From Digital Archaeology to Real-Time Triage: How Wayflyer Built a Self-Tuning Support Flywheel with Dust
- Industries
- Financial ServicesB2B SaaS
- Company Size
- 201-1000
- Departments
- Customer SupportITEngineering
Key Highlights
- Zero-friction triage in under a minute: Research completed before an engineer opens a ticket.
- Self-tuning knowledge flywheel: Runbooks automatically update post-resolution, keeping institutional context fresh.
- Over 90% reduction in runbook maintenance time: Automated drafting removes documentation overhead.
- Decentralized AI empowerment: Empowering the team to build custom agents drove near 100% monthly workspace adoption.
Support Loads Shift Like the Weather
At Wayflyer, speed gets capital out the door, while accuracy is what actually builds trust. As a Dublin-based fintech, the company analyzes real-time sales, marketing, and banking data to deliver financing decisions to businesses in a matter of hours. Delivering billions in funding to fast-growing brands requires operational infrastructure that moves just as fast.
Support loads in a scaling fintech shift like unpredictable weather. New products, market expansions, and API updates constantly alter issue patterns - one week bringing upstream provider degradations, the next a spike in merchant integration edge cases.
In a remote-first engineering organization built on autonomy like Wayflyer, knowledge naturally tends to scatter. With critical context fragmented across Notion, Slack, Jira, and individual memory, engineers were forced to play detective on every incoming issue. Gathering context meant spending valuable time on repetitive manual research. This was a core challenge for a distributed team striving to maintain operational speed.
Decentralized Building: Solving Friction at the Source
Rather than routing AI initiatives through a central data engineering queue, Wayflyer deployed Dust as a shared platform with minimal gatekeeping. This empowered the teams closest to operational friction to build custom solutions directly, without waiting for dedicated AI capacity or central approvals.
“The model is decentralized. Any team can build and publish their own agents.” — Mike Smith, Engineering Support Manager at Wayflyer
For support engineering, this meant directly tackling their research bottleneck. Because the engineers experiencing daily context-switching were the ones building on Dust, they designed tools custom-fit to their exact triage workflow - building a dynamic system that adapts as support demands evolve.
The Self-Tuning Flywheel: How Two Agents Close the Loop
To tackle the shifting support landscape, the team built two interconnected Dust agents that form a self-reinforcing knowledge loop:
- Support Triage Agent: The moment a new Jira ticket opens, the agent triggers automatically. In about a minute, it scans ticket histories, Notion runbooks, Slack conversations, and queries internal tools via MCP. It diagnoses issues using live, case-specific system data rather than relying on runbook generalities. The result is an internal summary with recommended triage steps and cited sources. Manual research time drops to virtually zero before an engineer even opens the ticket.
- Support Doc Agent: When a ticket resolves, this agent checks if the solution exposes a gap in existing knowledge. If a runbook is missing or incomplete, it automatically drafts an update. Engineers can also tag the agent at the end of a Slack thread to generate a structured runbook instantly from the discussion.
The Support Doc Agent delivers a massive efficiency win, cutting runbook creation and maintenance time by over 90%. More importantly, it powers the engine behind real-time triage: by continually capturing new resolutions, knowledge updates itself in the background as support patterns evolve, ensuring the Triage Agent always draws from fresh, accurate context.
Delivering Speed and Quality at Scale
With context research and runbook maintenance running automatically in the background, support engineers no longer have to compromise.
“Engineering traditionally forces a choice: move fast or focus on quality. At Wayflyer, we want both. Dust shifts our efficiency frontier outwards - now we can do more while maintaining our high standard of excellence.” — Mike Smith, Engineering Support Manager at Wayflyer
Support engineering’s flywheel is just one concrete example of what happens when teams are empowered to build for themselves. Because Dust was opened across Wayflyer without gatekeeping, other departments quickly followed suit - building custom agents for their own distinct workflows. By letting the people closest to the friction solve their own problems, Dust reached nearly 100% active monthly adoption across the entire Wayflyer workspace within six months of rollout.
Interested in learning more about how Dust can help your team? Visit our solutions page or reach out to our sales team.
