Explain the AI-first engineering shift—how are you building Saily going forward?

I want to start by sharing how we actually kicked off our transformation into agentic engineering. I felt that we were in the same spot as many others in the industry — reading LinkedIn posts claiming that "software engineering is dead," seeing crazy examples of someone building a browser in a record time, acknowledging that AI agents are a groundbreaking technology. That being said, constantly seeing all the excitement on social media has desensitized everyone — due to clickbaity articles, people (myself included) got overwhelmed and turned skeptical. Even though this technology is clearly the future, due to our sense about it, we moved with caution rather than shifting our entire engineering strategy to it and going all in.

We kicked off big changes after visiting Silicon Valley this February — I went there with a goal to better understand what's happening in the market. The general mood in San Francisco is that everyone's caught up in a feverish gold rush, trying to spin out "AI x [anything]" businesses. I've had the opportunity to talk to a bunch of builders, from fresh solopreneurs to engineers at Meta and Uber, successful startup founders, VPs of engineering, and the creators of the Agile Manifesto (literally). The picture became clear, and my mind was changed. Even though no one has it figured out, and no one knows where the industry is headed, AI agents are not the future, they're the present — you just have to get it right and apply them where it makes sense. I've found various businesses that have already implemented very specific agent applications in smart, practical ways.

To start our changes, we clearly stated to our whole team that we were doing this for actual gained value, not to chase the hype, and we did an AI workshop day where everyone (product people, designers, QAs, devs) had to set up Claude Code and fix at least one bug of their choice. A couple weeks later, we already had a dedicated Agentic Taskforce to implement company-wide changes, while engineers in those teams built their own tooling for specific tasks. We've made good progress with increasing code generation effectiveness and moved the bottleneck to the next phase, which is code review. After resolving reviews, we'll move to CI and monitoring, and we'll finish with product signals — collecting various user feedback from X.com, etc., and generating customer insights. That's what OpenAI Codex is solving, according to their leads (and let's be honest, by the time you read this, they've probably solved it).

So, in a month we have:
‍

• All of our engineers using agentic coding, implementing changes for other platforms than their own (iOS engineer coding backend, Android doing iOS, etc.),
• QAs oneshotting bugs rather than only identifying them
• Agents investigating outages, collecting release notes, and making calls.

Not too shabby for 4 weeks.

So where do we go from here? Two directions — one for software engineering, one for the organization overall. For software engineers, there's still a lot of areas to agentify. The role itself is already shifting to a full-stack builder in smaller teams, with more covered technologies by a single engineer, fewer handoffs, and closer proximity to customer problems. For the rest of the org, there's even more surface to apply agents, and the potential impact to everyone's work might be even higher than engineering. There are companies that already have AI ops people who work on identifying areas to introduce agents and implementing them. I see us doing the same — removing the tedious parts of work, leaving headspace for focusing on reaching results. I believe that the capabilities for agents will drastically increase every 6 months. The opportunities of what we can do are already huge, and the future will be even more exciting.

What are you optimizing for at the engineering team — speed, quality, volume, something else?

We're building an organization in hypergrowth — in 2025, Saily grew from 20 product and tech people to 100. Add AI transformation and having to ensure traveler connectivity all across the globe on top and you get quite the challenge that requires intense and rigorous processes and the necessity to constantly evolve. To make this work, we bet on smart people who are adaptive to change and are biased towards results. That's what we optimize for. Saily 12 months ago compared to now are two totally different companies and the increased efficiency and maturity is a combined result of all of our colleagues — they enable all the other qualities.

From eSIM to travel companion — should we expect multiple products, or perhaps some platform play?

Travel is a very wide industry — we're building Saily in two directions. Vertically, we're continuing our work related to connectivity and eSIMs. Horizontally, we're looking at what other traveler needs we can address — for example, why should Saily users have a separate freemium app for expense sharing? It's much more convenient to have everything about your trip in a single place, and this is something we're actively thinking about. There are several similar explorations we're doing — we don't want to limit ourselves only to eSIMs when we see that we can bring value elsewhere as well. On top of that, if personal agent tools such as OpenClaw and standards such as Shopify's Universal Commerce Protocol become a growing thing, another vertical will open up. There will be a new category of customers we need to support — AI agents representing their users. It's still the early days, so there's time to see if that actually picks up. With all of that said, at its heart Saily cares about solving travelers' needs in any way we can, and currently that leads us to one superapp. As for the future — we'll have to wait and see.

‍