Every June, Articulate’s IT team pauses its roadmap for a hackathon. This year, the theme was MCP, and the results are already shaping how the company as a whole uses AI.
A tradition built for a moment like this
Articulate’s IT department has run a hackathon for three consecutive years. The format is simple: clear a day, gather the team, and go after a real company problem instead of a hypothetical one. “Generally, we try to solve a company problem,” says Darren Brink, Senior Director of Enterprise Applications and Architecture. “That’s what we use the hackathon for.”
What makes Articulate’s version different is what happens after the trophy is handed out. The winning team’s idea gets carved out as a project, refined, and then shipped. “Every single year, first place not only gets the prize and the accolades of winning, but they get to see their item go into production,” Darren says.
This year’s twist: build it for Claude, govern it with Workato
This year’s hackathon took place at a pivotal moment: Articulate was rolling out Claude for Enterprise, and the IT team wanted a way to let Claude and other LLMs access company systems without giving up control over what those systems actually do. The answer was MCP, with a rule attached. LLMs could read and query freely through MCP, but anything that created or changed a record had to be routed through Workato, so IT maintained deterministic control over who could take which action.
That single rule reframed the day. Instead of a scramble to connect as many tools as possible, teams worked to turn recurring, well-understood requests into reusable, governed skills that plug into the company’s existing IT genie, adding capability without creating a sprawl of raw, ungoverned connectors.
What teams actually built
The ideas that emerged from the day ranged across the IT stack. Some teams added new skills to the existing IT agent, closing gaps in default capabilities they’d noticed since it first launched. Others built new MCP servers from scratch. One recurring pain point drove a standout project: as a fully remote, Slack-heavy company, it can be hard to track when decisions are made. A team spent the day building MCP-powered notifications to ensure key decisions surface automatically to the people who need them.
VP of IT Taruna Rai points to speed as the standout. “We had set aside the whole day for allowing the teams to get together, collaborate, brainstorm on their idea, and then play around and see what they could build,” she says. “It’s nice that the team is going to be able to come out with actual demos — not slides, but working prototypes we can quickly get out to the organization.”
Both Taruna and Darren credit Workato’s on-site team with making the day count. The Workato team ran hands-on training the day before, and then stayed on to coach, troubleshoot, and push back. “They come up with ideas, push back on concepts we’re thinking about, and say, ‘Is that the right way to do this?'” Darren says. “They’ve been a great model for how we do this better.”
That coaching matters because a hackathon project still has to survive contact with production. “It really helps us not only let the team learn the new tool very quickly, but also helps guarantee that when we come out of the hackathon, we’re going to have something shippable,” Taruna explains — built in a scalable way from the start, “despite being a hackathon.”
The ripple effect
The most surprising part of the day for Darren was what happened after the whiteboarding. Once other departments saw how fast IT could move, requests started pouring in: “Can I have access to this? Can I use this tool?” That appetite led Articulate to launch a Workato Center of Excellence to bring other teams onto the platform, and this year, departments outside IT are asking to run their own hackathons.
Taruna sees the same shift in how stakeholders frame problems in the first place. Conversations that used to start with “we need to build an integration” or “we need to buy a tool” now start with “how can we leverage AI to solve this”—a cultural change she calls well-aligned with where the company is headed.
The takeaway for other IT teams
Darren’s advice to other architecture and IT leaders considering their own AI or agentic push is to resist the urge to do everything at once. “Start small. Don’t try to boil the ocean,” he says. “Come up with a very concentrated problem: what’s a business problem you’re trying to solve? And then move fast.” For Articulate, that discipline, one focused hackathon, a governance rule for MCP, and a habit of asking whether something should be an MCP server, an agent, or something else entirely, is what turned a single day of building into a durable, company-wide way of working with AI.