About
Entrepreneur with multinational experience of founding and building in 3 different…
Activity
12K followers
-
Dan Lifshits shared thisYou know a $10bn round is not a rumour when you see this 😅 Another good Bending Spoons candidate soon! Instinct
-
Dan Lifshits shared thisWe planned our AI hackathon in October for 20 teams — and we filled that in 2 days. Since then, applications have come in at more than double our capacity. That says something: builders are tired of demos that only work in a sandbox. Give people a real operational mess to test against, and they show up. Now the less fun part — reviewing every application and picking the strongest fits. We'll get back to everyone soon, but if you applied: thank you. You're the reason this is a hard problem to have.
-
Dan Lifshits posted thisBuilding AI agents is fun. This week, a bug created 200 maintenance jobs that should never have existed. Some of our landlords received a message to renew their gas certificates. The only problem is that they did it a month ago. Agents were prepared to talk to them, get an engineer to come in, do the full thing. We canceled the jobs, rolled back the change, and apologized. When software does the work and contacts customers, any deployment mistake is harsh. At Dwelly, we operate the agencies where we deploy our software. So it’s us who have to handle this. But the benefit is it shapes how we think about building our product. We know deeply that automated action carries a lot of weight.
-
Dan Lifshits shared thisStep 1/ Claude "hottest SaaS startups now". Step 2/ - now please vote. Which of these companies would be acquired by Bending Spoons within the next 3 years? a) Wonderful b) Sierra / Decagon c) Mercor d) Lovable / Replit e) Factory / Blitzy f) Harvey / Legora / Wordsmith g) Clay h) Granola / Viktor / Instinct
-
Dan Lifshits posted thisWe turn agencies’ best practices into product features. Because software is scalable. Every agency we integrate can improve how the whole company operates. Some agencies in our portfolio collected photos and detailed descriptions of maintenance issues so contractors could provide an initial estimate before visiting. That gave landlords a cost to consider and helped contractors prepare for the job. We saw this practice and turned it into a product. Gathered all the previous data and organized it. Then we rolled it out across all our agencies. Any time something happens at a Dwelly managed home, we know how much it’ll probably cost right away. It’s not the first time it has happened. We found another agency with a well-developed approach to rent guarantees. Its team had figured out how to sell the product and manage the process. We’re now applying those lessons across the group. Each integration adds to our understanding of how to run these workflows. This is basically Helmer’s Process Power: we create moats by accumulating operating knowledge and injecting it into how the company works. Our long-term advantage comes from all the accumulated operating knowledge embedded in our software. And software is uniquely scalable. The smallest local improvement can become part of the shared AI operating system, then gets rolled out to improve service across the group.
-
Dan Lifshits shared thisLooking forward to Building Again tomorrow, 10th September, speaking with Jay Radia to a room of 100+ second-time technical and business founders about building an AI company in the UK, and what investors are really looking for right now. At Dwelly, we're doing it in a pretty unconventional way, buying letting agencies and rebuilding them around AI. I’m looking forward to comparing notes with a room full of founders who’ve been through the building journey before. For founders in my network, especially technical, there are a few spots left (link in the comments). See you there!
-
Dan Lifshits shared thisOn October 10, we're running our first AI hackathon with Anthropic, ElevenLabs, General Catalyst, and EQT Group — 20 teams, nine hours, London. Most AI agent demos look great in a controlled setting and fall apart the moment they meet a real customer, a real exception, a real mess. We're not interested in that kind of demo. We're giving builders a sandbox stuffed with synthetic calls, emails, tickets, and compliance headaches, and a real operational problem to build for. Then we throw your agent into that same data and see what actually holds up. If you build agents for operations and you want to prove yours works outside a controlled demo, this is for you. Link in the first comment! Places are strictly limited, so take this chance to apply with your team early.
-
Dan Lifshits shared thisIn August, Dwelly’s AI agents handled more than 1,000 maintenance tasks while reducing human work per task by around 90%. How? Maintenance is a great test for AI because it’s a tree of multiple actions. A case requires interpreting a tenant’s description and images, checking the property record and landlord’s instructions, assessing urgency, finding a contractor, arranging access, securing approval, managing changing availability, and confirming that the work was completed. Most of the work done by the letting agencies is spent on this. A conventional workflow requires around 30 human actions on average. With our agents handling the coordination, each task needed an average of five human escalations, with about three minutes of input per escalation. When an agent can’t determine the next step with sufficient confidence, it asks a person for a decision or missing context, then continues the workflow. We’re seeing the same shift in finance. More than 80% of payments are now recognized automatically, requiring no time from the finance administrator for matching and reconciliation. This saves our team another two to three hours each day. We have also integrated with ClearBank, which provides us infrastructure to further increase automatic recognition to 98%. We evaluate our agents through complete operating workflows. Human minutes per task and escalation frequency show how much work the system can handle on its own (say: "efficiency"). Accuracy, reopenings, speed, and customer outcomes tell us whether it succeeded (say: "evals, customer sentiment and brain").
-
Dan Lifshits shared thisThere's a type of engineer who, given a repetitive task, will spend three days automating something that takes twenty minutes a week. Not because the maths works out. Because leaving it manual is physically uncomfortable. If that's you, read this one properly. Dwelly is 15,000 properties and >$0.5bn of GMV (rent roll) AI native platform centrepiece of which run is a product an AI native team at the core, empowering the best agents out there. The only reason those numbers coexist is that technologies allow a breakthrough now — agentic loops running the rental lifecycle, evals telling us where they break, and every agency we acquire feeding the flywheel that makes the next one cheaper to absorb. Two things make it unusually satisfying to work on. 1️⃣ The first is that the automation is the product. Nobody has to be convinced that the tooling matters. It's the reason the company can grow at all. 2️⃣ The second is that it lands in the physical world. You ship something on a Tuesday and a real family gets their deposit back faster, or a boiler gets fixed before the weekend. Not a metric on a dashboard — someone's actual home. That feedback loop is rarer than it should be. Staff Engineer and Applied AI Engineer, both open. Details below. DM me if you'd rather just talk.Dan Lifshits shared thisMost AI companies ship an agent and hope it worked. We own the business outcome instead, allowing us to go deeper. Dwelly is building an AI-native operating platform for residential rentals - and we run the agencies ourselves, 15,000+ properties, rather than selling software to them. AI agents already handle large parts of the lifecycle: tenant enquiries, viewings, referencing, renewals, maintenance triage, arrears. Humans take the exceptions and the judgement calls. That ownership is the whole difference. Every tenancy gives us a real outcome - messy, delayed, sometimes ambiguous, but real, and ours. The other half is acquisition. We buy agencies, and every one arrives with its own legacy processes and its own pile of unstructured mess: inboxes, call recordings, PDFs, tribal knowledge. Absorbing that by hand doesn't scale. Done properly, each one makes the system better at absorbing the next. Which puts us on three problems I don't think anyone has clean answers to: How do you build one operating platform that unifies dozens - eventually hundreds - of distinct business processes, so launching the next one takes days instead of a quarter? How do you keep a multi-step workflow reliable when step six depends on a document that arrives whenever it arrives? How do you turn delayed, unlabelled, human-generated outcomes into an eval signal you'd actually ship against? We don't have this solved. We have it working, which is a different thing. I'm hiring across two roles, both remote across the UK, Ireland and European time zones: Staff Software Engineer - deep system design, 7+ years, and the judgement to take a vague business problem and drive it to production. You'd have a product manager and a team - the autonomy is over the how, not a substitute for support. The platform underneath the agents has to be as good as the agents. Applied AI Engineer - you've built agentic systems in production, not one side project. You've written evals you actually trust, watched a loop fail at 2am, and fixed the harness rather than the prompt. No years-of-experience bar; I care what you've shipped. For context: we've announced $260M+ raises since February, including a $170M Series B led by EQT Growth and General Catalyst. Runway isn't the constraint here — people are. Links in the first comment. Or DM me with the hardest agent failure you've debugged - I'll answer those first. #Hiring #AIEngineering #AgenticAI #PropTech
Dan Lifshits replied to a comment
1d
Michael Stysin we probably have different views on moats and long-term sustainable differentiation :)
-
Dan Lifshits reacted on thisDan Lifshits reacted on thisMost people think estate agency is about property. After a month inside one, I can tell you it's mostly about phone calls and emails. Thousands of them. Tenants chasing repairs, landlords chasing updates, applicants chasing viewings. The average agency drowns in this, and the industry's answer has been "hire more people and hope." Dwelly took the opposite bet: buy traditional UK lettings agencies and rebuild them around AI. Not a chatbot bolted onto a website. The actual operations. Calls, maintenance, the unglamorous middle of the business where the margin actually lives. Most AI products help a person do their job faster. We're betting on something different: AI that just does the job. We're now running 15,000+ properties and over $500M in GMV, which puts us among the top 10 lettings operators in the UK. We recently raised $170m to go further, and solving for the gap between "AI optimisation" and "AI running a regulated business at scale" is one of the most interesting problems I've worked on. Now we need more people building. Two roles in particular: Applied AI Engineer to build the agentic infrastructure that automates real operational work across our agencies: orchestration, memory, evals, observability. Pragmatic production engineering, not research. Staff Software Engineer to go deep on one core domain, work out what can be automated, and own it from concept to production. There's plenty more open across engineering and beyond, so if neither of these is quite you but the problem sounds interesting, please reach out. Link in the comments.
-
Dan Lifshits reacted on thisDan Lifshits reacted on thisComing off the MADRING Grand Prix, which showed Fever at its best, I'm proud and a little humbled to share another milestone that belongs to a lot of people here. We raised $250M, led by EQT Group, Point72 Ventures and joined by existing investors. It is the largest round ever raised by a live entertainment tech company. At Fever, capital has always had a specific purpose: to be able to build the most cutting edge technology there is in the world of live entertainment. A round like this is really a vote of confidence. It is a testament to the vision Ignacio Bachiller Ströhlein, Alexandre Perez Casares and Francisco Hein had from the start, and to ten years of work by everyone who has built on it. Seven years ago I was looking at Fever from the outside, trying to work out whether an ambitious startup could really change how people find things to do. The problem was hard and the market was enormous, and that combination is what got me. 100x growth later, I'm more convinced than I was then. We live in a world that is more and more mediated by screens, and isolation among younger generations is a real issue. Choosing to invest in live entertainment, in getting people into the same room, feels both compelling and encouraging. A lot has been said about what AI will do to our work. The optimistic version is that it gives us time back. What we do with that time is the real question, and I hope the answer is that we spend it with the people we love, doing something real. New funding by itself is not a goal or an achievement. But knowing there is the encouragement and the resources to keep delivering on the promise of democratizing access to live entertainment, that is a reward in itself. https://lnkd.in/e497U3mvFever raises $250m for live entertainment tech expansionFever raises $250m for live entertainment tech expansion
-
Dan Lifshits reacted on thisDan Lifshits reacted on thisHiggsfield is one of the most talked-about AI startups in Silicon Valley right now. I’ve known its founder, Alex Mashrabov, for more than 10 years. Recently, we spent a day talking about where technology is heading, and at one point I asked him: How can startups compete with giants like Meta or Snap? Here’s what he said: Sometimes an opportunity exists not because big companies can’t pursue it, but because nobody inside wants to own the risk. He used loneliness and mental health as examples. Meta or Snap may have the capital, talent, data, and distribution to build products in those spaces. But imagine being the PM who has to make that call. The upside may be huge, but so are the reputational, regulatory, and career risks if something goes wrong. At some point, staying away becomes the rational decision. Founders play a different game. They can take risks that would be almost impossible to justify inside a large organization. We usually talk about speed, focus, and innovation when explaining why startups can beat bigger companies. But sometimes the real advantage is simpler: founders are willing to take responsibility for a risk that nobody inside a big company wants to own.
-
Dan Lifshits reacted on thisDan Lifshits reacted on thisfirstminute capital summit is one of the many reasons why firstminute are the best in the business. Connecting portco founders, co-investors and industry thought-leaders on all things AI this year. Thanks for giving neno some airtime with fellow builders in the AI-native services category 🙌🏻 London we’ll be back soon with some exciting updates 🤠
-
Dan Lifshits reacted on thisDan Lifshits reacted on thisIntegral closes a €18 million Series A to innovate account and tax reporting Berlin-based accounting startup #Integral has raised €18 million in a Series A round co-led by Mosaic Ventures and Reid Hoffman, with participation from Cherry Ventures, General Catalyst and Puzzle Ventures. The investment brings the company's total funding to over €30 million. Founded in 2024 by Lukas Zörner and Anil Can Baykal, Integral combines AI agents with licensed human experts to deliver direct accounting, tax, and payroll services to SMEs. Through its affiliated firm, Integral Tax, AI handles routine tasks like invoice reconciliation and tax filings, which are then signed off by professionals. The approach has already slashed monthly accounting turnarounds from weeks to hours. Integral will deploy the capital to expand its engineering, commercial, and tax teams, while deepening its proprietary workflow automation to combat industry labor shortages. Source in the first comment.
-
Dan Lifshits reacted on thisDan Lifshits reacted on thisToday Jack & Jill announce their $40M Series A on their mission to revolutionise how we find work. For decades, the recruitment industry has faced a trade-off between scale and quality: J&J changes this by leveraging AI to combine the speed of software with human-like judgment. Creandum was fortunate to lead a $20M seed round under a year ago; since then, we have watched Matthew and Saaras move at lightning pace to excel across Product, Engineering, Hiring and more. They're assembling an ultra-talent-dense team and scaling across London, SF and New York. If you are interested in disrupting one of the largest markets on earth, you should join this rocket ship. It's a privilege to support this team, and we are so excited for what comes next. Peter Specht, Nathan Benaich, Karan M., Anna Chen
Experience
Education
-
The London School of Economics and Political Science (LSE)
-
First Honours
Joint degree programme - International College of Economics and Finance (ICEF) -
-
-
-
Courses
-
Econometrics
-
-
Introduction to Mathematical Finance and Insurance
-
-
Leadership as personal strategy course by McKinsey&Co.
-
-
MSc course in Strategic Management
-
-
Microeconomics: Industrial Organization
-
-
Supply Chain Management
-
Languages
-
Russian
Native or bilingual proficiency
-
English
Full professional proficiency
Other similar profiles
Explore more posts
-
David W
Raising a Pre-Seed round in the UK in 2025 for a Consumer Tech product feels a bit like Groundhog Day. You book the partner call. You show the vision, the consumer insights, the finished product, the first customers. They nod, smile… and then say the four words every founder dreads: “You’re too early for us.” Too early… for Pre-Seed. Let that sink in. The same funds that wrote £1-3M tickets into consumer apps 18-24 months ago on little more than a Figma prototype and a good story now want £200k-£500k MRR, 40% MoM growth for 6+ months, and ideally some profitability before they’ll even open the deck. Apparently “Pre-Seed” has quietly been redefined as “post-traction Seed”. We’ve heard it so many times it almost doesn’t sting anymore: “Amazing product, love the space — come back when you’ve got more traction.” So we’ve stopped arguing with the new rules and started playing an entirely different game. Instead of trying to convince funds that Pre-Seed still means “pre-traction”, we’re removing the objection completely. We’re going all-in on building traction that’s impossible to ignore: ✅ Focus on organic growth ✅ Turning our early adopters into an army of evangelists ✅ Saying no to paid growth until the unit economics are laughably good It’s slower. It’s scrappier. It’s occasionally painful. But when we do go out to raise again, nobody will be able to hide behind “too early.” We’ll have numbers that force the conversation to be about valuation and vision — not whether we belong in the room. Consumer tech isn’t dead in Europe. The era of funding hope is just over. Grateful every day for a team that’s willing to grind in the dark, and for the early users who keep reminding us why we’re doing this. See you on the other side of traction. #consumertech #startups #preseed #uktech #tractionbeforefundraising
1 Comment
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More