The patchwork mentality in travel tech

The problem we don’t name

In travel, we often mistake more inventory for more capability. I learned the hard way that what actually breaks at scale is not access to flights or hotels—it’s the plumbing between systems. Most agencies are stitched together with a dozen tools and a handful of brittle integrations. The quiet tax is paid every day in spreadsheets, copy-paste, and “could you resend that CSV?” I’ve seen teams burn roughly 30% of their margin on what I call data janitorial work: moving, reconciling, and rekeying the same information between disconnected silos.

I came into travel from AI and analytics. For years our work at ElkanIO looked like a research lab—experiments, models, pipelines. The constant was the same: if the underlying architecture is fragmented, every new feature becomes a one-off project and every handoff becomes a failure point. Margins don’t erode because your people aren’t working hard. Margins erode because the system asks them to work against its shape.

What patchwork looks like day to day

• Bookings created in B2C that never quite match what finance expects in the ledger.
• Supplier outages that cascade because routing is manual and reactive.
• CRM that can’t see the operational state of a PNR, so agents live in tabs.
• “Temporary” scripts that become permanent because nobody owns the schema.

None of this is glamorous, but it is the day-to-day reality of scaling travel businesses. You don’t beat patchwork with more patchwork. You beat it with architecture.

Architecture over inventory

Inventory is table stakes. The differentiator is a coherent system that treats every product—flights, hotels, activities, insurance, visas, cabs, ferries, cruises—the same way at the data layer. When the core model is unified, you can change the presentation without changing the truth. When the schema is consistent, you can automate routing, reconciliation, and yield.

That is why we built Synapse as the unified nervous system—one schema across products and verticals, one API for the rest of the platform to trust. It’s not a single monolith; it’s a stable core that reduces the degrees of freedom where inconsistency breeds.

Self‑healing instead of firefighting

Even with a unified core, integrations fail. Carriers change formats, suppliers go down, edge cases appear. The answer isn’t “open a ticket” and wait; the answer is an agentic loop that continuously improves the connective tissue:

watch → propose → prove → promote → route

• watch: observe payloads, performance, and exceptions in the wild
• propose: generate a candidate fix or mapping change
• prove: validate against real traffic and test data
• promote: apply the change safely to production
• route: select the best supplier path given price, reliability, and policy

When the system learns like this, your team stops doing janitorial work and starts doing customer work.

What this enables

With architecture in place, you can:

• Reconcile bookings and payments automatically because finance sees the same truth as ops.
• Survive supplier instability because routing is intelligent and resilient.
• Add a new vertical without inventing a new language each time.
• Expose consistent surfaces to Nucleus (B2B), Nova (B2C), Orbit (corporate), Cluster (groups), and the rest of the stack without bespoke glue.

The patchwork mentality is comfortable because it’s familiar. But comfort is expensive. I’m building the alternative: a travel platform where architecture—not headcount—does the heavy lifting.