What problem does it solve?
Jet fuel is a major operating cost and emissions source for an airline, and it is hard to manage day to day, because it is the sum of thousands of small decisions: the cruise cost index a dispatcher plans, whether a pilot taxis on one engine or two, when the auxiliary power unit is switched on and off, how much alternate and contingency fuel is carried, how close the aircraft's zero fuel weight sits to plan. Flight data recorders capture all of this, but turning it into a signal a pilot or a fuel manager can act on soon after the flight, rather than in a periodic report, is the harder problem.
The step change is software that scores every flight against a fleet and route level baseline and coaches pilots individually and non punitively. Icelandair has used one such platform, SkyBreathe, since 2014, running initiatives such as Cost Index Zero, single engine taxi out, alternate fuel planning, delayed APU start and zero fuel weight planning, and it reports that its total fuel savings rose 247% since 2018.
How does it work?
- Ingest flight data. Every flight's data, from the quick access recorder or ACARS, weight and balance, routing and weather, is captured and normalized across the fleet.
- Score against a baseline. Models compare each flight's actual fuel burn to a fleet and route level baseline and identify which specific levers, cruise cost index, taxi procedure, APU timing, alternate fuel, zero fuel weight, explain the gap.
- Coach the pilot. A pilot facing app shows, in plain language, where a flight over or under performed and what to try next time, framed as an opportunity rather than a mistake.
- Prioritize for the fleet. Fuel managers and flight operations engineers get the same data rolled up by route, aircraft type and procedure, to decide which standard operating procedure to change next.
- Track and report. Savings are tracked against the baseline over time and rolled into the airline's fuel and sustainability reporting.
- Audience
- Employee facing
- Autonomy
- Assist
- Adoption
- Early adopters
- Channels
- Internal tools
What is it worth?
Benchmarks are computed from the public deployments below: one data point per organization per KPI, with who made each claim.
| KPI | Median | Reported range | Data points | Claimed by |
|---|---|---|---|---|
| Cost savings | Not pooled | USD 300,000 | 1 | 1 vendor |
Value drivers: Lower cost to serve.
Indicative value
An airline burning 500,000 metric tons of jet fuel a year
USD 1.8 million to USD 9.4 million
Jet fuel cost avoided per year
How this is calculated
Formula: fuelTons * fuelPricePerTon * addressableShare * savingsRate. The low scenario uses every low input, the high scenario every high input.
| Input | Low | High | Basis |
|---|---|---|---|
| Annual jet fuel burn fuelTons, metric tons per year | 500,000 | 500,000 | The reference airline. |
| Jet fuel price fuelPricePerTon, USD per metric ton | 700 | 900 | Editorial assumption for a blended jet fuel price. Replace with your own. |
| Share of fuel burn addressable by pilot and procedure coaching addressableShare, fraction of fuel burn | 0.5 | 0.7 | Editorial assumption for the share of fuel burn influenced by cruise, taxi, APU and weight decisions rather than route network design. |
| Fuel saved on the addressable share, first year savingsRate, fraction of addressable fuel burn | 0.01 | 0.03 | Editorial assumption, replace with your own. Icelandair's 247% figure is cumulative since 2018, not a single year rate, so it does not set this range directly. |
What it leaves out: Gross fuel cost avoided only. It leaves out the cost of the platform, data integration and the fuel team's time, the value of the CO2 reduction under emissions trading or offset schemes, and any safety margin or operational trade offs behind a given saving.
Who already uses it?
2 public deployments, strongest evidence first. Grades: A regulator or audit, B the organization itself, C vendor case study, D anonymous or estimate.
Icelandair
Iceland · Travel and hospitality · 2025
Icelandair, which operates 52 aircraft to 58 destinations, has used OpenAirlines' SkyBreathe since 2014 and built its fuel program around SkyBreathe Analytics and SkyBreathe MyFuelCoach, covering initiatives such as switching from Long Range Cruise to Cost Index Zero, maximizing single engine taxi out, optimizing alternate fuel planning, delaying APU start and improving zero fuel weight planning. In OpenAirlines' case study, Helga S. Thordersen Magnusdottir, Icelandair's Program Manager for Fuel Safety and Efficiency, reports that total fuel savings have grown substantially since 2018 and that relative CO2 emissions fell between the first quarters of 2024 and 2025.
No outcome disclosed.
JetBlue
United States · Travel and hospitality · 2025
JetBlue, which operates 300 aircraft, set up a dedicated fuel optimization team and deployed SkyBreathe Analytics and the SkyBreathe MyFuelCoach pilot engagement app from OpenAirlines to turn flight data into fuel saving actions. OpenAirlines' case study headline claims a positive return on investment within three months; in the case study itself, Christopher Lum, JetBlue's Director and System Chief Pilot, reports that engine out taxi out compliance rose and fuel was saved over six months. Lum is quoted saying the airline is "not asking the pilots to be perfect" and wants to show them "where the opportunity may have been", which the case study frames as a deliberately non punitive approach.
- Cost savings: USD 300,000, in one month
"$300,000 in savings just from engineout taxi in in one month."
Claimed by: vendor
How do you implement it?
A model agnostic playbook: what to prepare, the order to build in, and what goes wrong.
Data you need
- Flight data recorder or ACARS data for the whole fleet, complete enough to compute a fuel baseline
- Route, weight and balance and weather data per flight
- A fuel policy that states what the airline can and cannot change (cost index range, contingency fuel rules) so recommendations stay inside it
Systems to integrate
- Flight data monitoring or quick access recorder system
- Flight planning and dispatch system
- Pilot facing app or portal for coaching content
- Fuel and sustainability reporting systems
Complexity: Medium
The analytics themselves are available from specialist vendors; the effort is in getting a clean, complete flight data feed fleet wide, agreeing a fair baseline per route and aircraft type, and building pilot trust that the coaching is developmental, not a scorecard used against them.
- 1
Start with one or two levers
Pick the levers with the clearest, safest savings, such as single engine taxi or APU timing, before tackling cost index or alternate fuel policy, which touch dispatch and safety margins.
- 2
Agree the baseline
Set the fuel baseline per route and aircraft type with flight operations engineering, not the vendor alone, so pilots trust that a flagged flight was really an outlier.
- 3
Make coaching non punitive
Frame every pilot facing message as an opportunity, not a report card, and keep individual results out of any performance review; a program pilots do not trust will not change behavior.
- 4
Route bigger changes through flight operations
Aggregate findings feed proposed standard operating procedure changes, which go through the airline's normal flight operations and safety review, not an automatic push to pilots.
- 5
Measure against the baseline, not the pilot
Track fuel saved against the fleet baseline over time, and separately track pilot app adoption, so a quiet program (low engagement) is visible before it shows up as no savings.
- 6
Extend fleet by fleet
Roll out a proven baseline and coaching program to the next fleet type or base once adoption and savings are established on the first one.
Guardrails
- Recommendations never exceed the airline's own fuel policy (cost index range, minimum contingency fuel); the assistant coaches within the policy, it does not set it
- Individual pilot results are not published or tied to performance review
- Every fleet level standard operating procedure change goes through flight operations and safety review, not an automatic push
- Baseline and coaching content are agreed with flight operations engineering, not the vendor alone
KPIs to instrument
- Fuel burn against baseline, by fleet, route and lever, month over month
- Pilot app adoption and engagement
- Savings by initiative (cost index, taxi, APU, alternate fuel, zero fuel weight), to see which lever is worth extending
- CO2 or fuel cost avoided against the program's own cost
Human in the loop
Pilots decide what to do differently on their own flights; the tool coaches, it does not fly the aircraft or override a pilot's operational decision. Flight operations engineering owns the baseline and any standard operating procedure change, and the fuel team reviews outlier flights before they are raised with a pilot.
Common failure modes
- Coaching pilots do not trust
- A program that feels punitive gets ignored or worked around. Keep it non punitive, explain the baseline, and show pilots their own trend, not a ranking.
- Savings that are really a mix shift
- Fuel per flight looks better because the route or aircraft mix changed, not because behavior did. Normalize the baseline by route and aircraft type before crediting a saving to coaching.
- Recommendations that ignore the fuel policy
- A model trained only on fuel burn can suggest cutting into contingency fuel or a cost index outside policy. Hard code the policy limits so recommendations never exceed them.
- A program that stalls after the easy wins
- Early levers (taxi, APU) get adopted fast and then engagement drops. Keep adding levers and keep the coaching content fresh, and track engagement, not only savings.
What are the risks and rules?
EU AI Act
Depends on design
Annex III point 4(b) covers AI used to monitor and evaluate the performance and behavior of workers. A tool that scores each pilot's individual flights against a baseline and shows a pilot where they under performed is a form of worker performance and behavior monitoring, so per pilot scoring and coaching can fall under 4(b) even when the score is never fed into a formal performance review; whether it does depends on whether individual pilot evaluation is genuinely in scope and on the provider's own Article 6(3) assessment. Fleet or route level aggregate analytics that never attribute a score to a named pilot is the lower risk design and sits outside Annex III, because it does not evaluate a specific worker. Safety critical decisions (how much fuel to carry, whether to divert) stay with the pilot in command under existing flight operations rules regardless of tier.
Guidance
- EASA Artificial Intelligence Roadmap 2.0: A human-centric approach to AI in aviation (European Union Aviation Safety Agency, Europe). Sets EASA's expectations for trustworthy AI in aviation, including human oversight of recommendations that touch flight operations.
Controls to put in place
- Fuel policy limits (cost index range, minimum contingency fuel) hard coded so recommendations cannot exceed them
- Individual pilot data kept out of performance review, with an accountable owner for the coaching program's use of pilot data
- Baseline methodology and any standard operating procedure change documented and signed off by flight operations
- Regular review of savings claims against actual fuel invoices, not only the platform's own dashboard
Frequently asked questions
- Does the AI decide how much fuel to load or where to divert?
- No, and it should not be designed to. This category of tool coaches pilots and flight operations engineering on efficiency; it should be scoped to stay within the airline's existing fuel policy rather than set it. Safety critical fuel and diversion decisions belong with the pilot in command and the dispatcher, under the airline's normal operational rules.
- What savings have airlines reported?
- OpenAirlines reports that JetBlue's Director and System Chief Pilot said the airline went from 19% to 45% engine out taxi out compliance and saved nearly 2,000 tons of fuel in six months with SkyBreathe MyFuelCoach. Icelandair's Program Manager for Fuel Safety and Efficiency reports that the airline's total fuel savings grew 247% since 2018 using the same platform, a cumulative multi year figure, not an annual rate.
- Is pilot coaching data used to rank or penalize pilots?
- In the deployments on this page, OpenAirlines' case study, based on Christopher Lum's testimony, quotes JetBlue's Director and System Chief Pilot saying the airline is "not asking the pilots to be perfect" and wants to show them "where the opportunity may have been", which the case study frames as non punitive. Keeping individual results out of performance review is a guardrail we recommend for any deployment of this kind, not a fact both case studies state.
- How mature is this technology?
- In production at two named airlines on one vendor's platform. Icelandair has used OpenAirlines' SkyBreathe since 2014, and OpenAirlines' case study headline for JetBlue claims a positive return on investment within three months of its own fuel program. That is evidence for two named airlines on one vendor's platform, not a claim about the vendor's full customer base or the wider market.
How to cite this page
Blits.ai AI Use Case Library, "AI for airline fuel efficiency optimization", last verified 30 September 2026, https://www.blits.ai/ai-use-cases/airline-fuel-efficiency-optimization. Licensed under CC BY 4.0. Method: how we verify use cases.
Changelog
- 30 September 2026: Published after review by an automated review workflow (independent skeptic review).
- 30 September 2026: Fixed adversarial review blockers. Added two OpenAirlines product pages (SkyBreathe Analytics, SkyBreathe MyFuelCoach) as sources on both evidence records, checked word for word, to establish that the specific products JetBlue and Icelandair run are AI based; neither case study itself uses the words AI, artificial intelligence or machine learning. Reattributed the "positive ROI in three months" claim from JetBlue/Lum to OpenAirlines' own case study headline, in faq[3] and the JetBlue evidence summary. Reworded faq[2] to quote Lum's testimony directly instead of paraphrasing "non punitive" as his own word. Removed the unsupported "conservative" and "well below the 247% figure" characterization from the savingsRate note. Softened the problem paragraph's unsourced "as soon as the data lands" and "rather than publishing a league table" claims. Reworded faq[3]'s opening to "In production at two named airlines on one vendor's platform", consistent with adoptionStage: early-adopters.
- 30 September 2026: Unpublished by an automated review workflow (independent skeptic review).
- 29 September 2026: First published