Thread logo and text

Thread is an early stage startup that provides intelligent service delivery to Managed Service Providers (MSPs). MSPs are businesses that serve other businesses's IT needs. Founded in 2019, Thread gives MSP technicians instant context about the reported issue, history, and suggested fixes to help IT teams work better, faster.

GIF of step 1

With no onboarding workflow,

Thread's Customer Success Managers had to set up multiple hour long calls with new MSPs and walk them through setting up Triage Agent, one of Thread's Pro plan offerings that handles incoming IT requests by communicating with clients before handing the issue off to a technician.

To implement a more guided approach to facilitate ease and self-service, I worked with the team to add a feature that used AI to simulate a conversation between the triage agent and a sample client based on the MSP's most common IT issues (that we also analyzed and identified). From there, the user could review and provide feedback that the AI model used to update its behavior dynamically. The loop of reviewing, providing feedback, and testing repeated three times to create stickiness and increase familiarity with a new tool without adding too much cognitive load.

GIF of step 2: select a board to analyze, generate 3 top themesGIF of step 3: review and provide feedback loop

To add delight into the experience,

I designed a way to personalize the agent through a customized name and avatar. I also designed micro-interactions like a text-shimmer effect and smooth transitions to enhance the luxurious feel of the experience. These pieces were integral to sustaining the standards the design team held ourselves to around design craft and product polish. (FYI: I worked with an brand designer for the avatars and illustrations)

GIF of step 1

To help users prompt better,

I designed a way for the user to leverage AI to better write prompts with agent friendly language, which helps the admin achieve more expected behavior out of their agent and increase user satisfaction. In addition, we also explained why the AI made those changes, which adheres to our core AI-design principle of transparency.

GIF: optimize prompt feature

We launched in June 2026

and tracked adoption for the next 30 days. Adoption was defined as the percentage of partners who deployed the triage agent to at least 4 of their clients and had 34+ agent sessions across that client base.

68% conversion, 7 days to adoption, 24 average end customers

Designing for AI

is a constant, evolving experience that will continue changing as people's relationship with AI grows and matures. Triage Agent is just the tip of the iceberg on AI features I work on Thread.

Thanks Amine (PM), Chakib (Eng Lead), and Daniel (Eng)