Jeppesen ForeFlight Airflow Is Less About AI Agents and More About Who Controls the Data Layer Beneath Them

The headline from Paris on October 6 was straightforward enough: Jeppesen ForeFlight unveiled the first of hundreds of planned agents for its Aviation Operating Intelligence platform, Airflow. A follow-up at NBAA-BACE ten days later added business aviation-focused capabilities, including safety management, flight tracking, paperless cockpit operations, runway performance, fuel, aircraft privacy, and new Airflow agents. Taken together, the announcements read as a rapid expansion push. But the more consequential decision in all of this isn’t which agents launched first — it’s the platform architecture Jeppesen ForeFlight has chosen to build them on.

The Architecture Is the Argument

Operators decide the pace of adoption, the level of automation, and whether to run their own agents, third-party tools, or Jeppesen ForeFlight’s agents — or any combination of the three. That’s an open-platform posture, and it’s a meaningful departure from the historically closed nature of the incumbent aviation software stack. The claim behind it is equally specific: every option reasons over the same certified aviation data and each operator’s own rules. That’s where the real bet is. Jeppesen ForeFlight isn’t arguing that its agents are the best agents — it’s arguing that no agent is worth much without access to high-quality, certified operational data, and that it controls that layer.

Airflow agents reason in the context of real flight operation workflows, draw on Jeppesen ForeFlight’s certified aviation data, provide transparency into the reasoning behind each recommendation, operate within established Safety Management System frameworks, and keep humans in control of decision-making. In practice, that means the platform is trying to do something the broader AI-in-aviation conversation hasn’t fully resolved: separate the model from the data substrate, and let the data substrate become the defensible moat.

I think that’s the smarter long-term play. From time spent on the vendor and delivery side of flight ops software, the hardest part of any AI integration isn’t the model — it’s getting the underlying data into a state where an agent can actually reason over it without surfacing nonsense. Certified aviation data, by definition, has been through a validation process that operational AI genuinely needs. Any third-party agent that plugs into Airflow inherits that foundation rather than having to rebuild it.

The Cockpit Angle That Deserves More Attention

Most of the initial Airflow coverage has focused on the OCC-facing agents — crew trip trading, disruption recovery — because those were the first to ship. The first tools are aimed primarily at crew scheduling and operational disruptions rather than functions used directly by pilots. Fair enough. But the NBAA-BACE announcement included something worth sitting with: ClearNOTAMs, an aviation-trained AI that translates abbreviated NOTAMs into plain language summaries throughout the Jeppesen ForeFlight experience, with the original source data always shown alongside.

That’s exactly the kind of lower-stakes, high-value cockpit application that AI should be proving itself on first. NOTAMs are notoriously dense, frequently misread, and a real friction point in preflight briefing — not a catastrophic risk surface, but a meaningful one. An AI layer that renders NOTAM content in plain language while keeping the source visible is genuinely useful to a pilot running through a briefing, and it doesn’t ask anyone to delegate a consequential decision to a model. The agent earns trust by being accurate and transparent on something manageable before it ever gets near anything higher-stakes. That sequencing is right.

The company also developed an MCP Connector that integrates with ForeFlight mobile for general aviation — another signal that the architecture is meant to extend toward the pilot-facing edge of the operation, not stay locked in the back office.

What This Means for the Competitive Landscape

Five business aviation partners have joined the Airflow co-development phase: Axis Aviation, Aircraft Service Group, Thrive Aviation, JetHouse, and Sparfell. That’s a modest but concrete early-adopter set, and the co-development framing matters — it’s the classic way established aviation software vendors socialize new capabilities before broad release, giving the platform real-world feedback while managing the liability exposure of early AI.

The broader competitive implication is that Jeppesen ForeFlight is trying to establish Airflow as the AI operating layer for the flight ops stack, not just another feature inside an EFB. Whether that succeeds depends on how many operators actually bring their own agents or third-party tools into the platform — which is the proof point the open-platform thesis requires. A platform with one provider’s agents is just a product with extra branding. A platform where operators genuinely mix and match is a different kind of infrastructure altogether.

From my perspective on how airlines evaluate and adopt flight ops software, the flexibility argument Airflow is making is well-calibrated to the current market mood. Airlines aren’t ready to hand the wheel to a single AI vendor’s roadmap, and the ones with mature procurement teams know it. An architecture that lets them start narrow, validate, and expand on their own timeline is a more credible ask than one that says “trust us, deploy broadly.” The agents are the visible part. The data layer and the adoption model are what Jeppesen ForeFlight is actually selling.

Jeppesen ForeFlight Airflow Is Less About AI Agents and More About Who Controls the Data Layer Beneath Them

The Architecture Is the Argument

The Cockpit Angle That Deserves More Attention

What This Means for the Competitive Landscape

Sources