Believe it or not, London is the place to be as a startup founder. According to the 2025 Startup Heatmap Europe Report, London was the most popular startup hub in Europe, chosen by 32.6% of founders.
When starting this business, Adam Barbera, who was born and raised in Barcelona, Spain, also felt this calling to move to London to help the company with its next development of growth.
The startup he co-founded, Dost, is a fintech startup that uses AI to automate a company’s paperwork and admin tasks. Instead of finance teams manually typing in invoice details or chasing approvals, the software does this work automatically.
Based on his experience running a startup and his move to the UK, Barbera offers insights into his journey, what his company does in greater detail, and why he believes moving to London was the correct move for his business.
Can you give me a brief overview of yourself and your background?
Before starting Dost, I worked across engineering, sales, and business strategy. I started as a telecom engineer but soon changed my mind as I didn’t want to spend my life behind a screen. I was more interested in solving real business problems so I then moved into project management for an IoT company before joining SAP as an account executive. That’s where I learned how the accounts payable ecosystem actually works – every day I was sitting across the table from finance teams, seeing what the software did and didn’t do for them.
From SAP I moved to Tech Data, on the European team focused on new technologies. That’s where I met my co-founders, Fernando Martín and Naqqash Abbassi. We left in 2021 to build Dost. We closed a $7.8M Series A in late 2025, led by Octopus Ventures, and I moved to London in early 2026 to lead our UK expansion.
Outside of work, I am very active. I’ve played water polo since I was nine, and I still train regularly. I also focus on mental clarity through meditation and structure.
You were pitching “digital coworkers” years before AI agents became mainstream. How does that early experience shape your view of what’s genuinely transformative versus what’s just a demo today?
At Tech Data I went to Naqqash, who’s now Dost’s CTO, and asked him if it was possible to build digital coworkers. I saw a gap for software that emulates what a human does in a process, end to end. This was before AI existed as a category, we were early adopters and I wanted to make it a real process, I have always been skeptical of demos.
Today everyone has seen an impressive demo but when I sit with CFOs and ask where they’re actually applying AI in finance, most of them aren’t. They’re using ChatGPT for productivity or search so AI has effectively become a new browser for them rather than something that helps to get a job done. The genuinely transformative version runs a workflow every day, it handles the exceptions, and nobody has to babysit it. If something can’t survive contact with a real finance process, it’s a demo.
What was the moment you realised you had to leave Tech Data and build this yourself, and how did that decision shape Dost’s DNA?
Tech Data had an Innovation Committee, an internal initiative where you could bring problems from the day-to-day and try to build solutions. Accounts payable kept coming back. The company was implementing another tool at the time and it wasn’t working. So we proposed building the digital coworker ourselves, pitched it to senior leadership, asked for re
That was the moment. Naqqash wanted to build something of his own. Our other co-founder Fernando had founded a company before. I managed to convince both to join me in building Dost.
The “no” shaped everything. Before we wrote a line of code, we interviewed more than 100 companies to validate that the pain was real and universal. We’d already been turned down once on conviction alone, so we made sure the evidence was undeniable. We closed a €40,000 friends-and-family round, built the MVP, and started closing customers. That discipline – validate first, build second – is still how we operate.
What does “AI-native” actually mean in practice, and why is it essential for finance automation?
OCR works by building a template. You show it an invoice format, it learns where the fields are, and it extracts them. If a supplier changes their layout, it breaks, and someone has to fix it. Bolting AI onto that doesn’t change the architecture – it’s just a template system with a chatbot on top.
AI-native means the intelligence is the platform, not a module added to it. Dost reads any financial document at line-item level, accurately from the very first document. No templates, no configuration. AI sits at the core, everything downstream so matching, coding and exception handling improves with every invoice processed. A rules-based system stays static while ours compounds.
Finance documents are messy. A 400-page airline invoice with a discrepancy on page 3. Your legacy tool tells you “there’s an error in this invoice” and leaves your team to find it. Line-item understanding finds it for you. That’s the difference between automation that saves time and automation that creates work.
What does autonomous AI look like inside a finance back office day-to-day?
An invoice arrives – by email, portal or any other channel and Dost reads it at line-item level, matches it against the purchase order and the delivery note, codes it to the right GL account, routes it for approval if the rules require one, and posts it to the ERP. The AP team doesn’t touch it unless there’s a genuine exception. The same logic applies on the receivable side, through to collections and reconciliation.
The numbers – 95% extraction accuracy, 80% cost reduction, 90% time saved – describe that shift. Across our customers we’ve processed more than five million transactions this way.
What it means for the team differs. One CFO said that what used to take his team a week now happens in minutes, and nobody checks anything twice. The point is not fewer people though, it’s that the same team handles a growing business without growing the back office. That’s the test of autonomy: volume rises but the headcount is flat.
Why did you choose London for Dost’s next stage of growth? What does the UK offer that Barcelona or Madrid don’t, and what does it still lack?
London puts you on the map. Every serious fintech player has a presence here. US companies coming into Europe route through London; European companies going to the US do the same. You’re not entering a market, you’re entering a network that connects every other market. Barcelona is where we were born and where much of the team still is but for this stage, the conversations we need happen here.
Finance here is deeply relationship-based. You don’t show up with a deck and leave with a deal. We’ve been hosting dinners to build relations with the wider finance community. You get introduced, then introduced again, then coffee, then trust. It’s slower but it builds.
What the UK still lacks, surprisingly, is e-invoicing infrastructure. For such a digitised country, invoicing is still largely PDF and paper-based, and the mandate timeline sits behind Italy, Poland and France.
With e-invoicing becoming mandatory, what do UK businesses misunderstand about AI automation, and where do you see the biggest readiness gaps?
The biggest misunderstanding is what governments are actually pursuing. E-invoicing mandates aren’t about digitising companies for their own good, they’re about control: fraud, money laundering, tax compliance. That changes who carries the risk. Finance directors become responsible for issuing, receiving and paying invoices correctly, and if something goes wrong, they’re the ones liable.
The second misunderstanding is thinking you’re further along than you are. When I ask finance leaders whether they’re using AI, they say yes but they mean personal productivity, not automating a finance workflow. Those are different things. Meanwhile roughly 80% of businesses are still manually keying invoice data, with error rates around 10%.
The readiness gap that worries me most isn’t technology, it’s process. The UK mandate arrives in April 2029, and PDFs won’t be valid. Companies that haven’t assessed how their invoice data actually flows, where it comes from, what fields are missing and touches it, can’t just switch on a compliant system in 2028. Fraud is the same: you don’t know you have a problem until you’ve already paid €10,000 to someone who doesn’t exist. Businesses who are starting to think about the mandate now are buying themselves the time to change processes, not just software.
Anything else?
One thing that might be controversial, but I think is true: when technology projects fail, companies blame the vendor. Most of the time, it’s us – humans. The main blocker in AI adoption isn’t the model or the integration but change management. People resist changing how they work, even when the current way is the thing they complain about. Sometimes the fix is as simple as asking a supplier to add one field to their invoice but those little things come down to humans, and that’s where transformation actually lives or dies. AI is just the enabler.
For more startup news, check out the other articles on the website, and subscribe to the magazine for free. Listen to The Cereal Entrepreneur podcast for more interviews with entrepreneurs and big-hitters in the startup ecosystem.
