I hadn’t heard of Dan Guido till a number of months in the past, once I got here throughout the video of a chat he gave at [un]prompted, an AI safety practitioners’ convention. Dan is the CEO and cofounder of Path of Bits, a software program safety analysis and improvement agency that works with corporations in tech, protection, and finance. However Dan wasn’t speaking about safety. He was speaking about what it takes to make an organization AI native, which is near the middle of the bullseye for many people proper now.
We’ve been making an attempt to determine how to do this at O’Reilly, however till I got here throughout Dan’s speak, we didn’t have a structured course of. We’ve been constructing alongside the strains he laid out ever since. So for this episode of Dwell with Tim I requested Dan to reprise the speak earlier than we received to the dialog. He was alleged to take twenty minutes, like his unique convention speak, however he took thirty-five, and I needed to minimize him off barely earlier than the tip to make room for questions. That was a tricky selection, since every thing he needed to say was golden.
Dan opened by reminding us of the present state of play in enterprise AI adoption. In February, Fortune reported on a Nationwide Bureau of Financial Analysis examine through which practically 90% of some 6,000 executives stated AI had produced no measurable change in employment or productiveness at their corporations over three years. Folks began calling it the brand new Solow paradox, after Robert Solow’s 1987 line that “you may see the pc age in every single place besides within the productiveness statistics.”
Dan’s perception is that this isn’t proof that AI doesn’t work. It’s proof that almost all corporations are deploying AI improper. They hand out ChatGPT and Claude licenses, after which management waits for the magic to occur. It doesn’t.
Dan began out by describing three ranges of AI adoption.
- AI assisted is the place everybody begins: “You give folks entry to ChatGPT, it drafts emails, it summarizes paperwork. It’s only a productiveness software, and your group doesn’t change. Your workflows are the very same as they had been earlier than. You simply have a bit of buddy that helps you with a few duties.”
- AI augmented is the place you begin redesigning workflows, in order that AI does the primary go on a code assessment and a human does the second.
- AI native is structural: “That’s the place you’ve redesigned the corporate and its workflows from the bottom up, assuming the AI goes to be there and that it’s a core participant. That’s not likely a software. That’s extra enthusiastic about AI as teammates.”
In his framing, the primary of the three is a software and the final is an working system. For Path of Bits, he stated that “working system” has a selected goal:
“I would like our safety experience to compound as code. Each engagement we do, all the abilities, the workflows, every thing that we construct makes the subsequent engagement quicker and higher.”
Worker resistance is the primary downside
Dan confessed how exhausting it was to get began on the ladder from AI Assisted to AI Native:
“After I introduced final yr that we had been all in on AI, that we had been going to be utilizing it throughout all of our workflows and redesigning the best way the corporate operates, I’d say solely about 5% of the corporate was with me. 95% was resistant.” About 20% was actively resisting. The opposite 75% had been resisting extra passively. “They’ll associate with it in public, however in course of they’ll sabotage it. They’ll hope that in the event that they maintain their head low, this can go over them, and that three months from now administration’s focus will change and it received’t be an issue anymore, and we will get again to doing what we had been doing. That’s the place nearly all of folks land when these initiatives occur.”
Moderately than argue together with his workers, Dan studied the literature on why folks reject new know-how and determined he wanted to deal with 4 biases towards AI: self-enhancing bias, identification menace, opacity, and intolerance for imperfection.
Self-enhancing bias is the behavior of crediting your wins to your personal judgment and your losses to circumstance, which is a specific downside for senior people who find themselves strongly connected to the years of expertise and instinct that received them to their current place. Opacity will not be having the ability to see how a choice received made. Dan’s commentary is that you simply don’t perceive your physician’s reasoning both, however one way or the other you belief the physician however get suspicious of the machine. Dan didn’t point out this work particularly, however intolerance for imperfection appears to confer with Dietvorst, Simmons, and Massey’s work on algorithm aversion, which discovered that folks abandon an algorithm after watching it err as soon as, even when it outperforms the human various. Their follow-up paper discovered that giving folks even a slight means to switch the algorithm’s output is sufficient to overcome the aversion.
Dan spent probably the most time on identification menace. He described a examine through which the identical kitchen equipment was marketed in two methods: “On one hand, it does the cooking for you. Then again, it helps you cook dinner higher. It’s the identical system. The individuals who recognized as cooks rejected the primary model and accepted the second.”
Most data work, Dan argued, and safety auditing particularly, is what he referred to as symbolic relatively than instrumental. That’s, it carries that means about who you’re. “So I’ve to border AI as one thing that makes you a extra harmful auditor,” he stated. “Not that it does the audit for you.”
In his work at Path of Bits, he intentionally constructed a countermeasure for every bias.
- Self-enhancing bias is addressed by “an AI maturity matrix” with seen ranges, as a result of you may’t declare you’re already adequate when there’s a printed ladder that identifies a distinct set of expertise as crucial.
- Identification menace will get expertise repositories, the place an engineer who writes a tough plugin will get credit score for encoding their experience. Hackathons additionally change the dynamic from resistance to exploration. I’m placing phrases in Dan’s mouth right here, however I believe he’d agree that when skilled builders are referred to as on as mentors in a hackathon, that additionally reduces their expertise of AI as an identification menace.
- Intolerance for imperfection will get a curated market, sandboxing, and hardened defaults, so everybody’s first expertise of AI isn’t a catastrophe.
- Opacity will get a written AI handbook that clarifies the utilization coverage and the danger mannequin relatively than simply saying “belief us.”
Right here’s Dan’s slide on “the cures that truly labored”:
Returning to certainly one of my passion horses, this can be a form of mechanism design. In my latest piece on the lacking mechanisms of the agentic economic system, I argued that we have to begin with desired outcomes and ask ourselves what mechanisms will assist to supply them. Dan’s method appears to be actually good at this. Most enterprises are treating AI adoption as a procurement downside or a communications downside. Dan handled it as a query of what incentives, defaults, and standing ladders produce the habits you need, given how folks really reply.
The final treatment on Dan’s record is that the CEO has to steer by instance. He famous, “I used to be the primary particular person via the door. My voice because the CEO issues much more than folks assume. The passive 50% of the corporate that isn’t positive if this initiative goes to achieve success, they’re watching to see what management really does, not what it says.”
A ladder, not a mandate
Path of Bits already tracked about 50 engineering expertise for efficiency assessment, issues like Python, git, Rust, and numerous safety auditing capabilities. Dan pulled AI expertise out into their very own matrix, with 4 ranges, from not engaged via succesful and adoptive to transformative. Every of those ranges is detailed individually and extra particularly for assurance, engineering, gross sales, and undertaking administration.
He famous that “The best degree of the maturity matrix will not be someone who makes use of AI probably the most. It’s someone who invents new methods to work and builds instruments with AI. So the identification of the skilled shifts from ‘I don’t want AI’ to ‘I’m the one who makes AI helpful for the corporate.’” This was his first essential design selection.
The second is what degree zero means. He stated “For those who’re at degree zero, in case you’re not engaged, which means you’re combating again towards the corporate. For those who dismiss AI as hype, in case you refuse to make use of AI for safety work, this can be a disagreement on rules, not on expertise. For individuals who had been caught within the not engaged class, we had exhausting conversations, and there have been individuals who left the corporate.” Ranges one via three are a talent difficulty, and the treatment is time with the instruments.
Whereas the slide describing the potential matrix is proven within the previous video clip, right here’s the place you will discover the total deck so you may examine it in additional element.
Driving adoption and expertise with hackathons
Among the best methods Path of Bits developed to maneuver folks up the ladder was to carry a hackathon each two months. Dan runs them with clear targets relatively than as a free-for-all. The main target space and studying goals are outlined prematurely and introduced per week forward, with separate directions for engineers and non-engineers. Folks work in pairs so every thing will get reviewed. There’s a demo session on the finish, after which follow-through. (It’s an essential a part of Dan’s massive thought, that it’s important to construct a system by which, in his phrases, organizational data and functionality compounds.) He famous that “Within the days afterward we maintain one or two folks round, and so they accumulate all of the reusable artifacts, construction them, and put them into the locations they must be.”
I requested what folks outdoors of product and engineering really work on, because the reply for an accountant at a hackathon was not apparent. Dan’s response is that the hackathon isn’t measured in artifacts shipped however in the place folks sit on the potential ladder the next week. Primarily, he’s working a coaching program that occurs to supply helpful output, relatively than a manufacturing dash that occurs to show folks one thing.
The primary hackathon, he instructed me, was the equal of a seaside cleanup: “It’s like these corporations that ship all people to the seaside with an enormous stick and say, let’s go decide up a bunch of trash and put it away, and then you definately get the massive workforce picture after with all of the contractor luggage of rubbish. That’s what we did with our public supply code repositories.”
He picked it as a result of open supply upkeep is the a part of the job that seems like a grind. No new options, simply closing points and off dependencies on public code the place nothing was in danger. “As an open supply maintainer, you simply get crushed down by the general public. This doesn’t work, I can’t use it, this factor sucks. Dozens of points stating flaws you already knew about. It feels burdensome. We needed folks to see that adopting AI would relieve burden.”
The second hackathon was about delivery impactful product updates, but it surely was additionally designed to maneuver everybody up the potential ladder by giving up management. Engineers needed to run Claude Code in bypass permissions mode, absolutely autonomous, on public repositories, inside sandboxes the corporate had ready prematurely. The one they’re working now could be about persistent background brokers that may be handed a process throughout an audit and are available again with a proof of idea exploit or a draft discovering.
Right here’s a take a look at Dan’s slack message saying the hackathon:
All the things the hackathons produce will get harvested into artifacts.
Path of Bits runs three expertise repositories: an inside one for firm workflows, a public one that anybody can use, and a curated one that vets third-party expertise earlier than they’re allowed in.
Publishing expertise to the general public repository isn’t just a advertising train. “It retains us trustworthy, and it forces us to jot down issues that different folks can use, not simply folks outdoors the corporate however inside too,” Dan stated. “It actually helps us take into consideration the tribal data that’s baked into the software.”
The curated repository exists as a result of Path of Bits is aware of how dangerous the availability chain is. They’ve revealed analysis on find out how to write malicious expertise, and so Dan will not be going to inform 130 workers to begin downloading code from strangers and working it on their laptops. “In order for you adoption, you want a protected provide chain.”
Turning scar tissue into infrastructure
Maybe much more essential than the abilities repository is, as Dan put it, “turning scar tissue into infrastructure.”
“Each single time Claude Code didn’t do one thing we needed, we might bake it right into a set of worldwide, copy-pasteable defaults. Identified good settings, really helpful patterns. I name it scar tissue. If I rent someone new tomorrow, I don’t need them to need to undergo your complete discovery means of the final yr of Path of Bits to determine find out how to use the software.”
The configuration repository, claude-code-config, is the place the amassed classes reside.
Dan constructed the primary model himself after which opened it to drag requests from the entire firm, assigning somebody after every hackathon to go accumulate what folks hadn’t contributed on their very own. “It’s simpler to place out one thing that’s unpolished than it’s to get it good on the primary strive.”
Briefly, an enormous a part of the Path of Bits “enterprise AI working system” method is a set of standardized instruments and hardened defaults. Standardization isn’t a straitjacket. It’s a basis.
On sandboxing, Path of Bits intentionally didn’t decide a single most well-liked resolution. There’s a devcontainer for builders, dropkit for disposable DigitalOcean droplets, COOP for remoted VMs, and the sandboxing now constructed into Claude Code for informal customers. “The purpose isn’t that everyone makes use of the identical sandbox,” Dan stated. “The purpose is that everybody has a protected sandbox to make use of, and that it’s straightforward for them to do it.”
One other of the hardened defaults is procedural. Path of Bits enforces a seven day cooldown on each bundle their builders set up:
“There are dozens of safety corporations scanning the web looking for a brand new cool weblog submit they’ll write about malicious code hiding on PyPI or npm, and so they normally determine on the market’s a provide chain difficulty inside hours. So we simply delay all of the packages that Path of Bits makes use of. Usually the malicious stuff will get picked up earlier than we ever get an opportunity to run it.”
That’s free-riding on a aggressive marketplace for safety analysis, and given the velocity of right now’s market, it’s a sublime resolution. There’s an entire class of defenses like this ready to be discovered, the place the mechanism will not be a technical system however a well-chosen delay.
Information, and DJ Patil’s “Tidy Home”
The issue we run into most frequently as we construct AI workflows at O’Reilly isn’t the mannequin or the tooling. It’s knowledge. Who has entry to which system, which system does that knowledge reside in, and who do I ask? In a 500 particular person firm that’s annoying. I’m wondering what it’s like at an organization with 50,000 workers.
I instructed Dan about DJ Patil’s Tidy Home framing. He agreed that knowledge entry for AI is an enormous downside. His reply begins with permissions:
“The permissions debt is invisible till an agent hits it. Making knowledge agent legible is a compelled permission audit. It’s important to really undergo and work out who can entry what…. It additionally raises the stakes for permissions errors. For those who overshare info, now an agent inside your organization goes to seek out it immediately. There are loads of these technical debt kind of issues the place, with brokers, all of it’s changing into due on the similar time.”
Each shortcut a company took with its knowledge over the previous twenty years is being referred to as without delay, and the businesses that may run the audit, make quick selections about boundaries, after which really share their knowledge are those that can get a drive multiplier.
Dan is towards letting a thousand flowers bloom, as a result of uncoordinated groups create overlap relatively than compounding. He’d relatively have one centralized basis, with innovation taking place on high of that. He recommended a helpful metric for making that work throughout workforce boundaries is what fraction of your workforce’s knowledge did you make reusable for everybody else, and the way a lot of it’s being utilized by groups outdoors your personal.
What post-AI jobs appear like
Earlier than the primary hackathon, Path of Bits ran hands-on periods to show its operations and go-to-market employees the fundamentals of git and the command line. Not mastery, simply sufficient to be a shopper of the factor. Right here we’re fifty years into my profession and the Unix command line nonetheless issues. Dan’s non-technical employees largely work inside Claude Cowork or Codex Desktop now, however he thinks the command line expertise was value it as a result of they know what’s taking place underneath the hood.
What occurs to a job when the software can do loads of what people used to do? Dan gave the instance of his personal technical editors. His editors used the hackathons to construct the instruments that received them out of line modifying, together with one which turns a public presentation right into a weblog submit within the firm’s voice. What the writers do now could be seek the advice of on find out how to body a narrative so it’s efficient with a specific viewers.
I agree. Human jobs aren’t going away any time quickly. This will get heard as optimism when it’s actually simply commentary. AI goes to exchange loads of what we used to do, however additionally it is going at hand us a considerable amount of new work, and far of that work hasn’t been understood but. High quality assurance for agent programs is likely one of the new jobs. So is expertise product administration, which is a task that didn’t exist eighteen months in the past and now has a headcount at a 130 particular person safety agency.
I requested a query in the direction of the tip about how we’re going to know which expertise and brokers are any good. What Dan has to this point is telemetry pulled from builders’ dot recordsdata via the corporate’s system administration system, which tells him what will get used and what breaks, plus one AI programs engineer whose job is product administration for the abilities repository, reviewing incoming pull requests and deprecating overlapping expertise.
What Dan thinks comes subsequent is analysis. He says: “When you make investments so much into these agent programs, you want proof that they do the job. The best way you do that’s you give all people a efficiency assessment. You give them an analysis knowledge set, a benchmark.”
Path of Bits is now constructing benchmarks for its core expertise. How properly can we discover bugs on this language? How properly can we write a press release of labor? Setting up these datasets is actual work, with constructive and unfavorable circumstances, and comparisons towards the algorithmic instruments that exist already.
Put the reps in
I requested Dan for the highest 5 errors he made. He stated there was just one. “It’s essential allocate an acceptable quantity of FAFO time. (That’s F Round and Discover Out.) A product comes out on Friday. There’s no documentation for it. There’s no coaching steerage for it. There’s no course on it. You may’t wait till someone systematizes the data. You simply must do it.”
Then he gave an analogy to going to the fitness center.
The recipe for fulfillment
Dan has a replicable recipe, which he summarized as follows:
- Standardize on one agent workflow that you would be able to assist.
- Write an AI handbook in order that threat selections aren’t advert hoc, and that everybody is enjoying the identical recreation.
- Create a functionality ladder that makes clear that enchancment is predicted.
- Run quick adoption sprints that drive hands-on utilization.
- Seize every thing as reusable artifacts: expertise + configs + a curated provide chain.
- Make autonomous brokers protected with sandboxing + guardrails + hardened defaults.
The Path of Bits expertise repository is public. So is the curated market, the configuration repository, the devcontainer, dropkit, and COOP (Continuity of Operations planning). He wrote up the entire playbook on The Path of Bits Weblog and gave a model of it to tl;dr sec. He thinks publishing makes the work higher as a result of it forces the tribal data out into the open the place it may be checked.
Which brings me again to the Solow paradox, which appeared to vanish by the late 90s, when US combination productiveness did lastly go up. That didn’t occur as a result of computer systems received quicker. It disappeared as a result of corporations found out find out how to reorganize themselves round what computer systems may do, and ultimately these organizational recipes unfold extensively sufficient to indicate up in combination statistics. The identical has to occur right now. The present AI discourse is obsessive about mannequin functionality and largely tired of diffusion. The issue will not be that the fashions are oversold. It’s that nearly no person has finished the required organizational work, and the few who’ve are largely retaining it to themselves.
If you wish to transcend the spotlight movies proven above, watch Dan’s whole speak right here. His slide deck is right here. And remember to take a look at the Path of Bits Github repository.
