Brian Leonard

The hottest role in AI startups right now isn’t Forward Deployed Engineers. It isn't GTM Engineers. It’s Deployment Strategists. Decagon calls it an “Agent Product Manager.” Harvey calls it a “Solutions Architect.” Palantir Technologies has had versions of this role for years. And the salaries are climbing fast: - Decagon: $200k–$285k - Palantir Technologies: $120k–$200k - Figma: $150k–$260k - Ramp: $100k–$180k - Harvey: $190k–$260k So who are these people? They are usually pseudo-technical -- CS or engineering majors, or folks with technical work experience. Many come from 2 years in consulting, IB, or PE, then jump into startups to get their hands dirty. They are young, hungry, polished, and comfortable being in front of customers. What do they actually do? They make sure enterprise AI deployments succeed. A $100k+ deal does not survive on a nice pitch or a self-serve onboarding flow. It survives if the customer sees value in the pilot. That means: - Embedding directly with the customer - Designing prompt logic for specific workflows - Working with engineering to align integrations and data flow - Helping exec teams define their AI roadmap - Running feedback loops into product and GTM Why does this role matter so much? Because enterprise AI is messy. Integrations, data transfer, and adoption make or break a deal. Most buyers are using AI for the first time, and each has unique workflows. Deployment Strategists bridge that gap. They own the outcome. They are accountable for making pilots successful, which often means millions in revenue down the line. At Conversion, Sam Bochner has been leading this work for us. We are now thinking about scaling it into a full team. Because a few successful pilots can fund an entire department, and the cost of failed deployments is too high to ignore. Is this just a rebrand of customer success? Not really. Success is about answering tickets and renewals. Deployment Strategy is about going deep with a few enterprise accounts, extracting maximum value, and ensuring the pilot closes into a multi-year contract. Call it Agent PM, Solutions Architect, or Deployment Strategist. Whatever the title, this is becoming one of the most important roles in AI SaaS.

646

91 Comments

It’s a bit surprising to see people in my feed this morning say OpenAI’s AgentKit “kills” a bunch of construction tech startups. It's incomplete and doesn't capture the whole picture. Yes, generic, horizontal-wrapper startups that just sit on top of foundation models are in trouble. But industry AI agents built for very specific, complex workflows, who understand and care about their customers are going to thrive as AI adoption accelerates. I played with Agent Kit yesterday for a simple task and here’s my take: 1. Agent kits are impressive but shallow without depth. To make them truly useful, you still need a quasi-technical team to build, monitor, and maintain them. In industries like construction, where margins are razor-thin and overhead has to stay lean, that’s a tough sell. Most GCs and subcontractors aren’t about to staff up a new AI department while trying to keep projects on budget. 2. Vertical Industries are highly complex and for good reason.  In construction for example: Permitting, liability coverage, inspections, safety, pay schedule and regulatory compliance all add layers of checks and balances. These workflows exist to protect businesses and keep projects safe and sustainably capitalized. You can make AgentKit work for you, you're just going to need people to do it.  And every construction company I have ever worked with wants to hire more builders and field professionals, experts in their domain, not more people in the back office. 3. Deep, domain specific challenges are where the real value is. The vertical winners won’t just bolt agents onto a foundation model, they’ll build depth with industry specific workflows. Nobody wants vendor lock-in, so they’ll be model agnostic, swapping in and out whichever LLM performs best for the task. They’ll integrate seamlessly with legacy players and every other tool the industry already depends on. They’ll own the ETL process, cleaning and structuring messy, fragmented data. They’ll capture proprietary workflows and the tribal knowledge unique to the vertical. And they’ll deliver enterprise grade privacy and transparency. 4. The organizational behaviors matter. In construction and many other verticals, middle managers often optimize for their resume and their next job while leadership optimizes for profit margins and retention. Sometimes it can be in direct tension of each other, one group wants experimental automation engineers, the other wants tools to drive productivity without new headcount. For example, software in general doesn't require additional health insurance. So yes, AgentKit and whatever comes next will eat the easy stuff like document search or report templates. But the real, hard, high value work will always go to the companies who know their vertical inside out and build for it.  Foundational models are democratizing access. Vertical AI is where the winners get created.  We’ve seen this in past technological shifts. 

209

17 Comments

Lovable just raised $200M at a $1.8B valuation. We spent 48 hours in Stockholm with the team and saw how startups will be built in 2025: 1/ WHEN IT WORKS, IT PULLS Their story is insane. They launched GPT Engineer. It exploded 🤯 50k GitHub stars in days! Lovable was born to build the UI on top. Same thing happened with Replit. It took them 8 years to reach $10M ARR. Then, they found “vibe coding”. They dropped everything else. And hit $100M ARR in under a year. In AI, PMF doesn’t creep in. It hits like a wave. 2/ MARKETS WITH 2 LEADERS MATTER Two leaders = duopoly = closed market. CRM: Salesforce + Hubspot. eCommerce: Shopify + WooCommerce Lovable is betting on becoming the 1st player. That’s where the door is. 3/ TIME IS RELATIVE Everyone’s working weekends. Not out of pressure, but alignment. When everyone’s pulling in the same direction, speed compounds. 1 year at Lovable = 10 years in SaaS. 4/ YOUNG TALENTS I used to believe experience was everything... Turns out, some of the most brilliant builders hit their stride in their early 20s. Cursor’s founders graduated in 2022. Lovable’s average age looked like 24 😅 I felt like a dinosaur. 5/ RULES ARE MEANT TO BE BROKEN I asked Fabian, co-founder & CTO, who their persona was. He said: “anyone who wants to build an app.” At first, I didn't get it. At June, we narrowed down to find PMF. You don't always need to niche down in AI. Lovable is replacing Webflow, Framer, and Bubble... at the same time. 6/ GTM IS OVERRATED WHEN PULL IS STRONG Lovable hired Elena Verna. One of the best growth leaders out there. But traction came before. Lovable go-to-market is its community. People use it, share it, pay for it. GTM is here to accelerate things. It puts fuel on the fire 🔥 7/ THE OPPORTUNITY ISN’T OBVIOUS “Vibe coding” was dismissed because of retention concerns. But I’ve seen the data, and trust me, it’s wrong. This opportunity wasn't obvious to most people. And it helped them. 8/ FUNDRAISING IS SECONDARY This raise didn’t happen on a roadshow!! It came from a trusted conversation. Focus on building, and the right money finds you. 9/ COST ISN’T A PROBLEM In SaaS, AWS gave you free credits. In AI, LLM providers do the same 😉 If your product is great, infra will follow. 10/ THINK FROM SCRATCH The"code on the left, app on the right" UI was their invention. I asked Fabian how he came up with it. He said “it felt like the best UI”. Great products invent their own interface. Others copy it later. (Tinder for X, Uber for Y) ______ These 48 hours were a glimpse into what the next wave of startups might look like. Thanks again to Anton and Fabian for having us. For being so open, and the cold swim 🥶 You're building a generational European company 🇪🇺 And to Elena for making the intro. You're one of the greatest 💜

527

34 Comments

Everyone laughed when YC gave us this advice. Turns out, it was genius. "The idiot and the genius usually end up saying the same thing. The midwit is the one who gets lost." Take feature decisions as an example. Idiot: “Just ask the customer.” Midwit: “Let’s do a 20-slide deck, market analysis, user surveys, competitive benchmarking, and pick the perfect framework.” Genius: “Just ask the customer.” When you start learning, overthinking feels like progress. You want to use everything you’ve picked up. But the more you pile on, the further you drift from the only signal that really matters: what your customer wants. YC calls this “midwit behavior.” The hard part isn’t complexity. It’s having the restraint to keep things simple, even when you know enough to make them complicated. The best founders I know don’t sound sophisticated. They sound obvious. Because keeping things simple takes more discipline than making them complex.

351

38 Comments

Lots of people made jokes when Atlassian bought The Browser Company a week ago. But I think there are some legit product strategy reasons... I'm a little late to the party on this. Tech Twitter had a field day with this news. It was highly entertaining, but mostly poked fun at the deal. Atlassian is a legendary SaaS company, so I analyzed this deal under the assumption that the people making it are smart, thoughtful, and deliberate. I think there are 5 legit possible reasons for the deal: Agents Are Existential To Atlassian's Core Business JIRA and Confluence make up the majority of the revenue. One is workflow management, the other is content management. Both are in the crosshairs of agents. There Are New "Entry Points" For AI Products ChatGPT, Claude, Cursor, etc have established new entry points for workers. They are starting their work more and more in those tools vs traditional SaaS tools. The browser is another potential entry point with no clear winner in the AI era yet. B2B Browsers May Make Sense In An AI World We mostly use a desktop browser for work. A lot of our personal consumption has moved to mobile. Existing desktop browsers were not built for work. AI may create the shift and opportunity to create this. Atlassian Needs Moonshots To Keep Growing Atlassian's historical playbook has been to build/buy and then sell adjacent products to JIRA and Confluence. That strategy may be running out of fumes. What adjacencies exist that could maintain growth when the baseline is $4B. They may need a new S-curve. They're Losing The Next Generation Of Companies First pointed out by Varadh Jain, the next generation of companies don't start with Atlassian products. They are choosing Linear over JIRA, Notion over Confluence, Slack over Hipchat (RIP). Dia and Arc are both products that had followings of the next generation. All of this is interesting to me as product strategy has shifted in the world of AI. Full Post Here -> https://lnkd.in/gknesaZB

111

31 Comments

OpenAI’s Product Lead has a blunt warning: most AI startups aren’t companies, they’re expensive demos 😬 Chegg waited too long and lost 90 percent of its value. Jasper raised at a $1.2B valuation but collapsed when costs outpaced revenue. Duolingo bolted on AI the wrong way and lost user trust overnight. AI is unforgiving. Features get copied in weeks. Costs scale faster than revenue. The market does not give second chances. Miqdad Jaffer, Product Lead at OpenAI, lays out the playbook for founders who want to survive: ▪️ Moats come from data, distribution, and trust ▪️ Pricing is strategy, not an afterthought ▪️ Every feature must scale economically ▪️ Differentiation has to go deeper than “adding AI” 📖 I am sharing the full guide and frameworks here: https://lnkd.in/exXTNhdR AI Resources you might find helpful: • OpenAI’s New Playbook for Product-Market Fit in AI Startups https://lnkd.in/dy-HgVtX • 2,500+ Angel Investors Backing AI Startups - https://lnkd.in/djjN-TD2 • 50 AI Agent Startup Ideas for 2025 - https://lnkd.in/dy9E-MCn • AI Co-Pilots for Startups & VCs - https://lnkd.in/dSX3ss2q • The Agentic Revolution - https://lnkd.in/dadFm25a • 118 AI Communication Agent Use Cases - https://lnkd.in/dr9cM5BT

492

105 Comments

Here are 5 NEW companies paying $100k-$400k for FULLY-REMOTE devs: 1. Checkly 2. Lumos 3. Attio 4. Wave Mobile Money 5. Svix They get it. - They don't have the FAANG brand. - They even pay less than FAANGs in the US. Yet, they're able to easily hire and retain FAANG-level talent. Want to know about other companies in this category? EuroTopTechJobs.com lists thousand jobs from 100+ such companies. It also tracks thousand top-paying onsite, hybrid and country-remote jobs. In Europe. It also provides data regarding compensation and saving rates. Check it out if it sounds interesting 🙂 ♻️ Like/Comment/Share if you want to help people in your network.

257

26 Comments

Last Thursday, Figma IPO'd at $50B with a 158% first-day pop. Dylan Field still owns majority control after 13 years. In an era where many founders treat startups as a means to an end, this feels like watching someone see their vision through - no matter how long it takes. Some background: It was September 2022. Adobe announces they're buying Figma for $20B. $20 billion. For a 10-year-old design tool. Everyone assumed it was done. 15 months later, regulators kill the deal. Figma gets a $1B breakup fee. Many companies would've played it safe with that windfall. He quadruples R&D to $751M. Ships Make, Sites, and Dev Mode. Goes all-in on product while everyone else battens down the hatches. Takes the company public at $50 billion (2.5x Adobe's offer). That's what happens when you refuse to abandon your mission. (13 years of refusing) I keep thinking about what made Field stick with it for 13 years. It's not (just) intelligence or opportunity. It's something I saw in his IPO letter: "Expect us to take big swings. We will bias toward long-term compounding over quarterly share-price appreciation." This from the guy who just delivered a $50B IPO after regulators killed his $20B exit. Look at the 5 moves Field made after the Adobe deal collapsed: 1. He ate the rejection like breakfast. Regulators kill the Adobe deal? Use their $1B to build something worth $50B. Turn a forced “no” into the best “yes” of your career. 2. He kept his team when everyone said cut. After Adobe collapsed, conventional wisdom would say downsize, especially in the hiring environment at the time. But Field avoided forced layoffs. Most stayed. This means the same core team that built a $20B company built a $50B one. 3. He treated the breakup fee like jet fuel. Many companies would've hoarded that $1B. Figma quadrupled R&D to $751M. Shipped Make, Sites, and Dev Mode while everyone thought they were damaged goods. 4. He built AI into the product roadmap. 2023-2024 was peak "pivot to AI or die" hysteria. Field shipped AI-native products (Make, Sites, Dev Mode) that actually solve design problems, not just chase buzzwords. 5. He built the company to be un-acquirable from day one. Dual-class shares. Majority voting control. The legal infrastructure to say no. When Adobe finally got through, it was because Field chose to let them. — Adobe offered Figma a dream acquisition of $20B. Regulators said "no." So he bet their $1B consolation prize on going bigger - and built something worth $50B instead. $20B in 2022. $50B in 2025. The "setback" was actually a massive gift.

311

43 Comments

SaaS monetizes features. Rails monetize mistakes. The main objection to Rails: "Isn't this just network effects?" No. Rails don't create network effects. They create learning effects. Every time work flows through the rail, it doesn't only deliver an outcome, it makes the rail smarter. SaaS scaled revenue. Consulting scaled judgment. Rails scale error correction. While LegacyCos bolt AI agents onto existing workflows, NewCos build differently. They know you can't buy five years of resolved edge cases. That compound learning becomes the moat. Learning Effects > Network Effects Classic SaaS defensibility came from network effects. Slack gains utility with more teammates. But SaaS compounding is bounded. Each instance learns locally. Operating Rails compound differently. Every GitHub Copilot rejection, every Stripe fraud flag, every CrowdStrike attack blocked makes the entire network smarter. Expensify example: When you correct a miscategorized expense, it only improves your workspace, despite 15 million users generating errors daily. Imagine Expensify as a Rail: every correction across thousands of companies teaching the entire network. Your operating costs benchmarked against similar companies. Edge cases from one company preventing errors at another. This isn't about user count. It's about work count. Every pattern resolved becomes reusable logic with provenance. Rails monetize mistakes across the network. SaaS monetizes features in isolation. The Three Laws of Operating Rails Law 1: Learn faster than they copy  Your rail must improve faster than competitors can imitate. A competitor can copy features in weeks. They can't compress five years of edge cases resolved, versioned, and rolled back. The moat isn't what you built; it's how fast you compound. Law 2: Make it tacit, not portable  Perfect documentation is easily copyable. The real moat lives in tacit knowledge, patterns that only work with your specific context, governance, and audit trails. Law 3: Power without transparency kills trust  As rails automate more decisions, they need more governance. One bad auto-execution can destroy years of trust. Constitutions, vetoes, one-click rollbacks aren't nice-to-haves. They're essential. NewCo Playbook: Start Boring, Compound Fast (Exclusive to newsletter subscribers) The winners won't be the ones who automate fastest. They'll be the ones who learn from failure fastest. LegacyCos are adding AI agents to workflows, hoping for magic. But intelligence in local instances doesn't compound. You don't build moats by making each silo smarter. NewCos who grasp Operating Rail laws will target boring shared burdens. They'll turn every error into network intelligence. They'll shoulder bounded liability to earn trust. And they'll compound learning at rates LegacyCo can't match. Strong operators compound errors into moats. Weak operators add features and hope. (Full version sent to subscribers)

55

24 Comments

Do engineers need to learn unit economics? This wasn't a thing in traditional SaaS. Engineers built features, marginal costs were essentially zero. Gross margins were 80-90%. AI broke that model. I got to review quite a few AI SaaS companies ("wrappers") over the last two years - impressive growth rates, but most of them with pretty low gross margins in the 50-60% range. Some even had traditional seat-based pricing without usage limits that could have caused negative gross margins in the worst case. Luckily it didn't. From experience, I know many SaaS founders aren't super scientific about pricing (especially bootstrappers). But if you're building AI-heavy features, the pressure to become obsessed with pricing and unit economics is rising fast. Now engineering decisions could have huge P&L impact: → Choose GPT-4? $60 per million (output) tokens → "Free" open-source model? Infrastructure costs might surprise you → Build a "simple" AI feature? Could cost $1,500/day at scale The problem: Founders usually let engineering teams pick tech stacks purely on technical merit. With LLMs, there's big risk they'll discover unit economics don't work. What's different now: • LLM API calls cost real money that scales with customers/usage • Gross margins highly depend on architecture choices • Model selection = pricing strategy decision Not sure if this means you should let your engineers sit in pricing discussions or involve finance in model selection. But if you get this right, smart infrastructure choices could give you a massive competitive advantage. For example, you could offer more generous free tiers than your rivals, and/or undercut them regarding price. Even if language model inference costs keep dropping, the fundamental shift remains - your technical decisions now directly impact unit economics. Your CTO and CFO need to become best friends. What's your strategy for managing AI costs as you scale?

73

22 Comments

LLMs aren’t engineers. They’re just convincing guessers. Especially when it comes to your backend. AI doesn’t understand your infra, configs, or dataflows. It hallucinates what it can’t see. That’s the fatal flaw in most AI dev tools. But what if AI had context? Not just codegen—actual system awareness? 🧠 Tetrix AI by Deskree is launching soon: A live context layer that lets AI reason like your best engineer. It knows your stack. Knows where code fits. Knows how to ship it. Ethos Fund backed them twice. This isn’t hype. It’s infra. Full story (and wild founder journey) in this week’s LinkedIn Legendaries—https://lnkd.in/gz7wNsfp

52

20 Comments

The latest AI Product Kitchen is live - with Nan Yu, Head of Product at Linear!! Linear just closed their Series C ($134M total), are taking on Jira, and power the world's fastest-growing tech teams: Cursor, OpenAI, Vercel, Notion, and Replit. We cover: - Why Linear is building AI agents as teammates - Why backlogs should be deleted - How Linear wins with fewer features -- Here's 6 of my fave spicy takeaways -- 1. Linear wins by doing less, better They don’t try to match Jira. They do the opposite - focused, opinionated, solving the right problems, beautifully. Less surface area = more velocity 2. Every product org is a series of bets You’re not shipping a roadmap - you’re placing wagers. Linear treats every decision as a hypothesis: test quickly and adjust fast. Speed > perfection 3. Backlogs are where good ideas go to die Most teams hoard tickets “just in case.” Linear isn’t afraid to delete them. If it’s not clear and actionable, it doesn’t stay in the system. Clarity > clutter 4. If your system doesn’t help future-you, it’s useless Linear designs for retrieval, not perfection. They ask: "Will future me understand this?" If not, it gets rewritten or tossed. Build for context that’s long gone 5. Good PMs ship features. Great PMs ship clarity Anyone can ship a feature. The real value is aligning people, defining the “why,” and reducing ambiguity. Clarity scales. Confusion compounds 6. Define ownership before outcomes Linear bakes ownership into the system. Without clear responsibility, even great strategies fall apart. Every initiative starts with “Who owns this?” If you care about speed, clarity and building impactful product orgs - this ep's for you. See the full ep: "Agents as Teammates: Linear's AI Vision" on YouTube or links in comments below

71

23 Comments

Startup exploitation reaches peak absurdity when non-technical founders offer $60k SF salaries for "founding engineer" roles with 0.5% equity. This systematic labor exploitation targets technical workers through aspirational language while providing starvation wages in expensive markets. The founder manipulation playbook: - Use "founding" titles to justify below-market compensation - Offer minimal equity while demanding maximum technical responsibility - Exploit learning opportunity narratives to justify professional underpayment - Position technical work as privileged access rather than essential business contribution Your startup evaluation framework: - Does compensation reflect market rates or founder equity preservation strategies? - Are equity offerings proportional to responsibility levels or designed to minimize founder dilution? - Do learning opportunities justify below-market wages or serve founder cost optimization? - Are technical contributions valued appropriately or treated as discounted labor? The most expensive technical talent mistakes happen when skilled engineers accept equity theater instead of market-rate compensation. Non-technical founders benefit through reduced operational costs while technical employees subsidize business development through discounted labor. Demand market-rate compensation regardless of equity promises that typically provide minimal financial return. Choose companies that value technical expertise through appropriate compensation rather than founders who exploit professional development anxiety. Build career progression through market-rate positions that provide genuine equity participation proportional to technical contribution levels.

238

73 Comments