Why not use a pure AI service?

Pure AI services hit a ceiling at the first carrier denial. Negotiation, regulatory escalation, and small claims filings require human judgment. Pure AI services often stall or refund the claim rather than fighting it. Human-plus-AI services maintain higher recovery rates on contested claims.

Can AI write my appeal letter?

AI can draft a template letter, but successful appeals require case-specific argumentation, citation to specific regulatory provisions, and sometimes reference to recent case law. A pure AI letter has a measurable lower acceptance rate than a human-reviewed letter.

Is TravelStacks an AI service or a human service?

Both. AI handles eligibility classification and document parsing instantly. Human reviewers validate eligibility verdicts and handle escalation. The combination produces faster claims at higher recovery rates than either approach alone.

What is the AirHelp approach to AI?

AirHelp uses AI heavily for first-pass filing on its standard plan. Human teams handle escalation. The 35% commission reflects the cost of the human escalation infrastructure. AirHelp+ subscription bundles additional services that are partly AI-driven and partly human.

Did ClaimCompass fail because of bad AI?

Partly. ClaimCompass marketed as pure-AI claim service. The model could not handle dispute escalation. Combined with funds-routing-through-the-service structure, the service shut down in 2023, leaving some passengers exposed. See [airhelp alternatives after claimcompass shutdown](/blog/airhelp-alternatives-after-claimcompass-shutdown).

Are AI services cheaper than human services?

Sometimes. TravelStacks combines both at $19 flat for US DOT and 25% EU261, which is roughly competitive with pure-human services. The benefit of AI is speed, not price. Most claim cost is in the regulatory entitlement, not the service overhead.

What does FTC say about AI legal services?

The FTC took enforcement action in 2024 against DoNotPay for the AI lawyer framing. AI services that imply legal representation without actually employing licensed attorneys are subject to FTC scrutiny. TravelStacks does not market as legal representation; it markets as a claim filing service with optional human escalation.

← Back to blog

ComparisonsApril 29, 2026•10 min read

AI-Powered Flight Compensation: How TravelStacks Uses Technology

Q: Does AI actually help with flight compensation claims?

Yes for two specific tasks: classifying claim eligibility against the full regulatory criteria, and parsing boarding passes, FIDS photos, and carrier emails to extract claim-relevant data. Outside these tasks, AI does not meaningfully outperform a knowledgeable human.

Loren Castillo

Founder, TravelStacks

AI flight compensation technology is mostly marketing, with two exceptions: claim eligibility classification and document parsing. TravelStacks uses targeted AI for the parts of the claim where machine learning genuinely outperforms manual review, while keeping legal escalation in human hands. This guide explains where AI helps, where it does not, and the honest framework for using technology in passenger rights.

AI Flight Compensation Technology: The Honest Framework

AI flight compensation technology is heavily marketed but rarely well-defined. Most services labeling themselves AI-powered are doing simple rule-based matching against regulatory criteria. Genuine machine learning helps in two specific places: (1) classifying claim eligibility against the full body of EU261 and DOT regulatory criteria, and (2) parsing boarding passes, FIDS photos, and carrier emails to extract claim-relevant data automatically. Outside these two narrow tasks, AI does not meaningfully outperform a knowledgeable human reviewer.

The myth: 'AI will write your appeal letter and the airline will pay'. The reality: AI extracts data from your evidence, classifies the eligibility, and assembles the file. Legal escalation, dispute negotiation, and edge case handling remain human. Both pieces matter.

Where AI Genuinely Helps: Claim Eligibility Classification

EU261 and US DOT eligibility involves multiple interacting criteria: distance band, delay duration, carrier nationality, route geography, extraordinary circumstances claims, codeshare operating carrier identification. A trained classifier can ingest a passenger's flight details and produce an eligibility verdict in seconds with higher consistency than a human first-pass review.

›
Distance band determination: great-circle calculation between IATA codes. Trivial for AI.
›
Delay measurement: cross-reference scheduled vs actual arrival time. AI extracts from FIDS or BTS data.
›
Carrier nationality: EU261 only applies to EU-licensed carriers. Carrier database lookup.
›
Codeshare operating carrier identification: parse the ticket number and operating carrier code. Common edge case where AI catches errors humans miss.
›
Extraordinary circumstances classification: AI matches the carrier's stated reason against the case law database (Wallentin-Hermann, Sturgeon, Pesková, etc.) for first-pass classification. Human review on edge cases.

Where AI Helps: Document Parsing

›
Boarding pass extraction: OCR the booking reference, flight number, departure/arrival airports, and date. Saves manual data entry.
›
FIDS board photos: extract scheduled vs actual times. Cross-reference with carrier-stated delay reason.
›
Carrier email parsing: extract delay reason, rebooking offer, refund offer. AI flags evasive language patterns.
›
Receipt categorisation: itemize meals, hotel, transport during delay. Calculates Article 9 right of care reimbursement.
›
Multi-passenger ticket consolidation: AI identifies family bookings and consolidates claim filings.

Where AI Does Not Help: Legal Escalation and Negotiation

Once a claim is denied or stalls, the work shifts to negotiation, regulatory escalation, and sometimes litigation. AI cannot substitute for:

›
Negotiating with airline legal teams: human relationship and case-specific argumentation.
›
National enforcement body filings: AESA, ENAC, LBA, DGAC, CAA processes are bureaucratic and require careful presentation.
›
Small claims court filings: jurisdiction-specific procedural rules, evidentiary requirements, statute of limitations calculations.
›
Edge case handling: the 5% of claims where the carrier invokes a novel extraordinary circumstance defence or where evidence is ambiguous.
›
Multi-jurisdiction complex cases: trips spanning multiple regulations require human judgment on which framework to invoke first.

Pure AI claim services hit a ceiling at the first denial. TravelStacks routes denied claims to human escalation immediately. Pure-AI competitors stall.

The TravelStacks Approach: AI Plus Human in the Right Order

1
AI ingests flight details and classifies eligibility instantly.
2
AI parses uploaded documents (boarding pass, FIDS, emails, receipts) and pre-fills the claim file.
3
Human reviewer validates the eligibility verdict, especially for extraordinary circumstances cases.
4
AI assembles the formal claim and submits to the carrier portal.
5
If the carrier responds within 8 weeks with payment: AI processes, human verifies the payment matches entitlement.
6
If the carrier denies or stalls: human escalates to DOT or NEB. AI assembles the regulatory filing template; human writes case-specific argumentation.
7
If small claims litigation is required: human attorney engagement. AI assists with evidence assembly.

Red Flags: Pure-AI Claim Services to Avoid

›
'Fully automated': usually means the service cannot escalate when the carrier denies. The first 'no' becomes the last word.
›
'AI lawyer': lay representation in court is not legal in most jurisdictions. Marketing copy that suggests otherwise is misleading.
›
'Approval rates above 95%': industry baseline is 50-70% on contested claims. Inflated approval rates suggest the service only files easy claims and discards hard ones.
›
'Settlement in 24 hours': regulatory minimum is 7 business days for US DOT credit card refunds. EU261 typical is 4-12 weeks. Faster claims are possible but not the norm.
›
'Powered by GPT-4' or similar: vague AI marketing is a tell. Reputable services explain which specific tasks use AI and which do not.

Comparing AI Use Across Major Services

›
TravelStacks: AI for eligibility classification and document parsing; human for escalation. Transparent about which steps use AI.
›
AirHelp: heavy AI for first-pass filing; human team for escalation. 35% commission reflects the human cost.
›
Compensair: lighter AI use; primarily template-driven workflow.
›
ClaimCompass (defunct): pure-AI promise; failed in 2023 partly because the model could not handle dispute escalation.
›
DoNotPay: marketed as AI lawyer; FTC enforcement against the AI lawyer framing in 2024 (federalregister.gov on consumer protection).

Get a Realistic Eligibility Estimate

AI helps where the work is repetitive (eligibility, document parsing) and human helps where the work is judgment-heavy (negotiation, escalation). The best services use both. Use the delayed flight worth calculator to start your eligibility check, see best flight compensation services that guarantee payouts for the broader service comparison, and the EU261 passenger rights pillar for the regulatory framework. Start a claim.

Think your flight qualifies?

Check in 30 seconds. Free to find out.

Check my flight