Why did Anchor Browser move away from credit and token-based pricing?

Anchor Browser used to have tokens and a credit system, which fits how tech companies want to buy — paying per gigabyte, per LLM token, or per VM. When the company started focusing on enterprise customers, that same technical model didn't align well. Pricing is one of the most complex questions in AI right now because demand is strong from both tech companies and enterprises, and a single model rarely fits both.

What makes an AI product defensible when anyone can build with LLMs?

According to Anchor Browser, the foundational principles of SaaS moats haven't changed — but everything that wasn't a moat is now eroding much faster than it used to. The defensible path is focusing on very hard problems. If something feels easy, it will eventually get commoditized. Anchor's agentic controller, which parses a website's DOM and decides what action to take, is an example: a year ago, no solution existed.

Why does Anchor treat hard problems as a positive signal?

The team explicitly says "if it's hard, it's good" — they want to work on hard things, not annoying, mundane, operational tasks. Hard problems filter out competitors because not everyone will tolerate the pain. They also create iteration cycles, and even as models get smarter, those iteration cycles are inevitable. The more iteration cycles you accumulate, the more of a moat or edge you build over time.

How is Anchor Browser using its own product internally?

Over the past three to four months, the team began using Anchor Browser through Claude Code as part of their own stack. This dogfooding gives them direct exposure to the workflows their customers run and feeds back into the product as they grow.

Do enterprise and developer customers really need different AI pricing models?

Yes. Tech companies want consumption-based pricing — per token, per gigabyte, per VM — because it maps to how they think about infrastructure. Enterprise organizations don't respond well to that same model. Anchor is seeing simultaneous pull from both segments, which makes it very difficult to design a single pricing model that aligns with both audiences at once.

From Tech Enthusiasts to Enterprise: Repositioning an AI Browser Product

How do AI startups transition from developer adopters to enterprise buyers?

Anchor Browser recently transitioned from helping tech enthusiasts to focusing on enterprise customers. The company is seeing strong demand from both segments simultaneously.

Tech companies want to pay per gigabyte, per LLM token, or per VM — very much on the technical side.
Enterprise organizations don't align well with that same technical, usage-based model.

As the Anchor team puts it, pricing is one of the most complex sort of questions right now in AI, and different models work very well for different customers. The difficulty is finding a model that aligns with both audiences at once.

What changes in product and pricing when you move upmarket in AI?

The biggest shift is away from purely technical, consumption-based pricing. Tokens, credits, gigabytes, and per-VM billing resonate with developers but don't translate cleanly when presented to an enterprise buyer.

Anchor Browser previously used a credit-based system, but found that taking that same model into an enterprise org doesn't work. The team is actively working through how to price an AI tool when it's no longer just a simple markup on top of LLMs.

For founders navigating the same shift, the underlying lesson connects to building a defensible moat: pricing has to reflect the value of the hard problem you solve, not just the infrastructure you pass through.

“I think pricing is one of the most complex sort of questions right now in AI. Different models work very well for different customers.”

— Idan · Business AI Explained @ 13:14

How do you maintain technical credibility while serving enterprise?

Anchor's answer is to keep solving genuinely hard technical problems, even as the buyer profile shifts upmarket. The team's narrative is explicit: if we feel something is pretty easy, something is going to be eventually solved, eventually get commoditized, we try to veer away from that.

A concrete example is the agentic controller — the layer that parses the DOM of a website and decides what action to take. A year ago, there was no solution for this. The team treats difficulty as a positive signal:

Hard problems filter competition. Not everyone is going to work on it because it's such a pain.
Iteration cycles compound. Even as models get smarter, the iteration cycles are inevitable — and the more you have, the more edge you build.
SaaS moat principles still apply. Nothing fundamentally changed about what's defensible, but everything that wasn't a moat is diminishing much faster than it used to.

“if it's hard, it's good. That's like we want to see ourselves working on hard things and not like just annoying, mundane, operational things”

— Idan · Business AI Explained @ 19:56

How does Anchor use its own product to grow?

Over the past three to four months, the Anchor team started using their own product through Claude Code to support their go-to-market workflows. Dogfooding the product through an AI coding environment lets the team build internal AI workflows on top of the same browser infrastructure they sell — reinforcing both the product roadmap and their own growth motion.

Frequently asked questions.

Why did Anchor Browser move away from credit and token-based pricing?: Anchor Browser used to have tokens and a credit system, which fits how tech companies want to buy — paying per gigabyte, per LLM token, or per VM. When the company started focusing on enterprise customers, that same technical model didn't align well. Pricing is one of the most complex questions in AI right now because demand is strong from both tech companies and enterprises, and a single model rarely fits both.
What makes an AI product defensible when anyone can build with LLMs?: According to Anchor Browser, the foundational principles of SaaS moats haven't changed — but everything that wasn't a moat is now eroding much faster than it used to. The defensible path is focusing on very hard problems. If something feels easy, it will eventually get commoditized. Anchor's agentic controller, which parses a website's DOM and decides what action to take, is an example: a year ago, no solution existed.
Why does Anchor treat hard problems as a positive signal?: The team explicitly says "if it's hard, it's good" — they want to work on hard things, not annoying, mundane, operational tasks. Hard problems filter out competitors because not everyone will tolerate the pain. They also create iteration cycles, and even as models get smarter, those iteration cycles are inevitable. The more iteration cycles you accumulate, the more of a moat or edge you build over time.
How is Anchor Browser using its own product internally?: Over the past three to four months, the team began using Anchor Browser through Claude Code as part of their own stack. This dogfooding gives them direct exposure to the workflows their customers run and feeds back into the product as they grow.
Do enterprise and developer customers really need different AI pricing models?: Yes. Tech companies want consumption-based pricing — per token, per gigabyte, per VM — because it maps to how they think about infrastructure. Enterprise organizations don't respond well to that same model. Anchor is seeing simultaneous pull from both segments, which makes it very difficult to design a single pricing model that aligns with both audiences at once.