Faro Research · June 2026
We ran Faro’s AI readiness audit on 7 major fintech companies, checking raw content depth, pricing transparency, API discoverability, llms.txt quality, and structured data. Here’s what the data shows.
Key Findings
Stripe is the only fintech with a native MCP server. It's not just AI-ready, it's AI-native.
Wise has the best pricing transparency in the group. Their fee comparison table sits in plain HTML on the homepage. That's exactly what an agent needs.
Plaid and Adyen have the largest llms.txt files (112K and 110K characters). Both are clearly maintained by someone who owns AI discoverability as a job.
Starling Bank filed an empty llms.txt. The file exists but has zero content. That's almost worse than not having one at all.
Remitly scores well on content depth for a consumer company but has no developer API. There's nothing for an agent to actually integrate with.
The 37-point gap between Stripe (91) and Remitly (54) isn't a budget gap. It's an attention gap.
Scores reflect what Faro’s audit engine can verify from publicly available signals, no logins, no insider access.
Each company audited on five signals: raw content depth (can an agent read the page without JS?), pricing transparency, API readiness, llms.txt quality, and structured data/metadata.
stripe.com
STANDOUT FINDING
Only company in the group with a native MCP server. Already building for the agent era, not retrofitting for it.
wise.com
STANDOUT FINDING
Best pricing transparency in the group. A fee comparison table against Bank of America, Wells Fargo, and PayPal sits on the homepage in plain HTML. No clicking required.
adyen.com
STANDOUT FINDING
One of the largest llms.txt files in this audit at 110K characters. Someone internally owns AI discoverability. It shows.
plaid.com
STANDOUT FINDING
Largest llms.txt in the group at 112K characters. The most structured AI-readable context of any company we tested.
revolut.com
STANDOUT FINDING
Only neobank with all plan prices on the homepage in plain HTML. Standard (free) through Ultra (£55/mo). No clicks, no login wall.
starlingbank.com
STANDOUT FINDING
The llms.txt file exists but has zero content. A missed signal for a bank with genuine developer ambitions via Engine by Starling.
remitly.com
STANDOUT FINDING
Well-structured llms.txt for a consumer company. But with no developer API, there's nothing for an agent to actually do once it reads it.
Stripe, Plaid, and Adyen were built API-first. That maps directly to AI readiness: structured content, discoverable documentation, maintained llms.txt files. They didn't need to retrofit anything.
Remitly's core product is moving money between people. There's no reason for a developer API, and their pricing is corridor-based by nature: a human-readable FX rate, not a structured SaaS price. The AI readiness gap isn't a failure. It's a product category mismatch.
Four of seven companies have verified llms.txt files. The quality gap is striking: Plaid's 112K file versus Starling's completely empty one. Having the file there isn't enough. What's in it matters.
Revolut and Starling both have developer programs, but neither has treated AI discoverability as a priority. Revolut's homepage pricing is genuinely excellent. It just isn't matched by the API or llms.txt investment.
Each company was audited using Faro’s AI readiness engine on June 29, 2026. We fetched each company’s homepage and evaluated five signals, each scored out of 20:
Raw content depth
Can an AI agent read your homepage without executing JavaScript? Measures information density in raw HTML.
Pricing transparency
Are prices findable and machine-readable without navigation, authentication, or 'contact us' gatekeeping?
API readiness
Does a public developer API exist? Are docs discoverable from the homepage within 2 hops?
llms.txt quality
Does the site have an llms.txt file? Is it populated with meaningful, structured content?
Structured data
Are OG tags, JSON-LD schema, Twitter cards, and canonical URLs present and well-formed?
Scores are based on what Faro’s engine can independently verify from public signals. No login, no insider access, no assumptions. llms.txt verification required an active fetch; “unverified” means the check timed out or returned no response, not that the file doesn’t exist.
Most AI readiness conversations focus on content. Can ChatGPT find and describe your product? That matters. But fintech companies face a harder problem. Agents aren’t just reading about financial products, they’re being built to take action: compare loan rates, initiate payments, pull transaction data, move money between accounts. A well-written homepage gets you to the starting line. What actually matters is whether your API is agent-accessible, your pricing is machine-readable, and your docs are structured enough for a model to build an integration plan from scratch.
Stripe understands this. Their MCP server lets agents running Claude or similar models access Stripe functionality natively, not by scraping a webpage, but through a first-class integration. That’s AI-native product design. The rest of the industry is still optimizing for human discovery.
Most of the gaps on this list are fixable. An empty llms.txt can be populated in a day. Pricing transparency is a copywriting and architecture decision. API docs buried three navigation layers deep can surface with one footer link. The harder work is building a developer API for a consumer product, that takes longer. But it’s the bet worth making as agents become the primary way people interact with financial services.
Check your score
Run Faro’s AI readiness scan on your own domain. Takes 30 seconds. You’ll see exactly which signals you’re passing, which you’re failing, and what to fix first.
Scan your site free →No sign-up required for the first scan.