Youssef Kalad | Alleycorp

Venture Wishlist shares ideas and themes VCs want to fund. Brought to you by Purpose Built venture studio.

Youssef Kalad is an investor at AlleyCorp, one of New York’s oldest and most active early-stage venture builders and funds. AlleyCorp both incubates companies in-house and invests from pre-seed through Series A out of its latest $250 million fund. Youssef supports AlleyCorp’s Economic Infrastructure vertical—targeting regulated, under-digitized markets where successful interventions can unlock outsized social and financial returns in how people learn, earn, and access care.

Youssef’s venture wishlist

🏛️ Next-Gen GovTech Customer Service: Let citizens “call the government once” and resolve issues immediately while remembering everything about them and their needs. Bonus points if you chart a vision for the future of the government’s operating system.

⚒️ Good Jobs for the American Worker: Full-service marketplaces with advantages on either time-to-train or cost-to-train across sourcing, training, evaluation, and placement for the physical “Atoms” economy where human skill, intuition, and service will remain valuable (health, construction, climate, care, education, manufacturing).

📈 Enterprise-Grade Tools for SMBs: AI-driven marketing, CFO, training, and tax platforms for small business once reserved for the Fortune 500. AI unlocks expert talent small companies could never pay for or attract.

💊 Modernizing Government Benefits: Real-time eligibility, care-coordination, and home-based support in Medicaid—especially for high-need, low-income, and high-cost populations.

👵 Aging & At-Home Care Infrastructure: Operating systems, tooling, financial infrastructure, and copilots for informal caregivers plus reliable home-health and skilled-nursing for an aging population.

🤖 AI for the Frontline: Turn smartphones into on-site experts for frontline teams in energy, healthcare, construction, waste management, and retail. Stand out by showing uses beyond surveillance and closer to worker superpowers and a more dignified, safer worker experience.

Other insights

🏗️ Domain Expertise > Pure Code: As software creation is democratized, deep workflow knowledge becomes the durable moat.

🚀 Founder Traits: Healthy obsession, relentless velocity, true customer empathy, and hands-on industry experience matter most—especially in regulated or relationship-based markets.

🕰️ Patience Pays in Regulated Sectors: Higher compliance and capital hurdles slow go-to-market but create defensibility when solved. It helps to have worked for your customer.

🌟 Inspiring, Human-Centric AI: The future lies in tools that augment labor—eliminating the dangerous, dirty, and dull tasks—rather than replacing workers.

📚 Capturing Tribal Knowledge: There’s white space to preserve retiring experts’ know-how, turning PDFs, spreadsheets, and lived experience into searchable, trainable data sets for the next generation of field AI.

---

---

Full Interview

So usually we just start with your description of the fund and your strategy—whenever you want to share, kind of overview-wise.

I'm an investor with AlleyCorp. We're one of New York's oldest and probably most active early stage funds. The strategy is to be early and to bet on New York across sectors and categories. We also have incubation in our ethos. Our founder, Kevin Ryan, was the godfather of New York City tech back when the city was the center of everything else—finance, law, retail, media—but not tech. He founded a string of awesome New York City-based companies from MongoDB to Business Insider, Zola, Gilt, et cetera.

That’s really the fund thesis. We're investing out of our latest vehicle, a $250 million fund from pre-seed to Series A. A little more than half of our founders are based in New York City and half are elsewhere.

I joined a couple of years back to help build out our Economic Infrastructure vertical with my colleague Tanya Beja, who leads our team. The thesis there is simple: we saw an opportunity to invest in non-obvious, generally regulated and under-digitized sectors where, if you have an intervention that works, the impact upside is tremendous—for families, individuals, and institutions—and there's a lot of financial return.

In practice, it means we back founders who change the way people learn, both in K-12 and beyond; the way they earn—accessing earnings opportunities, growing careers, navigating changes in labor; and the way people access care, both care for underserved populations and broadly things like childcare and benefits, particularly government benefits.

In practice, it has looked like creating entirely new categories - upskilling / access to career advancement education (Stepful), pharmacy infrastructure in low-income communities (Alchemy), changing the landscape of government RFPs and vendor selection (ToughLeaf), or building tooling for SMBs where solving that pain point is useful for them and their labor pool (Upwage).

What is different about investing in regulated industries?

A couple of things. One, you need to have more patience because the friction when going to market is just higher. You have many more compliance requirements, in some cases capital requirements for the startup to demonstrate that, hey, if I'm going to work with the government or a bank, I need to be around for more than a couple of years. Often—and this is actually the case in education, government, and healthcare—there's a real emphasis on impact and outcomes for the end user. Increasingly, AI for labor is one of those categories where companies also want you to demonstrate that there's no disparate impact on their labor pool. So you should demonstrate that you've trained on really strong, comprehensive data.

Are there any concepts or mental models that you rely on the most when you're doing this investing?

There are four characteristics I over-index on when investing in founders, and then I’ll give you an example that came to mind from one of our startups.

The first is we really like to see someone who has a healthy obsession or compulsion—ideally it's aligned with their startup, but not necessarily.

The second is data points from that person's life that show they have energy, velocity, and can execute at really high speed. So there's purpose, but also the ability to move quickly, fail quickly, succeed quickly, and learn.

The third is that they have very clear, deep empathy for their customer—ideally because they've been the customer. It's different from sympathy, which is like, "I have an idea of your pain, I’ve run a survey." Empathy is really knowing it because you've experienced it.

Then on experience—experience matters. If you're in a regulated industry, a high-friction industry, or a relationship-based industry, it helps to have serious experience. Ideally, you’ve been in a leadership role within an incumbent institution or a high-growth startup that served that institution.

So those are the four things we look for.

On the obsession piece—Carl Madi, who's the CEO of Stepful, was a colleague of mine at Uber. He runs a company that's dramatically changing how people access healthcare careers. It’s a digital, asynchronous, AI-enabled school for physician assistants, medical assistants, phlebotomists, and higher-level technician roles in healthcare.

He said he first started working on it during Covid. He would do his day job from 7 or 8am till 7 or 8pm, and then work on Stepful from 8pm till 1am. He did that for months. His obsession was figuring out how to stem the tide in healthcare careers at a time when hospitals couldn’t hire enough people, and patients were going without care because there weren’t enough nurses. New York was importing nurses from other parts of the country at 2-3X premium. He just had that obsession and worked on it nonstop until, according to him, he found it unbearable to do anything else but Stepful.

What was surprising about what you just said is you said the obsession doesn't have to be aligned with the startup. Is that what I understood?

Yeah.

So what would that be? What's an example of that?

A good example would be someone who is obsessive about a topic area that demonstrates increasing levels of intrigue, excitement, and ability to go deeper. In some sense, it can be corny. You could be obsessive about a sports team and analytics for sports, and talk about how you've built out a spreadsheet over the years of characteristics for the type of basketball player you should have on your fantasy basketball team, and the things you've honed in on as a basketball fan. I share that because I'm an obsessive basketball fan. The ability to not just be at this high level and say, "I like basketball, that's a cool thing, I watch it," but to demonstrate depth, intrigue, and a desire to build something to improve your understanding of the game and your depth and experience—I think that's exciting. Do I plan on building a basketball startup? No. But I think we look for that obsession in people. Ideally, it's a healthy compulsion or obsession. In Carl's case, of course, he was obsessed with serving the very people that are deeply connected to his startup. So, not a bad bet for us.

You've shared your obsessions and I’m wondering if you have any non-consensus beliefs that drive your investing so far.

Yeah, I think there are two. I mentioned experience before. I want to back people who've been in an industry for a bit. One is actually around— a lot of folks are deeply interested right now in finding the highest level, highest capacity technical expertise for people building in AI or near AI. I think that's important if you're building, call it, a foundation model. But my perspective is that as software development becomes fairly democratized, the emphasis on founders will shift to domain expertise more and more. Anybody can whip up a great app or tool, but can you build something that has deep knowledge embedded in it about the workflow of the person who's supposed to use it? The small accounting shop you're supposed to help with their day-to-day or customer acquisition. Or the community health clinic that really struggles to schedule people and follow up with seniors. That's where I think value will accrue—not just with someone who's highly technically competent. You're going to need both working together.

The second, and this is just a personal kind of desire of mine—I want to invest in AI that's inspiring. I think we've talked about this one-on-one. The people building AI tooling are just as important as the thing they're building. I want to work with people who are looking to supercharge others' work using AI and who see an opportunity to have AI and labor work together versus apart. I don't buy a future where software solves everything and takes over people's jobs. I want to invest in a future where people's jobs—especially the most dangerous, dirty, challenging, monotonous parts—are made easier or eliminated entirely, so they can do stuff that's fun and exciting. That's the type of inspiring AI I'm looking to back.

You're saying that programmers won't have jobs?

I wasn't saying that.

I thought you were saying that if you know the customers with AI, you can write the product without product expertise.

Production-level code is different than a super basic demo or app. But it is true that, as a founder or small business person, you can do way more with less, especially around building software. Jevons Paradox is an interesting one. You bring down the cost and the ability to do things, and then there's just way more demand for software.

Yeah. I think the best essay I've seen on that recently was Tim O'Reilly's—about how you make software more accessible and cheap to build, and then everyone will do it. I think it still requires clarity of thought.

Yes.

Even if AI handles all the race conditions and scaling—maybe eventually even highly secure, highly scalable production code—you still have to think clearly about what you want. A lot of people know an industry and their users, but don't practice that kind of clarity about what they actually want the software to do.

Yes.

And AI could fill in something. It'll do something, but it may not be exactly what you want if you haven't thought clearly about it. That's my sense of it. But maybe that's because I still value my computer science education, which may be obsolete at some point.

So let's get into the ideas. Are there any specific ideas you're excited about right now?

Or things you'd like to find or fund?

Yeah, I think there are themes that are top of mind for me and that touch on this learning, earning, care angle, but also my background. If the last three years have been about building AI tooling for office workers, I think the next two, three, four years are for bringing AI to the field—anyone who's not at a desk and working in a taxing, laborious, high-stress frontline job. So it's for low- or moderate-income people in energy, construction, waste management, retail, health. How do we give these powerful tools to them in a way that dramatically improves what's arguably a more stressful and more challenging way to do your job, which is out in the field versus behind a computer screen?

The second is around government effectiveness, specifically state and local government, which spends something like $140 billion a year on tech alone—not including people, paper, or all of these non-automated processes. There’s an idea I’ve toyed around with: government’s not known for its customer service, but what’s next-generation customer service for government? Does picking up the phone and calling the government become the primary mode of interaction with services in a way that's easy, seamless, remembers who you are, understands your family situation, and can pull up the paperwork you've submitted before?

There’s a whole host of other things: procurement, asset management, especially where regulatory permitting limits innovation. When I was at New York State, we had a process for allowing telecommunications companies to rent space on light poles and physical infrastructure we owned. It was a very slow process. The ability to bring broadband to people and scale it was limited because we had a very manual process for knowing where the asset was, what they wanted to do with it, and how we could quickly allow and charge for it. So that’s government effectiveness.

The third is around Medicaid and modernizing Medicaid. 35% of Medicaid members are disabled. A very high proportion of Medicaid spend is specifically on long-term care. How do we focus on high-need, high-cost populations more effectively—especially people aging at home rather than in a hospital? Can we build technical rails for eligibility verification? A good example: upwards of 40 states have eligibility systems that are over a decade old for Medicaid redeterminations. If we live in a world where there’s less money for Medicaid, we need to put those dollars to greater use—less on admin and overhead, more on serving people.

The fourth theme, and this gets at the democratization of software, is that a lot of enterprise tools that were once only available to big companies can now be made available to small companies. These tools are connected to processes that SMBs needed much more money and people to do but couldn’t access—like marketing and sales support, CFO office and financial management, employee training and onboarding, tax. If I'm a small business, I can’t hire a design person or a marketing person. That’s exciting because small businesses are by far the largest driver of employment growth over the last two decades, and my hunch is they’ll remain so.

The fifth one, which I touched on with Medicaid but has broader implications, is the aging population. What do we do, especially for the part-time and unofficial caregivers in their lives—the child, the niece or nephew, the partner managing their care? It's about bringing a much higher quality, more reliable supply of home health and skilled nursing support to that population. It’s managing their eligibility, especially if they're dual eligible for Medicaid and Medicare. It's like a companion operating system for people in my generation and the one ahead, juggling responsibilities of care for both parents and children, and in some cases, other loved ones.

I think founders building in those spaces are going to be pretty well served because they feel like fairly resilient, lasting categories.

That's a lot there. I'd love to talk more about AI in the field because I have a sense that many workers have experienced computers as tracking them and telling them what to do. I think of gig workers—like, okay, got my assignment, go do the next thing. And you're talking about a situation where field workers are telling the computer what to do. That seems like a big change.

Am I thinking about that the right way?

Yeah, I think the best way to think about it is ChatGPT has made my life as an investor, and someone who is doing a whole lot of diligence on a bunch of different sectors at any one time, way easier and simpler. It’s allowed me to read more, think more, and converse with another party—even if nobody's in the office at the time—and trade ideas. Extending the capacity of people in really challenging jobs in the field is an area I think is amazing.

So I'll talk about the most recent investment we made. It's a company in stealth, but it gives an idea of what this might look like for the $150B home care industry. It’s an AI operating system to modernize home care agencies, who dispatch caregivers and home health aid to someone’s home to care for them, if they’re elderly, recovering from a severe injury, or live with a disability. The range of services varies, from doing physical or swallow therapy, bathing and moving people, or checking that they’ve taken medications and eaten. Today’s process is broken and painful for discharge facilities, families, and staff: whether an agency is scheduling a home health aide for the first time or a cancellation occurs last-minute, a 24/7 on-call scheduler sifts through lengthy PDFs and spreadsheets to find names and phone numbers for aides who might be available to take the shift (one-by-one, she texts and calls them to see who might be free). The process is both stressful, slow, and inefficient, adding immense stress to the scheduler’s life and resulting in some painful delays for families: when interviewing one aide who cares for children with disabilities, I learned that some families resort to calling an ambulance for an overnight facility stay when their caregiver cancels. AI dramatically improves this process, giving back the scheduler her off-hours, while intaking new requests for caregiver matches seamlessly and bidding out cancelled shifts quickly and effectively based on someone’s zip code, availability, training, and the family’s preferences.

Another example, one I found really exciting, was in climate, where many good-paying jobs are being created for working class Americans. Oftentimes, someone working on a well or a construction site installing solar panels, when something breaks, has to call someone in an office and ask, “What do we do? Can you send over the documentation? Can you talk me through it?” The information lives in PDFs, old SOP documents and diagrams, and email attachments. This results in delays, safety issues, and for small contractors, missed deadlines.

So is there a way to put the guy in the office in your pocket? Can you take a photo of the site and ask, “Has this been built correctly? Why is this down? What does our SOP or safety manual say?”—without delaying progress? That’s exactly what you mentioned earlier. Rather than people taking orders from a robot, they’re giving the orders to the AI in a way that helps them.

Yeah, I think getting the data is really key because it's not like you can just crawl the web and suddenly know how to repair this piece of energy infrastructure.

The domain expertise is useful here. I talked to a founder a few months ago who's building in this space. They had been at a leading climate and sustainability company for probably a decade and a half, then worked at an AI company building tooling in-house, and now they're building their own company. They knew with immense precision how these contractors and companies operated. They also recognized that most of the insight lives in PDFs, Excel spreadsheets, and emails. Training on that company’s internal data was really key. It wasn’t something you could just Wikipedia. Getting back to the prior mental model, I think depth in an industry and experience is going to be really important now.

I’ve heard of a company doing the procedure manual approach for law enforcement. There’s law, there are policy procedures—cops are supposed to go back to their car and flip through a big manual and be like, “The suspect's inside, can I break this door down? What’s the legal test?” That’s one type of problem where it's all written down, so you can imagine an LLM going through it and knowing that body of knowledge. That’s useful.

What I was talking about in terms of the data is some of this field work, I imagine, is sort of written down, but not really. Like, how do I do maintenance on this device? How do I repair this thing? How do I build this thing? What if someone made a mistake in construction—how do I debug? Some of it's written down, but I think a lot of it’s not.

So that's a big challenge in building these types of companies.

Totally. And that leaves room for data capture and data infrastructure. There are a few companies I’ve talked to who are solving the tribal knowledge problem in these sorts of industries. Their whole job is to ask, how do I make sure that—especially if your workforce, like in the case of CPAs, where something like 75% are retiring over the next five to ten years—but if we're talking about field work, it could be construction, it could be logistics—how do I make sure that when that man or woman retires in the next five years, we don’t lose all their tribal knowledge? Can I shadow them, interview them, digitize the content they’re writing down in a way that extends their capabilities for the desk worker?

We’re getting more and more used to being recorded all the time in our Zoom calls. For field workers, I think there’s still resistance to being tracked, although it does happen—GPS tracking, for example. But is someone going to wear AR goggles to record all this tribal knowledge? I don’t know. Cops have had to wear something now in many jurisdictions. They turn it off when they can. It's an interesting question.

I think that’s where the incentives really matter.

So if I felt like, Miles, you were going to penalize me by having your Grain note taker on this call, I would have said, hey, no, no Grain note taker. Sorry, I don't want that. But if I think it's going to be valuable for your ability to recall the conversation and actually go back and forth with me in a dialogue, that's awesome. I might ask you for the transcript so I can go back and critique myself on it. The incentives are super important in these industries.

The other thing I’ll say is there are industries that are so supply and labor constrained that the last thing employers should be doing is upsetting their labor force and coming across as overly invasive or punitive. Instead, they should be thinking, how do I make you stick around longer? How do I make this a long-term career for you?

Whether it’s the skilled trades or the care economy, employers are well served not to alienate their labor force here.