Big features like Localization used to give you a 12-18 month moat. And we just did it in a Waymo ride.
Back in the day, pre-AI, we used to beat DocuSign head-to-head. At Facebook. At Google. At Yelp. At Twitter. At a lot of the biggest tech companies of the era.
Reason #1 sounds almost comically simple:
We were localized in 20+ languages. They weren’t.
That was it. That was the 10x killer feature we used to win six figure deals at Facebook, at Google, at Twitter, at Yelp, at 100s of other top tech deals. If they had customers in Tokyo or São Paulo or Frankfurt and needed signing flows that worked in their language, we had it. The competition didn’t.
It took DocuSign almost 18 months to catch up. To see it, to prioritize it, to plan it, to prototype it, to test it, to ship it.
Eighteen months where we won deals we often had no business winning on pure product parity. Eighteen months where “we support your global customer base in their native languages” closed deals that feature-by-feature comparison tables would have lost.
Today? They could do it in a weekend. Anyone can. For real.
What Localization Actually Cost in 2010-2020
At Adobe Sign / EchoSign, localizing into 20+ languages wasn’t a weekend project. It was a 6+ month slog to rebuild the product, add the localization frameworks, and we had a dedicated engineer on it plus support from our CTO and Director of Eng.
You also had to hire native speakers. And/or contract translation agencies and manage them. You had to build an i18n framework into your codebase. You had to figure out right-to-left rendering for Arabic and Hebrew. You had to handle character sets for CJK. You had to deal with legal review for every jurisdiction because “electronic signature” means different things in Germany than it does in Japan. You had to localize not just strings but formats, date orderings, name orderings, address formats, phone number formats.
And then every new feature, every UI change, every new error message had to get re-translated, re-reviewed, and re-deployed across 20+ locales.
The companies that could afford to do this had a real structural advantage. The ones that couldn’t, didn’t.
It wasn’t a nice-to-have. It was an 18 month moat. Back then, pre-AI. Maybe Even in 2024.
It even took Shopify over a decade to localize its product.
The Shopify story is even more striking. Shopify was founded in 2006. One of the best pure product engineering organizations of its generation. Obsessed with global commerce from the start.
They didn’t ship native multi-language selling on the storefront until around 2020. Roughly 14 years after founding.
Shopify, with thousands of engineers, a world-class product org, and a direct financial incentive to serve international merchants, still needed more than a decade to properly ship multi-language. The admin got a 6-language beta in 2018. The full storefront multi-language capability came in 2020. Shopify Markets, the real cross-border management layer, didn’t arrive until 2023.
Seventeen years from founding to full localization stack.
And this was not because the Shopify team was slow. It was because localization is genuinely, legitimately hard. Translation. i18n framework. Currency handling. Tax logic per jurisdiction. Locale-aware SEO. Theme-level string extraction. Third-party app compatibility across 20+ locales. Every new feature has to ship in every language from day one, or you accumulate debt you never pay off.
This was hard for us in 2010. It was hard for Shopify across an entire decade. It has been hard for every serious B2B company that ever tried to do it.
What Localization Costs in 2026. 20 Minutes To Start, Hours to Proof. And Maybe $200 of Tokens.
This week, Amelia and I localized our entire AI VP of Marketing app, “10K,” into Chinese, Spanish, and a handful of other languages. On a phone. On Replit. During one Waymo ride.
That’s not an exaggeration. One Waymo ride. We talked about it on The Agents #001 podcast.
10K is a real production app. It’s the thing that designs every campaign, every offer, every marketing action for SaaStr AI Annual 2026. It’s been running for months. It has real data, real workflows, real API integrations. And localizing it into multiple languages was… a conversation with Replit while we were in the back of a car.
The AI read the codebase. Identified every string. Generated translations. Handled the i18n framework. Tested the rendering. Shipped it.
What used to take 18 months and seven figures took under an hour to get to a v1 and cost nothing beyond the Replit subscription we were already paying for.
Your Killer Feature Moat Might Last a Weekend Now
If localization or other killer, “tough” features were your moat, well … they aren’t anymore.
And if you’re sitting there thinking “well, localization was never really my moat” — ask yourself what else on that list has just evaporated.
- Integrations with long-tail tools? AI can write those now. In minutes.
- Industry-specific workflows? An AI can study the industry and generate them.
- Admin panels, reporting dashboards, custom fields, role-based permissions? Commodity. All of it.
- Mobile apps to match your web app? Commodity.
- Translations of your docs, your help center, your onboarding emails? Commodity.
Every single thing that used to take an 8-engineer team 18 months to ship is now table stakes that a solo founder can have running by Friday.
The durable moats in B2B in 2026 are not the ones that used to matter. They are:
- Distribution. Who already has the customer’s trust and attention.
- Data. Proprietary datasets that compound over time and that competitors genuinely cannot replicate.
- Network effects. Real ones, not marketing-deck ones.
- Brand. Actual brand, built over years, that customers trust when they’re nervous.
- Speed of iteration. The ability to ship 10x faster than the incumbent. (Which, ironically, AI both enables and commoditizes at the same time.)
Notice what’s not on that list: your feature set. Your localization footprint. Your integration catalog. Your admin panel.
None of that is a moat anymore. None of it.
Everyone Can Move Even Faster Than Just 90 Days Ago. You Have to Find A Way to Keep Up. And Even … Pull Ahead.
When we beat DocuSign at Facebook back in the day, the product gap was real but the work gap was larger. We had put in months of work they hadn’t, and seen the market gap months before that, that they didn’t. The work itself was the moat.
AI has collapsed the work gap.
The work that used to define competitive advantage, the unglamorous, patient, years-long accumulation of translated strings and localized legal flows and custom integrations and industry-specific reports, that work is now free. Or close enough to free that it doesn’t matter.
This is why pure B2B is getting squeezed and why AI-native companies are re-accelerating. The old B2B playbook was partly about out-working the competition on a thousand small things. That playbook doesn’t work when your competitor can replicate 18 months of your work in a weekend.
What is your actual moat in a world where a lot the “hard work” of B2B is now a Waymo ride?
If the answer is “our product is more localized” or “we integrate with more things” or “we have more features”, it’s time to start over. Those are not answers anymore.
The answer has to be something AI can’t commoditize. Distribution. Proprietary data. Network effects. Brand. Shipping velocity.
Everything else is a weekend project now. That’s no longer theory or folks hyping AI on twitter. It’s very real.
Try It Yourself
Pick the hardest engineering project on your roadmap. The one that’s been slotted for Q3 because it’s going to take a whole team a quarter to ship.
Then open Replit. Or Lovable. Or Cursor. Or whatever. And try to ship it this weekend.
You might not succeed. But I promise you will be shocked at how close you get.
And when you see how close you get, you’ll understand why the moats you thought you had are gone. And you’ll start thinking much harder about the ones you actually need to build.
