.svg)
At OutSystems World Tour Las Vegas 2026 on October 7, John Issa, Head of Technology at Ascent Developer Solutions, and Jared Wesley, VP of US Go-to-Market at Moxie, presented "The engine behind every loan: Evolving a live lending backbone on ODC" on the Tech Stage. The session showed how Ascent, a construction real estate lender with more than $4 billion in loan originations since launch, built Alpine, its lending platform, as one shared workflow on OutSystems Developer Cloud (ODC), and how Ascent and Moxie keep changing it while loans are live.
The session spoke to a question OutSystems put to the whole room that morning. Opening the keynote under the theme "own your agentic future," CEO Woodson Martin said the demands on every business will keep changing, and asked: "how will you deliver the engine to stay ahead of that change in your agentic future?" Ascent's answer, from inside a live lending system, was groundwork first.
At the start, Excel supported loan structuring. As Ascent grew, more people worked on each loan, and the business needed a shared workflow that made changes easier to trace and kept teams working from consistent information. So Ascent built Alpine.
Engineering was never the bottleneck. As John explained, the hardest part was getting the workflow right: who owns each decision, what happens when something changes, and how it affects the next team. A platform doesn't solve those hard things. It gives your best people more time for them. In John's words: "OutSystems gives our team more time to focus on the business problems we need to solve."
It's the reality Woodson put first in the keynote: delivery beats speed. As he said, "generating code faster doesn't matter if you can't put it into production." The way lenders build has changed, from spreadsheets, custom code or a rented loan origination system (LOS) to a platform with faster building and AI-assisted review and testing. The hard things haven't: knowing what you're building, getting the requirements, and making sure every team knows the workflow.
The keynote's headline announcement connects directly to how Ascent already works. Luis Blando, Chief Product and Technology Officer at OutSystems, announced that OutSystems Agent Experience is generally available. According to OutSystems, it opens the platform to any AI coding agent, including Claude Code, Cursor, Codex and Kiro, grounded in each organization's own context. Luis said 140 organizations helped co-design it in early access.
Ascent took part in the early access program for the OutSystems MCP server, part of Agent Experience. In John's words:
"We're part of the early access program for the MCP server and we've used it with Claude, Codex, and Cursor. We use it heavily for prototyping, building libraries and sandbox apps, testing, and code review. It helps us validate functionality and overall user experience."
He also shared where it's heading: "As support for larger apps and the overall server matures, I expect this to become our primary development approach."
Now any team can do the same. In Woodson's words: "Agent Experience is generally available to every OutSystems customer and any developer."
The keynote's theme was ownership. Ascent's decision started from the same idea, applied to lending: know what's yours. For Ascent, that meant the terms, the math and the workflow.
Two questions settled it. Could an off-the-shelf system be configured to fit? Not well enough; a rigid system ends up dictating the deal. Could Excel keep working as a shared tool? Not once more people worked on each loan; it was outgrown.
So Ascent built, and chose OutSystems to build on. For John, security and how easily an application could get up and running were major factors, along with fast prototyping: putting ideas in front of the people who would use them early, to validate what the business needed.
The approach had three steps. Prove, with a small proof of concept on a subset of the work. Build the structuring and the math, which is where the flexibility lives. Evolve, changing the system live, with loans in flight. During the build, Excel and Alpine ran side by side until structuring moved into Alpine.
The result is one version of the loan. In John's words: "Alpine keeps the loan's context intact as work moves from one team to the next." Context was a word the keynote kept coming back to, as the thing any AI needs before it can help. In Alpine, that context lives in one loan record, fed by Structuring, Pricing & Fees, Conditions and Approvals. Data is entered once, reviewed in one place and approved with the loan.
In the keynote, Luis Blando asked the room to raise a hand if their company had one system almost everybody depends on and nobody wants to touch. Plenty of hands went up. Alpine is a system the business relies on every day, with loans at every stage. Ascent and Moxie change it anyway, release after release.
That was Moxie's part of the session. Moxie's team on Alpine is led by Miguel Aparício. For them, the job is never what to build. It's how to change a live lending platform without disrupting the business. The session set out the evolution playbook Moxie uses, part of Moxie's delivery work:
As John put it: "We work very closely with Miguel and the Moxie team to make sure the business context behind each change is clear."
Region and branch show how it works. As Ascent grew across the US, it needed both on loan records to track and report on its originations. The change shipped behind a feature flag, with no downtime. Existing loans were updated in the background and checked before switch-on, and new loans get region and branch from day one.
In John's words: "Our people need to understand what changed, who changed it, and what it means for their work."
Alpine's change history records every change, on every loan, across Structuring, Pricing & Fees, Conditions and Approvals. Each record holds five elements: field, old value, new value, who and when. Every change traces to a person and a moment, so when someone asks what changed on a loan, the answer is a lookup, not a meeting. Today, a limited group of people at Ascent can see this history.
Woodson described the teams getting real value from AI as those building it into workflows "where the guardrails and the human checkpoints are already working." At Ascent, those checkpoints exist today: a name on every change, and approvals with a name on them, in place before any agent goes near a loan file.
The slides about agents were marked "Potential." They describe what could be possible, not something in production.
Alpine's structuring engine keeps the rules in one place. A person can read the rules and change them with approval. An agent, if one were used, would only read them, never rewrite them, and the same controls would apply: the same rules, the same change history and the same approvals, with a person accepting first. It fits another reality from the keynote: not every task needs AI. In loan math, there's no room for guesswork.
The session also sketched how a task could flow if it were handed to an agent: verify documents, bring in data, check required steps. If the task needs no judgment, the work would move on because the criteria are met. If it does, the authorized person would decide.
Readiness for production would take business validation, technical validation and an independent security review, and John added strong evaluations and clear human ownership. A mortgage insurer set a similar bar in the keynote: MGIC's Director of Enterprise Architecture said AI that goes in front of the lenders MGIC serves has to be "defensible, deterministic, and reproducible."
What comes next sits on three foundations, in this order:
Liam Doyle, SVP of Product Management at OutSystems, cited McKinsey data in the keynote: 88 percent of enterprises use AI, and only 39 percent see an impact on EBIT. His explanation was that most pilots stay at the edges of the business. Alpine is the core, the engine behind every loan, and that's where Ascent is laying its groundwork.
The session's closing slides carried two statements from John Issa. The first: "For us, the value is control over our roadmap and the ability to keep evolving the platform." The second, his closing line before questions: "Agent autonomy starts with a foundation of trusted data, clear rules and approval controls."
For Ascent, owning its agentic future starts with owning the groundwork.
John Issa is Head of Technology at Ascent Developer Solutions, a construction real estate lender with more than $4 billion in loan originations since launch.
Jared Wesley is VP of US Go-to-Market at Moxie, an OutSystems-exclusive partner and OutSystems Elite Delivery and Gold Partner, founded in Portugal in 2015. Moxie helps clients move from strategy to production on OutSystems, from advisory and architecture to delivery, QA and long-term support.
John Issa, Head of Technology at Ascent Developer Solutions, and Jared Wesley, VP of US Go-to-Market at Moxie. They presented "The engine behind every loan: Evolving a live lending backbone on ODC" on the Tech Stage on October 7, 2026.
Alpine is the lending platform of Ascent Developer Solutions: one shared workflow built on OutSystems Developer Cloud (ODC), where structuring, pricing and fees, conditions and approvals feed one loan record with a change history.
Yes. OutSystems made Agent Experience generally available on October 7, 2026. According to OutSystems, it opens the platform to any AI coding agent, including Claude Code, Cursor, Codex and Kiro. Ascent took part in the early access program for the OutSystems MCP server and uses it with Claude, Codex and Cursor for prototyping, testing and code review.
General availability of Agent Experience; Agent Spotter, a free tool that recommends where agents could add value in existing OutSystems applications; a strategic collaboration agreement with AWS; and an Insurance Agent Kit on OutSystems Forge, with Agentic First Notice of Loss coming in November.
No. The session presented agents in the lending workflow as potential only. Ascent uses AI coding agents for prototyping, testing and code review. If agents are used in Alpine, they would read the rules without rewriting them, write to the same change history and wait for a person to approve first.
The slides and the groundwork checklist from the session are at moxie.pt/vegas. Planning to change a system your business relies on every day, or to bring AI coding agents into your OutSystems delivery? Talk to us.