedIn, Wikipedia).
- Datadog
Andrew becomes team lead for information security and compliance (LinkedIn).
- Harness
Andrew joins as director and head of security and compliance (LinkedIn, on this page).
- Log4Shell reported
The Log4j flaw is privately reported to the Apache Software Foundation (Wikipedia).
- CVE-2021-44228
Log4Shell gets its CVE identifier, with a CVSS score of 10 (Wikipedia, on this page).
- Episode 17
Sean Cassidy on Log4Shell and a high-trust security culture at Asana (episode 17).
- Senior director
Andrew becomes senior director and head of security and compliance (LinkedIn).
- This episode
Andrew on the show (Podbean).
- A mentor
Andrew mentors in the Cyversity Mentorship Program's Winter 2023 cohort (LinkedIn, on this page).
- Episode 36
Feross Aboukhadijeh of Socket on dependency updates and the open source supply chain (episode 36).
- Security & GRC Decoded
Andrew shares a conversation on compliance as a security baseline (LinkedIn).
- A GRC whitepaper
Andrew praises the Harness GRC team's whitepaper on automation in governance work (LinkedIn).
- AIUC-1
Andrew joins the consortium behind the AIUC-1 standard for AI agents (LinkedIn, AIUC-1).
What to do this week
Five things, in the order of the episode.
- Start with common ground. On your next review with a team you don’t know, open with a casual conversation about what interests them, as in A couple of hours to build trust.
- Script one repeat task. Pick the spreadsheet or manual deployment that comes back every week and hand it to a script or the pipeline. The calculator shows what the whole list is worth.
- Know where every library runs. Keep a bill of materials for your software current, so the next Log4Shell starts with a lookup (Wikipedia). The scorecard shows the other gaps.
- Let bots open dependency fixes. Turn on dependency pull requests, and have a person approve each one, as in Pipelines that fix themselves.
- Get help with it. Our 24/7 Monitoring & Response team watches your infrastructure and responds hands-on, Product Security builds security into the pipeline, and a vCISO can lead the team. If you want to talk it through, book 15 minutes with us.