How Raycon made AI accessible to their whole team
In three months, Raycon turned scattered AI experiments into shared company infrastructure: teammates collaborating in shared Spaces, 127 reusable skills, and 26 automations running unattended.
Raycon had a clear idea of what it wanted with AI: a positive feedback loop that got smarter with more team usage.
The hard part was getting AI into the hands of the whole company. Early adopters were effective, but their work stayed locked in individual chats. Building the system in-house was priced and rejected. Point tools could produce work, but they could not accumulate it into something the company owned.
Three months after adopting Type, Raycon runs 26 unattended automations, ships website changes from Slack to production, and has a team of 50+ people building with AI together.
“Now you're not relying on just your AI power users. With Type, the whole team benefits.”
Team-wide AI initiatives
Raycon was early and deliberate about AI. Ray Lee, the co-founder and CEO, set the AI north star to help the team make higher-quality, faster decisions that make an impact on the P&L.
People who found AI overwhelming needed a way in. The work of people already good at it also needed somewhere to live, so the rest of the company could inherit it.
AI, Organized by Departments
Raycon's Space list now reads like an org chart: Analytics, Marketing, Product, Website, Supply Chain, CX, Creative, Business Systems, Operations, and Executive.
“The vision of what we sought in a team-based AI infrastructure aligned very closely with what Type offered.”
Replacing a tool that could not compound
Two of Raycon's three AI initiatives, marketing and website, were already running on Viktor. It could do the work. Keeping and extending that work was the problem.
Landing pages lived on disconnected preview URLs. When Raycon connected Type to its CRO channel, all four active landing pages were rebuilt as live, editable Type artifacts within days. They now sit alongside the Intelligems test data, Notion roadmap, Shopify store, and GitHub repo that govern them.
The landing pages stopped being deliverables and became assets. Iteration no longer requires a handoff.
Shared systems that report, decide, and ship
26 unattended automations
Daily paid media reports, a six-page performance PDF, product-level reporting, inventory health checks, ratings scorecards, and executive briefs now run on a schedule.
“It not only just has the data points, which we did before, it also adds the context of the business, of why this happened and what other things are going on.”
CRO that ships from Slack
Raycon's website Spaces account for 665 runs. The same shared surface reads Intelligems results, coordinates the roadmap, and ships changes to the live store.
“We've been able to push changes extremely quickly — we have the power and flexibility to go from Slack to production.”
A counterpart in the room
Teams pull Type into Slack threads and live discussions to validate assumptions, answer open questions, and keep decisions moving with company context attached.
“Now when making decisions, we just say ‘Let’s ask Type’ instead of waiting days for an analysis.”
Accessibility, not fluency
For the Raycon team, the gap was not AI fluency. It was access.
By August, the rest of the company was doing more work in Type than the three people who brought it in. The month's heaviest user had not been one of those original power users.
Since July 1, nine people have worked in Analytics, seven in Marketing, and six in Product.
“It's made it much easier to leverage AI for people who were overwhelmed by it.”
Usage shifted beyond the original power users
| Month | Power users | Everyone else |
|---|---|---|
| May 2026 | 103 | 11 |
| June 2026 | 419 | 220 |
| July 2026 | 637 | 445 |
| August 2026 estimate | 364 | 928 |
August estimates extrapolate first-half usage across the full month.
Three months in, adoption is still climbing
Monthly usage grew 9x from Raycon's first partial month to July. Based on first-half usage, August is estimated to reach 1,758 agent runs, another 31% increase.
| Month | Agent runs | Active Spaces | People active |
|---|---|---|---|
| May 2026* | 141 | 5 | 4 |
| June 2026 | 896 | 20 | 17 |
| July 2026 | 1346 | 23 | 19 |
| August 2026 agent runs estimate | 1758 | 16 | 33 |
- Peak active Spaces
- 23 in July
- People active
- 33 in August
* May begins on the 20th. August agent runs estimate extrapolates first-half usage across the full month; active Spaces and people are actual first-half counts.
“The combination of accessibility, cross-collaboration, and persistent memory is where Type shines.”
127 skills, including 38 built in the last 30 days. The team is teaching Type more now than when it started.
Let Type set up your first space for you.
In just a few minutes you'll be rolling with your first agent.
Start a 2-week free trial“The more the team uses it, the better the entire infrastructure gets. Your whole departments can benefit, rather than just individual team members.”
Ray Lee, Co-founder & CEO