Inside the Silicon Mind

Adam Fisher and Liran Eshel discuss AI adoption gaps, agentic engineering roles, and societal impacts of productivity shifts

41 min Listen to the original
Artificial IntelligenceVenture CapitalSoftware EngineeringOrganizational ChangeLabor ProductivitySaaSAgentic AIInvestment Strategy

Summary

Adam Fisher and Liran Eshel discuss the divergence in venture portfolios between companies rapidly adopting agentic engineering and those lagging behind. They outline organizational strategies for AI adoption, including managing employee resistance and redefining engineering roles around human-centric oversight. The speakers highlight how AI compresses investment cycles and enables horizontal expansion, while also warning of the psychological toll of constant AI interaction. Finally, they address the societal implications of wealth concentration and the potential for small AI-native teams to outpace large enterprises.

Episode sections

  1. 1:07 – 5:14

    The Agentic Awakening Playbook

    Adam Fisher explains that Bessemer's portfolio has split into two groups: those rapidly adopting agentic engineering and those lagging behind. To bridge this gap, Fisher and Liran Eshel co-authored 'The Agentic Awakening,' a playbook designed for C-level executives to understand the organizational shifts required for AI adoption.

  2. 5:14 – 12:27

    Managing Employee Resistance to AI

    The speakers discuss how to handle employees who refuse to adopt AI tools. Liran Eshel suggests that managers must first lead by example and provide training, citing methods like hackathons and 'hero moves' where leaders rewrite code to prove AI's value. He notes that while conversion is preferred, persistent resistance may lead to termination.

  3. 12:27 – 16:35

    The Shift from Coder to Manager

    Liran Eshel argues that software engineers are transitioning into managers of agentic processes. He identifies four human-centric roles that remain essential: Chief Architect, Chief Security, Chief Design, and Chief Product, emphasizing that AI cannot replace the accountability and judgment required in these areas.

  4. 16:35 – 21:19

    Investing at AI Speed

    Adam Fisher describes how AI has compressed investment cycles, making traditional 8-10 year SaaS timelines obsolete. He notes that companies no longer need to stay in their lane, as AI enables rapid horizontal expansion into adjacent markets, making speed of execution and product building the primary criteria for investment.

  5. 21:19 – 25:35

    The Addictive and Draining Nature of AI Work

    The speakers describe the psychological toll of agentic engineering, comparing it to a slot machine or the endless loop of email. They note that while AI increases output, it creates a cycle of constant feedback and token consumption that can lead to exhaustion and 'vibe coding' without proper architectural oversight.

  6. 25:35 – 28:47

    Building AI-Native Products

    Liran Eshel shares that he built an agentic business intelligence platform called 'Brain' to replace legacy BI tools like Power BI. Adam Fisher identifies information sharing among investors as a key area for automation, suggesting that AI could better leverage the collective knowledge of a venture fund's portfolio.

  7. 28:47 – 31:50

    Two Against Fifty and Future Outlook

    Both speakers agree that small teams with AI can outpace large organizations, with Eshel predicting that a handful of people could eventually put hundreds-employee companies out of business. Fisher adds that while small AI-native teams can be highly efficient, they will still need to add humans to maintain and grow at scale.

  8. 31:50 – 33:25

    Playbook Impact and Timing

    Speaker_1 describes being intrigued by the playbook due to personal consulting experience and exposure to multiple teams. They note the critical timing relative to recent AI model releases and the rapid pace of change.

  9. 33:25 – 34:55

    Emergent Patterns in AI Adoption

    Speaker_1 observes that despite companies paving their own paths, they converged on similar strategies. Speaker_2 clarifies these companies lacked a pre-existing structure, instead replaying emergent patterns that formed the basis of the playbook.

  10. 34:55 – 35:39

    Book Recommendations: Accidental Empires

    Speaker_1 recommends 'Accidental Empires' by Robert X. Cringely, drawing parallels between the origins of the tech industry and the current emergence of AI agents as new 'accidental empires'.

  11. 35:39 – 39:00

    Cognitive Dependence and Human Skills

    Speaker_0 expresses concern about over-reliance on AI affecting cognitive abilities and reducing human-to-human interaction. Speaker_2 agrees, recommending books on human nature like 'The Laws of Human Nature' to maintain essential interpersonal skills.

  12. 39:00 – 41:05

    Wealth Concentration and Societal Impact

    Speaker_1 identifies the concentration of wealth and productivity among small groups as a major concern, citing the expected IPOs of Anthropic and OpenAI. They compare the scale of this intelligence revolution to the industrial revolution.

Assets discussed

AssetKindStanceContext
Anthropicprivate companyneutralIf you see how much concentration around the two new the the the expected the IPOs of uh anthropic and OpenAI.
OpenAIprivate companyneutralIf you see how much concentration around the two new the the the expected the IPOs of uh anthropic and OpenAI.

“Stance” is how the speakers talked about the asset, not a recommendation.

Claims worth verifying

Adam Fisher is a partner at Bessemer who backed Wix and Fiverr before they went public.

Unidentified speaker (speaker_0) ·0:14

How to verify: Check Adam Fisher's LinkedIn profile and Bessemer's portfolio page for Wix and Fiverr investment announcements. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Liran Eshel is the co-founder and chairman of CTERA.

Unidentified speaker (speaker_0) ·0:14

How to verify: Check CTERA's corporate website or investor relations page for leadership biographies. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Fisher and Eshel co-authored 'The Agentic Awakening' playbook.

Unidentified speaker (speaker_2) ·1:07

How to verify: Search for the publication 'The Agentic Awakening' by Adam Fisher and Liran Eshel on their personal websites or LinkedIn. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Bessemer's portfolio showed a gulf between companies adopting agentic engineering and those that were stuck.

Unidentified speaker (speaker_2) ·1:27

How to verify: This is an internal observation; verify by checking public case studies or press releases from Bessemer regarding portfolio company performance during the AI shift. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

One company had employees write 100 CRMs in two days during a hackathon.

Unidentified speaker (speaker_1) ·9:23

How to verify: This is an anecdote; verify by searching for news or blog posts from companies known for AI hackathons around the time of the podcast (late 2026/2026). For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

A CTO rewrote a core module over a weekend using AI to convert a skeptical team leader.

Unidentified speaker (speaker_1) ·10:03

How to verify: This is an anecdote; verify by checking engineering blogs or LinkedIn posts from CTOs discussing AI adoption stories. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

Liran Eshel built an agentic business intelligence platform called 'Brain' to replace Power BI and Tableau.

Unidentified speaker (UNKNOWN) ·26:01

How to verify: Search for 'CTERA Brain' or 'Liran Eshel Brain platform' on product hunt, CTERA's website, or tech news sites. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Adam Fisher states that two people with AI can do the work of 50 or 100 people.

Unidentified speaker (speaker_1) ·0:45

How to verify: This is a claim of productivity ratio; verify by checking industry reports on AI coding assistant productivity metrics (e.g., GitHub Copilot studies) or Eshel's other public statements. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Investment cycles in SaaS have shrunk from 8-10 years due to AI.

Unidentified speaker (speaker_0) ·19:39

How to verify: Verify by comparing historical SaaS company growth/exit timelines with recent AI-native startup trajectories using Crunchbase or PitchBook data. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Liran Eshel is convinced that a handful of people with AI could put a company with hundreds of employees out of business.

Unidentified speaker (speaker_0) ·29:15

How to verify: This is a prediction; verify by tracking market share shifts in software sectors towards AI-native startups versus legacy incumbents. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

The turning point for AI adoption was October/November 2025, six months after Claude Code version two and Opus 4.5 came out.

Unidentified speaker (speaker_1) ·32:39

How to verify: Check release dates for Claude Code v2 and Opus 4.5 via official Anthropic and OpenAI blogs or press releases from late 2026/early 2026. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

The playbook was written after conducting research five months after the October/November 2025 turning point.

Unidentified speaker (speaker_1) ·32:59

How to verify: Cross-reference the publication date of 'The Agentic Awakening' playbook with the stated timeline of research completion. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Anthropic and OpenAI are expected to have IPOs.

Unidentified speaker (speaker_1) ·39:20

How to verify: Check SEC EDGAR filings for S-1 registration statements from Anthropic and OpenAI, or review recent financial news reports regarding their IPO plans. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Primary source

The current revolution is of the size of the industrial revolution.

Unidentified speaker (speaker_1) ·40:00

How to verify: This is a subjective comparison; verify by checking if other economists or historians have made similar comparisons in publications or interviews during the verification window. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Two people with AI can do the work of fifty.

Unidentified speaker (speaker_1) ·39:40

How to verify: This is a claim of potential productivity; verify by looking for case studies or productivity reports from companies adopting agentic AI that demonstrate such ratios. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

'Accidental Empires' talks about the origins of the tech industry.

Unidentified speaker (UNKNOWN) ·35:00

How to verify: Read the book 'Accidental Empires' by Robert X. Cringely and confirm its content covers the history and origins of the tech industry. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

'The Laws of Human Nature' is by Robert Greene.

Unidentified speaker (speaker_2) ·37:01

How to verify: Check the publication details of 'The Laws of Human Nature' to confirm the author is Robert Greene. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

'True Believer' is by Eric Hoffer.

Unidentified speaker (speaker_2) ·37:53

How to verify: Check the publication details of 'True Believer' to confirm the author is Eric Hoffer. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

AI made writing code ten times faster.

Unidentified speaker (speaker_0) ·31:50

How to verify: This claim is in the description, not the transcript part. However, if attributed to the speakers, verify via industry reports on AI coding assistant productivity metrics (e.g., GitHub Copilot studies) from 2026-2026. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

The playbook 'Agentic Awakening' was published.

Unidentified speaker (speaker_1) ·33:19

How to verify: Search for the publication of 'The Agentic Awakening' playbook on the speakers' or their firm's website. For recent events, verify 2026-08-18 through 2026-11-18. Use another period only when the speaker explicitly names it.

Do your own research

  • Verify the claims in 'The Agentic Awakening' playbook by reviewing the document or related case studies from Bessemer Portfolio.
  • Check the performance metrics of companies in Adam Fisher's portfolio that have adopted agentic engineering versus those that have not.
  • Research the job market trends for 'Chief Architect', 'Chief Security', 'Chief Design', and 'Chief Product' roles to assess demand.
  • Review the technical capabilities of agentic BI platforms like 'Brain' compared to legacy tools like Power BI for enterprise adoption.
  • Analyze the psychological impact studies on 'vibe coding' and agentic engineering workflows to understand the reported exhaustion and addiction risks.
  • Investigate the IPO prospects and financials of Anthropic and OpenAI to assess the wealth concentration claims made by the speakers.
  • Read 'Accidental Empires' by Robert X. Cringely and 'The Laws of Human Nature' to contextualize the speakers' historical and psychological analogies.
  • Verify the timeline of AI model releases (Claude Code v2, Opus 4.5) mentioned as turning points for adoption in late 2025.
Full transcriptexpand

0:00Unidentified speaker (speaker_0): Would you fire a really good employee for refusing to use AI?

0:04Unidentified speaker (speaker_1): They fall under the category that we call resistance. They defy the new religion, then you don't have a choice, unfortunately. If you enjoy the writing of the code, then that world is gone.

0:14Unidentified speaker (speaker_0): Liran Eschel, co-founder Sotera. Adam Fisher backed startups like Wix and Fiverr long before they went public. Together they wrote the Agentic Awakening. I wanted to understand how companies can get ahead with AI and what that demands of the people trying to keep up. If AI is doing more of the work, why can the human end up more exhausted?

0:33Unidentified speaker (speaker_1): It's like a slope message with a coin that's the prompt and you're waiting for the answer. It's the most addictive thing I've done in my life. It's

0:39Unidentified speaker (speaker_2): not moving so fast that what I think is relevant today could be irrelevant in a few weeks. If you don't do it, your competitor is doing it.

0:45Unidentified speaker (speaker_1): I see today already there's this two against 50 fifty happening. Two people can do 50 or 100 or 1000 thousands quite and what would other people do and how it would impact the society.

0:57Unidentified speaker (speaker_0): Adam, I'm really fascinated about a paper that I recently read, and I'm really keen to get your insight on what is a gentic awakening.

1:07Unidentified speaker (speaker_2): Agentic awakening is our playbook for the C sea levels at various types of software companies. uh, startups in particular. Uh, we realized about nine months ago that uh a vast gulf had opened up within our portfolio between companies that were really taking agentic engineering to the limit and moving at a very rapid pace to adopt and innovate and implement, and many others that were stuck behind behind. Obviously, we realized it's a spectrum, but from a fund perspective, we wanted to find a way of both educating and making uh companies really feel comfortable about moving forward and moving forward faster. This is an enormous change for how RD functions, uh, the organization, the roles, uh, even some of the a lot of the tooling and the expenses. And we realize that sometimes the best way to start is actually by sitting down and either reading or listening to something quite comprehensive, but that is also speaking directly to the C level. Not necessarily the most technical person in the organization, not necessarily even the founder, but somebody who who understands what is happening

2:26in the in the industry, what excellence should look like, but also what it involves. Because a lot of people uh understood uh incorrectly that it's just about adopting A AI coding agents or certain AI tools. And what we tried to reveal is that it's much more than that, but also what awaits, meaning what what what does it mean for the organization when you reach that that point? And so we realized that we needed somebody who was highly technical, that understood what it meant to be a C level, to be a CEO, and what it means to be responsible for the organization, because we're talking about quite significant, if It's not profound, changes for the organization, uh that many are are hesitant to take. Um, but also somebody who themselves was hands-on with the latest uh uh kind of building, was it was was himself a a builder. Now that's a rare find, because if you find them they're usually running their own startup and they don't have time to do what what we wanted to do. So uh

3:31Unidentified speaker (UNKNOWN): I

3:31Unidentified speaker (speaker_2): I reached out to my friend Liran who was uh a founder of a company uh that we invested in, Citera, Citra, and had moved on to the the chairman role. And I knew he'd been building quite a bit over the last three years, both internally and externally. And I realized like this is this is the guy because he's both kind of

3:52Unidentified speaker (UNKNOWN): part

3:52Unidentified speaker (speaker_2): of the of the new movement, but also experienced and has that business side. And then we set about trying to learn from as many companies as we could, which is interesting because it meant both some of the most advanced companies, which includes both native AI companies, those that were born in the current kind of AI era and could start from scratch, but also more established companies that had successfully made that leap forward. And then importantly, Uh also with companies that were struggling to actually understand where and why they're struggling. or where they get stuck. Um and trying to learn from as many as we can so that we could build a playbook. And that's the that's what Agentic Awakening is. It's a playbook that is for the C level exec. Uh it is long, it is detailed, but it's all but it's very relevant, again, for the C level to understand what needs to be done.

4:44Unidentified speaker (speaker_0): Very fascinating document. And I

4:47Unidentified speaker (UNKNOWN): f

4:48Unidentified speaker (speaker_0): found myself resonating with a lot of areas. within the document, but also some areas made me think, okay, what does this actually mean? One of the things that your playbook recommends is support and a deadline for employees to adopt AI, as if they need to do this really quickly. It's important, especially with the pace of what's happening in this new shift in the market. Now, would you fire a really good employee for refusing to use AI?

5:14Unidentified speaker (speaker_2): And I can say from the venture side, it's it's it's less it's it's it's less relevant, but as as Liran uh you know will will point out, I assume, uh yes, the the you when as we spoke to companies, there we understood very clearly that uh uh there were some quite a bit of holdouts um when it comes to making this making this change. It's a big mental shift for people. in particular those who are more experienced, but not only. And that you have to reach that kind of point where a majority have been so-called awakened. And there are various methods in which to help overcome that. It's not simply about forcing them or firing. And so a lot of the great anecdotes that we were able to pull out of companies were were really revealing because there are there are various ways. of doing it, both through peers, uh by example. Um maybe I'll let Liran give kind of expand on the answer.

6:11Unidentified speaker (speaker_1): Yeah, I think uh

6:13Unidentified speaker (UNKNOWN): no

6:13Unidentified speaker (speaker_1): everybody everybody's focusing on the you know how many people are gonna let go, you know, who's gonna make the transition or not. But also think about the companies. It's not like you have so much talent out there available that you can just hire AI-pilled engineers to your 10-year-old company and expect to fill the ranks so quickly. It's new for most engineers, and some take more time to transition. So I think response responsible managers first give proper chance to their teams to make this transition happen and give them the tools to do that.

6:51Unidentified speaker (UNKNOWN): And

6:51Unidentified speaker (speaker_1): if you're not doing it yourself as an executive, if you don't have that personal experience first and give that the opportunity, don't blame your employees for not making the jump. That's what I would say.

7:02Unidentified speaker (speaker_0): I've experienced this myself just recently. Uh this is why the word awakening caught my attention. We have been quite intentional with trying to automate a lot of tasks. And um

7:16Unidentified speaker (UNKNOWN): I

7:17Unidentified speaker (speaker_0): know this probably doesn't sound lands really well, but I I let my operations manager go about two months ago, and we just got to a point where we felt the company was outgrowing his role.

7:28Unidentified speaker (UNKNOWN): Um, and

7:29Unidentified speaker (speaker_0): I'm left with a lot of tasks that I don't know what to do with these tasks. Do I hire somebody? But I thought let me just figure this out. And I spent probably about a week. weekend on perplexity computer on my own, sitting there in my room, and automating the majority of everything this individual is doing, pretty much.

7:49Unidentified speaker (UNKNOWN): Um,

7:49Unidentified speaker (speaker_0): something very simple as an invoice comes in, it gets flagged up, filed away, and I don't need to worry about it. It it it it speeds things up. That's why the word awaking awaken was interesting. And another example of that, we hired somebody recently who came into the business. He's got all these tools and he just started building things very quickly. It's um it's interesting, it's fascinating.

8:12Unidentified speaker (UNKNOWN): Uh,

8:13Unidentified speaker (speaker_0): which is why the word awakening caught my attention. Now, what's your thoughts about that? Because look, the world is changing and things are moving very, very fast.

8:26Unidentified speaker (UNKNOWN): The reason

8:26Unidentified speaker (speaker_0): I asked that first question is we're kind of forced to become agentic, but I I've personally seen a lot of value in it. What happens if somebody doesn't feel comfortable or isn't willing to

8:41Unidentified speaker (UNKNOWN): go

8:42Unidentified speaker (speaker_0): through that change management? What do you do about that individual in your company? Do you try to coach them? Do you move change their role or do you decide that maybe this person isn't right for the business?

8:54Unidentified speaker (speaker_1): We actually uh you know we're exposed to several very interesting stories from how people how different managers handle that. Um

9:03Unidentified speaker (UNKNOWN): you

9:03Unidentified speaker (speaker_1): know, the first things we saw a company doing is doing this off-site hackathon, you know, taking people outside of their normal environment, maybe in an off-site off site or a few days. It needs to be more than just one-hour session presentation on how to use AI. Okay. For example, one company said everyone writes a CRM. And they got like a hundred CRMs at the end of these two days. Uh or another one said, We're gonna do a competition where everybody's gonna gonna come up with a new application the company needs, and you're gonna build that. Um And and they gave them the tool. tools and instructions how to do that. And it was over like maybe two, three days. And then then I think people were more empowered to know what they can do. But even that was sometimes not enough. And I talked to uh the CTO of a company and he said I had my the team leader of my core team would not nudge, like I we would say, there's no way we call this uh with this typecast the legacy guardian. Okay. He says, I understand the AI is good, but it's not good for our company because our core is too complex. If we put AI here, just gonna ruin everything and we're gonna have a big security and compliance issue, and so on. So in that case, the CTO uh we call it the hero play, the hero move. Like he actually wrote over all weekend,

10:23he rewrote one of the main parts of the code from scratch using AI. And he came on on Monday morning and he gave him this work-in uh module. And the guy just operated that, and at the end of it says, I can't believe that you did that. And he was converted. The awakening, you know, it's a theological thing. Like people get converted, like they see the light and they're converted. It And then and then so uh so so you do what you can, you train, you do the hackathon, you do the offsite, you do the hero move, right? At the end of all this process, if that person has not you do internal competition, you show you show how one team outpaces the other teams and you show them it is possible because look at the other team. But if at the end of all that they fall under the codeg the category that we call resists resist and they defy the new religion, then you don't have a choice, unfortunately. Yeah.

11:15Unidentified speaker (speaker_2): Right. But I I I think it's about finding ways to either remove their concerns,

11:21Unidentified speaker (UNKNOWN): um,

11:22Unidentified speaker (speaker_2): which in the example that Liran gave, uh, the engineer in question was pointing to the legacy and saying, This is what's preventing us from doing it, or preventing me from feeling comfortable. And it was so that the the the CTO removed that concern by essentially rewriting the legacy uh Or it's about making or uh showing that employee how AI can improve their life, their their work life. Um and there are various methods of doing that as well. And everybody has something in their work that they want to improve. Uh we we naturally feel better Go when we're more productive, when we when we're able to do things faster and more efficiently.

12:01Unidentified speaker (speaker_0): I definitely subscribe. to that. I believe in this so much. I think the productivity aspect, right? I think part of our roles becomes more fulfilling because we get to do the things that we enjoy doing. Um but there is another part of the the paper. You describe a data scientist leaving the profession after eight years because he felt AI had surpassed him. considerably. And

12:29Unidentified speaker (UNKNOWN): it

12:29Unidentified speaker (speaker_0): makes me think: what do you recommend people who feel exactly the same as that data scientists do if they feel like AI is surpassing them?

12:39Unidentified speaker (speaker_1): Eventually we are becoming managers. The new engineering work is orchestration and management of agentic processes. So you are becoming the manager.

12:50Unidentified speaker (UNKNOWN): You're

12:50Unidentified speaker (speaker_1): no longer. the developer, the individual developer.

12:54Unidentified speaker (UNKNOWN): And

12:55Unidentified speaker (speaker_1): if you don't subscribe to that course of work, then maybe it's right for you to do other things. But uh people that have building engineering in their soul, in their heart, I think, should love it. I personally love it. I find it more interesting because you can do so much, your productivity, like becoming a 10x or 20x of what you do is fascinating for me. If you just like the like if you enjoy the the writing of the code, then that world is gone. Uh

13:26Unidentified speaker (UNKNOWN): if

13:27Unidentified speaker (speaker_1): you want to build big things, it's better times than ever.

13:30Unidentified speaker (speaker_0): A lot of software engineering, if not all of software engineering, has become a commodity play now, right? I think everyone effectively can use these tools such as cursors. Cursor and Core Co to write software. And it's accelerating at such a fast pace. Our CTO said what can be done today is very different to what can be done three months ago and what could be done in three months, uh three months from now.

13:53Unidentified speaker (UNKNOWN): Um

13:54Unidentified speaker (speaker_0): what about those type of roles? Are we seeing a world where a lot of what is being done across engineering will become pretty much taken over by AI? You describe the role is orchestration and management. What do you think the software engineers will end up doing? We we index a lot on words such as taste and judgment and first principles thinking. What does that mean a year from now, potentially?

14:17Unidentified speaker (speaker_1): Uh I uh we outline in the paper three main pillars. Each organization needs to have the four food, it's called probably the four chiefs.

14:26Unidentified speaker (UNKNOWN): Okay.

14:27Unidentified speaker (speaker_1): One is uh design. Okay. So let's start with chief architect. Okay, the architecture of the product. You still here you need to have your voice of the architecture because if you if you have you'll have engineers, teams using multiple tools, each tool will use a different database, would With use using different uh kinds of architecture, different cloud architecture. That can be very, very risky, you know. just saying AI will solve everything. So one group is the architecture group, okay? The CTO, the chief architect, and the people around it. Small group, but very, very important. And they supervise the architectural decision. Then you have the chief security. That also is a human decision. You know, AI can write lots of code, but That ownership and accountability is not something that you can just hand over to AI. You need people that will sign off the architecture and make sure that you know compliance and security still prevails. Then you have the chief uh chief design. That's also design, I mean graphical design. If you noticed, all the products look the same now. They all have these small cat's eyebrows on the top, they have these squares with the line on the side. I can just see you know claud-based clod-based output immediately. And so what's your unique? You know, you want to have character to your product. By the way, it's it's all both graphical design, but sometimes also

15:47conversational character. You know, do you relate to the agent that talks to you? You know, so the the character in the product will only come from human because otherwise it's all the same and you're not differentiated. And the fourth is the chief product. And you what used to be the old product organization, but it's also being transformized, transformed to something new. Products today build betters. They they they they issue PRs, right? They they're becoming part of the team, but they set the direction of the company to what you want to build, what you think you can sell, you know, what your customer needs. So I think these four pillars are human pillars. They're not necessarily necessarily very large teams. but uh people but they need humans uh for sure.

16:30Unidentified speaker (speaker_0): I'm interested to dive a little bit on the investment side. Adam, you you just

16:35Unidentified speaker (UNKNOWN): you've

16:35Unidentified speaker (speaker_0): been doing this for a very long time and you've you've had a huge success as an investor. Um you work with Laran a lot, you've written his paper together. But how does this philosophy and this way of thinking change the way you think about investing?

16:49Unidentified speaker (UNKNOWN): And

16:49Unidentified speaker (speaker_0): I'm interested interesting here, interesting to double click on this one because a lot of our listeners are founders who are thinking about building

16:56Unidentified speaker (UNKNOWN): and

16:56Unidentified speaker (speaker_0): what would attract the attention of in investors in the space. How do you look at investment today? How has this changed your philosophy around how you look at deal flow, where you spend your time with the type of businesses you want to invest in?

17:12Unidentified speaker (speaker_2): So it's less the type of business. and more what I expect from the company. And what I mean by that is I don't want to go into the various sectors and segments that I think may or may not be interesting. It's all moving so fast. It's what what I think is relevant today could be irrelevant in a few weeks. Um but the notion of of product building and what the what the customer wants and and even uh focus has changed a lot. I used to say in the world of SaaS, uh product was paramount. Uh you would stay within your lane, you would focus on a certain buyer and user within the organization who was senior but not too senior, um, who would champion your product, you would change their organization from in terms of uh what your product delivered, and you'd you stay there for years. That's no longer the case. Um but the good news. is that you don't longer have to stay in your lane. You can develop so quickly adjacent products and expand, kind of think of it kind of horizontally

18:16into into adjacent fields that perhaps there was a competitor there or perhaps there was uh you know a larger player um and you can do so overnight. Uh what makes it more interesting, of course, is that those those those products themselves are at this point typically AI native and kind of built as serving both users and agents so that it actually serves the customer in the end. And so what that means is that we're looking for companies that can move very quickly, that can find that initial wedge uh through product and focus, but can very quickly then uh leverage that to expand. Um, 'cause my kind of view is that if you don't do it, your your competitors doing it. Um and so that speed and execution uh is is what we're we're ultimately looking for. Uh and and not everybody is is up to the task. Um, whether it's uh whether it's the the agentic engineering aspect of it or or or just the pace.

19:16Unidentified speaker (speaker_0): Yeah, the reason I asked that question is in the past, in the SaaS era or the cloud era, the the investment tur timescales cycles tend to be eight to ten years. Typically, you see some sort of event or or str exit for a business uh that that an investor wants to see an ROI on to

19:38Unidentified speaker (UNKNOWN): uh um

19:39Unidentified speaker (speaker_0): to that effect. But we're seeing those time scales shrink significantly now with AI. And the article discusses or at least asks the reader to imagine two engineers with AI outpacing a team of 50. From an investment standpoint, I want to get your opinion on have you seen this agentic awareness? awakening

20:04Unidentified speaker (UNKNOWN): uh

20:04Unidentified speaker (speaker_0): theory or or idea? You see that in your investment portfolio of companies you invested in, the ones that are applying this thinking have significant growth in their revenue and their product market fare and and and how their ARR has grown.

20:20Unidentified speaker (speaker_2): Yeah, I mean several of the companies that we spoke with are are are in my portfolio or in or in the wider bestmar portfolio. Um that's part of the gulf that we saw being, you know, widening between the ones that it understood what is what is possible and pushing the limits, and the others that were still waiting for directions from somewhere or or or some kind of forcing function. Um and and and we see it. And if you pay attention, you'll also see the leading startups these days are releasing products. at uh at a furious pace. Uh, and they're not just features. We're talking about new products, products that uh perhaps displace a supplier or products that uh displace uh uh a company that was supposed to be you know doing something very different. Um and and and so that that widens the opportunity, but I think that's also what customers are looking for. I I think they're looking for fewer vendors to deliver more impact. So

21:19Unidentified speaker (speaker_0): I want to take this back to you. So you in the paper you describe working with AI agents as addictive and draining, um, with people supervising more than they can fully understand.

21:30Unidentified speaker (UNKNOWN): The

21:30Unidentified speaker (speaker_0): question I have is if AI is doing more of the work, why can the human

21:37Unidentified speaker (UNKNOWN): I

21:37Unidentified speaker (speaker_0): guess

21:38Unidentified speaker (UNKNOWN): end

21:38Unidentified speaker (speaker_0): up more exhausted?

21:40Unidentified speaker (speaker_1): It's a good question. You have to experience it to see that. And you're waiting for the answer. And what happens is that the longer the cycles become, it can take a few minutes It's for the task to finish. So you put a few more, a few more coins, right? And then they start. popping back and they ask you back for feedback. And it's like you have these employees that all the time coming to you, give me another thing to do, give me another thing to do. And you and you just give them another coin and another coin, another coin. So I don't know why, but it's psychologically is the most thing I've done in my life. So I can explain it better than that. But it's anyone that's been that until like 2 a.m., 3 a.m. in the morning, still feeling feeding his agents, will probably relate to what I'm saying.

22:29Unidentified speaker (speaker_2): I'll tell you, I I can relate to it having uh been around, you know, when when email came out or uh or instant messaging. You know, at first it was like, wow, I don't need to wait for uh for a fax or for a meeting. Uh but after everybody was on email and you were kind of shooting these things back and forth, you realize like the faster I respond, the faster I'm gonna get a response, the more work I've created for myself. And you never get out of it. Um and so I think there's also that aspect. Um it's just you literally cannot send an email or a or an instant message uh without without getting a response, which uh depending on your line of business just means you've just created more work for yourself. Is

23:06Unidentified speaker (speaker_0): it is a little bit like gambling, right? Because to your point, it gets addictive, but at the same time there's the token spend and a lot of it. And now I'm I'm getting that message from Claude: you've reached 75%, your team are requesting more usage spend. And where does it end? Right? How do you even know where the value is? Because it's it's exciting and you're building and you can automate so many things, but you also get trapped in this world of all these age agentic tools around you. Does it get exhausting to the point that you don't know you can't really attach the value to where the real value is?

23:45Unidentified speaker (speaker_1): You certainly build things. There's like software scroll where you build things that you don't necessarily need. The agent can decide to build much more than you originally Really wanted, and you just unknowingly build a whole new feature that these things happen and to that's why you you try to have the judgments. I have a section there that we talk about the difference between uh I said, uh were you vibe coding or were you agentic engineer? Were you engineering? Like were you vibing or were you were you were engineering? Because I think vibe coding is the term that uh Karpatri, I think, coined. coin. It's more like just someone going and building his own app with a few prompts, right? Without knowing what he's doing much. And I think agent engineering is also understanding what you're doing. Ask not necessarily reading reading the code, but asking the right questions and making sure that what you built is the correct architecture based on the what the company needs to do. And also, you know, token maxing, because uh I think initially we saw companies saying, just use, you know, use as much as you can. because they wanted to

24:51get people to get to understand what they can do because people were holding themselves. They were not even using the maximum of the technology because maybe they were worried about budget or you know, not sure what's going to do. So they pushed the aggressive company pushed people to use even more and they didn't care about budget at all. Once they started start getting the bills, uh now they're calibrated. calibrating. Okay, okay, now let's also check the quality, the accuracy, the the need for what you people are building, and maybe do model routers where we send simple tasks to less expensive models, or we start saying don't, you know, you're overspending here and let's let's focus that. But that's the second phase, not the first phase.

25:34Unidentified speaker (speaker_0): I have a question for both of you and I'm keen to get you your uh

25:37Unidentified speaker (UNKNOWN): like

25:38Unidentified speaker (speaker_0): the first thing that comes to your mind, something you've built that has been, I guess, really profound, has helped you become far more productive, has helped you in your day to day. Could be something personal, could be something professional, but just something that you've personally built for yourself and you've walked away from it and thought this has been a real game changer for me. Whoever wants to go first, but I'm I'm keen to get your thoughts on that.

26:01Unidentified speaker (UNKNOWN): I

26:01Unidentified speaker (speaker_1): build built a company Britain. brain. I uh I decided that the business uh the BI systems were antique. And after using so many years of you know power BIs and tableaus and the likes of these, I said if code can agents can write code from scratch and do such amazing things, they can probably analyze my data without having to go through a middleware tier. And I build a full agentic business intelligence platform that I first use in my company and eventually turn it into a product. So this has been my I think this is the holy grail because you see lots of solutions that are basically replacing simple tasks. This is the workflow agents. And then you have this brain concept where I want to make better decisions. and so that's what that's what intrigued me.

26:54Unidentified speaker (UNKNOWN): I don't

26:54Unidentified speaker (speaker_2): have I don't have a good answer unfortunately. Uh uh a variety of excuses, but I I think in the world of of uh of of venture capital, at least when when you know you lead investments and sit on boards, there's uh types of of things we w need to build are not um they're not groundbreaking. Uh Um there are various types of uh sourcing tools. and uh and and and new types of dashboards, but ultimately uh so much of what we do is also just working with with uh with CEOs and founders.

27:24Unidentified speaker (speaker_0): If you had something you wanted to build, because deal flow is such an important piece of what you do as an investor and you look at great deals, but is there something in your day-to-day that you feel uh this should be automated? Aaron

27:37Unidentified speaker (UNKNOWN): Powell

27:38Unidentified speaker (speaker_2): I I think what needs to be automated, or one of our biggest challenges is information share. You know, in our world, each individual investor, whether it's a partner or somebody else, uh, is being exposed to different companies, um, working with different individuals in their portfolio and and and learning. And there's a lot of information processing. So there's a lot of information that's that's that's lost and and not sure. shared. Um that's really one of the advantages of a large fund in particular, or also a fund with a lot of so with a large portfolio and and and individual histories. Those trying to learn from one another is um is what really what we bring again to to the rest of the partnership, but then also to that portfolio. Uh so I have no doubt that there are ways to improve that. Um Similar to what we did with uh Agentic Uh Awakening, you know, in terms of gathering information from from portfolio companies and trying to kind of passing it from from from those in the node to those that are still or just learning, that's what I would want to find ways to to automate uh um much much better than we currently do.

28:47Unidentified speaker (speaker_0): Guys, I want to go back to Agentic Agenic Awakening with one more final follow up question to the to the last I guess topic we had discussed around the two engineers outpacing a team of fifty.

29:01Unidentified speaker (UNKNOWN): I

29:01Unidentified speaker (speaker_0): guess the question would be within

29:03Unidentified speaker (UNKNOWN): it's

29:03Unidentified speaker (speaker_0): hard to put timescale to these things, but let's say two, three, maybe five years time. Could do you believe we could see a world where a handful of people with AI

29:13Unidentified speaker (UNKNOWN): to

29:13Unidentified speaker (speaker_0): its optimum capacity. could realistically put a company with hundreds of employees out of business.

29:19Unidentified speaker (speaker_1): I'm convinced. I'm totally convinced of that.

29:23Unidentified speaker (UNKNOWN): Uh

29:24Unidentified speaker (speaker_1): I think access to agentic resource I think you'll need resources because eventually to create big, large projects you'll need hundreds, even sometimes thousands of agents running.

29:37Unidentified speaker (UNKNOWN): And

29:37Unidentified speaker (speaker_1): you need to have the capability to orchestrate and to run them efficiently. But if you have these, then you basically I I I see today already this two two against fifty happening. And I think you're gonna see it even at a larger scale.

29:53Unidentified speaker (speaker_2): Yeah, I agree. It's it it it it's it's it's happening. Um it's not always felt in the form of a company um vanishing. Um You know, oftentimes it's taking a part of some business from a hundred different companies, which maybe they don't all feel as as big an impact, but over time it grows. But I also do with the Ron that at the end of the day, that new small company that's only one or two individuals, in order to maintain it and to grow, they have to add people. It will be different. people, it will be uh structured differently. Um but I I I do think it's very hard to grow and maintain um that business and preserve it without without adding humans.

30:38Unidentified speaker (speaker_0): Now we end every single episode with asking our guests for a book recommendation. Part of why we called it Inside the Silicon Mind, mine, getting into the mind of our guests, our founders, our vest investors, and have them recommend a book Now, before I ask this question, because this particular episode is so much about Agentic Awakening as a paper, as a as a as a playbook,

31:00Unidentified speaker (UNKNOWN): I

31:00Unidentified speaker (speaker_0): guess what for both of you, what made you come together and and at this study?

31:07Unidentified speaker (UNKNOWN): You know,

31:08Unidentified speaker (speaker_0): why was that important to you?

31:09Unidentified speaker (speaker_2): I've known the Liran Iran for many years now, so so it wasn't about uh uh you know it he was actually the first person I approached. Uh so we I I I knew this was a good match um from the get go. And ultimately, you know, we're we we wanted to help the the industry. That's uh it it might sound uh um cliche, but um but we know that that's that's part of our role is uh we're where w when you're at the forefront, when you're seeing things, you you know, you can pass that information on and obviously it requires sometimes a lot more work than we than we expect. And in this case it did. It took a lot more time uh and work than we imagined. Um but we're very very pleased with the result.

31:50Unidentified speaker (speaker_0): Daron what about you? What what what you know Adam comes along with this idea and and this it it's an incredible pie piece of document. I mean the the the playbook is something that I um

32:02Unidentified speaker (UNKNOWN): would

32:02Unidentified speaker (speaker_0): recommend anyone to read, all our all our audience. But for you, what what was so important and in fact impactful about this particular playbook?

32:09Unidentified speaker (speaker_1): When Alan approached me about it, the the the main thing it was that I found it interesting. So it intrigued me. So uh and because I had my personal experience and I was already talking to several companies. So I've already been consulting to a few companies or friends in different companies that asked. me about that. But opportunity really through the pessimist portfolio and beyond to get exposed to multiple teams of the front and to see what they're doing, that was, you know, and also at the timing, because you know, this can be in six months, the story could be different. Maybe we need the new version, right? It moves so fast. But getting this in the at this critical point in time. six months after Clod Code version two came out and uh Opus four point five, really really the turning point was October, November twenty-five. And we'd be able to conduct that uh you know five months after that. And then also, by the way, I had another touch point with some of these companies following the publication to see how they proceed how they they proceeded since then. And so for me it was uh I I really f it was intriguing. I I was interested in that more than just the work of doing that. By the way, we didn't even

33:19know that we're gonna publish a playbook. But the the after about a dozen of these, I already knew what they're gonna say. And each one paved their own road, but even but even so they went into the same lanes. So that then I said okay there is something here that uh you know repeats itself and so here here's the story. And then we started writing it. So

33:41Unidentified speaker (speaker_2): but it's very important to understand that the these companies didn't have the structure in mind. In other words, they didn't have really their own playbook. They were just replaying what they had done. Um and that's the patterns that Liran saw that r ended up being the the playbook that could be repeated by others. Yeah,

33:56Unidentified speaker (speaker_0): I mean like I said from the very start, the the Having read it, the it has had a profound impact in terms of how I think about our day-to-day in our business. The word awakening is so perfectly chosen for this playbook. It is an awakening. You think about the different studies, you think about where the market is heading, and and it's a conversation that probably needs to be had in all industries, in all sectors, not just in technology and engineering. I think there's going to be an impact. across it is agnostic, right? It goes everywhere. Um, but guys, I really have loved and really enjoyed having you on the show. We always wrap up with a question around a book recommendation. We've got a section on our website where we attach this book was recommended by this get guest. used. And I we'd would love to have one from you. So maybe start off with Leran. I'd love to get a book. a recommendation for you outside of Agentic Awakening, one that you would highly recommend.

34:55Unidentified speaker (speaker_1): I think when I wrote it I thought about Accidental Empires.

35:00Unidentified speaker (UNKNOWN): You

35:00Unidentified speaker (speaker_1): know that book that talks about the origins of the tech industry. So I had that book in my mind when I wrote it. Uh because I But I was thinking almost like we are this is the new agent this is the new Microsoft building here. This is the new you know. So these are the accidental empires of the AI age. So that was what I had in mind. This is the nonfiction, like the the the uh the other books, but I think it's good. This one is to be could be good.

35:30Unidentified speaker (speaker_0): Love it. Thank you, Laran. I don't know, what about you? What would be a book recommendation from you? Uh you know, one of my biggest concerns with AI. is the over reliance of AI to the point where people become.

35:44Unidentified speaker (UNKNOWN): Um I

35:45Unidentified speaker (speaker_0): I know I don't know if anyone's ever done a study on this, but I I the worry is what if we become so dependent on AI to the people that becomes becomes a an over No dependence independence that will affect people's cognitive ability, right? Everything is screenshot this, send it. to AI, respond to this, right? So I think reading and studying and and keeping the mind going is so important. And that's why we ask for a book recommendation is to keep people expanding their minds, expanding their intelligence. So this is why it could be genuinely anything, fiction, nonfiction, truly anything.

36:21Unidentified speaker (UNKNOWN): Yeah,

36:22Unidentified speaker (speaker_2): I mean at the end of the day, I I I have I have a lot of book recommendations. Um and uh I I share your your concern. Um I I I don't spend as much time reading books about business and technology. That's like that's the first thing I'll say. And so um when I think about uh our our daily interactions with computers and with agents, I realize, you know, it what's gonna Going become to an imm most important skill is dealing with people, dealing with humans. Uh because people will find ways to actually avoid humans from an early age. Uh, you can probably already see it in the young uh population. So there are various types of books that uh, you know, I recommend people to understand both themselves but also others that they have a hard time. Um, the one that I've recommended um uh previously is um the the The Laws of Human Nature uh by Robert Green. Greene, Um which gets deep into the psychology of various types of individuals you're likely to meet throughout your life,

37:23especially kind of leaders and people who have of great influence. uh um um how um how important it is like not necessarily go with the crowd um i'm still very moved by a book called um

37:45Unidentified speaker (UNKNOWN): um

37:45Unidentified speaker (speaker_2): true believer by uh I think it's Eric from sorry no Eric uh Hoffer um which is also kind of uh an eye-opening book in terms of mass movements and how everybody can be wrong except you.

38:02Unidentified speaker (speaker_0): Fascinating There's something you said there that caught my attention. The the curve is shifting and it and it shifted with the internet era and then the mobile era and the cloud era. Now we have AI. And social media definitely, I think you probably can attest to this and you've seen it, right? It has pushed people slightly away where people are relied relying on these phones as a way to communicate as opposed to human-to-human interactions. And AI is now pushing us to a world where we're interacting with agents and and and robots, right? Um whilst this episode has come to an end with the book recommendations, just curious. I mean, is this something that that you think about as a concern? And maybe this is part two where we come back and and talk about this, but

38:48Unidentified speaker (UNKNOWN): is

38:49Unidentified speaker (speaker_0): that a worry? Is that a worry where we are becoming so over reliant on technology and now agents and and potentially eventually robots that human interactions become less and less

39:00Unidentified speaker (speaker_1): I don't know why, but I think we're always concerned. I think if you we have the same thing with mobile phones and with social networks, right? And people ad adapt, right, to new technology and people people change, but they th there's there there's also bigger gaps. big advantages in the new technology. It's difficult for us to invent to to to understand how things would appear with the new technology from where we are right now. I think my my bigger concern is about the concentration of wealth and uh concentration of uh uh uh productivity in small grip groups of people and the inclusion I think is a big problem. If you see how much concentration around the two new the the the expected the IPOs of uh anthropic and OpenAI. And it also goes back, you know, if two people can do fifty works of fifty or a hundred or a thousand, what would other people do and and uh and how it would impact uh the society. I think for me this is uh this is this is a big thing. And we have never seen this. I think it's we're talking about the intelligence revolution or the data revolution, the size of the industrial revolution. If you want another book on that, is the as Noah Harari's thought about the 21st, 21st century. I think he has some

40:20interesting topics there. And I think we haven't seen this the start thought of what that could be to do to our society. And I think uh that could be a challenge bigger than the technology, I think.

40:33Unidentified speaker (speaker_0): I appreciate the uh the thoughts and a great way to end the episode. Perhaps leave it with a cliffhanger for part two. Um but Loran Adam it was actually actually uh uh one of my you know looking forward to this episode a lot because we are moving in such a such a fast rate in this world of AI. So thank you so much for bringing Agentic Awakening to the world and and putting this pip this this playbook together and thank you for being on the show.

41:02Unidentified speaker (speaker_1): Thank you. Thank you so much.

41:04Unidentified speaker (speaker_2): Thank you

This page was produced automatically: the audio was transcribed with nvidia/parakeet-tdt-0.6b-v3+sortformer-v2.1 and interpreted by Qwen3.6-35B-A3B-UD-IQ3_S.gguf. Transcription and interpretation can contain errors; always confirm against the original episode. Nothing here is investment advice.
All insights