tray.ai

AI broke the old data pipeline model. New joins. New sources. Constant change. Legacy ETL wasn’t built for agent-driven workloads. We just introduced Data Engineering in Tray, with a new SQL Transformer for in-flight bulk data shaping. 🔗 bit.ly/4qX2dvA

Behind every great AI agent, there's an architect. Vignesh Velan built Zuora's Customer Briefing Agent — 23 hours to 5 minutes, 170+ hours back to the team weekly. 🏆 Final Agenty: The Architect Award.

AI orchestration is exposing the limits of SaaS-era integration platforms. Many were designed for predictable APIs and deterministic workflows. AI demands something very different. Tray CEO Rich Waldron explains the shift in our latest blog: bit.ly/47sq2ET

Most teams are asking the wrong question. It's not "Which LLM should we use?" It's "How do different models perform across real workloads?" We broke down ChatGPT vs Claude vs Gemini 👇

Tray has been named a Visionary in the 2026 Gartner Magic Quadrant. But the real signal is the shift underway. Our CEO explains why this could be the last Magic Quadrant focused on traditional iPaaS: bit.ly/47sq2ET

Podium used to take up to 14 days to set up a single integration. So they built a self-serve integration marketplace instead. What they found was that customers who connected their tools behaved very differently from those who didn’t. Full story: bit.ly/4ocXdSc

What happens when AI becomes a shared service inside the company? @jwpepper shifted from scattered chatbot use to governed AI workflows employees actually rely on. Watch how they did it.

Congrats to the winners of the first-ever Agentys 🏆 @kchakkarapani and @Zuora Matt C. and @Life360 Marcus Dubreuil and @jwpepper Ramiro Meyer, Gabriel Laporte, and the team at @useapolloio Amazing work turning ambitious ideas into real AI agents running in production.

It’s awards season. So we gave out a few of our own. 🏆 Welcome to The Agentys. Celebrating the teams building real AI agents in production. From orchestrating hundreds of SaaS apps to resolving IT issues in seconds. See who took home the trophies. 🎬

Apollo’s CEO set a productivity mandate. IT responded by identifying 80+ opportunities for AI agents across the business. What happened next turned one experiment into a blueprint for how work gets done. See the story. ⬇️

HubSpot runs revenue. But what runs everything around it? Watch. ⬇️

We’re growing the team at Tray. Open roles: • Office Manager (London) • Senior Product Marketing Manager (Bay Area) Come build the future of AI orchestration with us. See the roles: bit.ly/4ojqf3B

MCP makes it easy to connect models to real systems. That also means context can influence real actions. As agents gain privileges, governance has to live outside the prompt. Rich Waldron explains why MCP changes the enterprise attack surface. 🔗 bit.ly/3ZF5mVN

MCP is here to stay. But right now it feels a bit like a gold rush. New tools and “agent platforms” popping up everywhere. @jwpepper's architecture team took a different path: start with the iPaaS layer already running your integrations. Here’s why.

Many teams feel like they’re behind on AI. So the first move is often turning on tools like Copilot or Gemini. That’s just the beginning. In this clip, Nate Gemberling and Stephen Stouffer break down what teams are actually experiencing with AI adoption.

🚨 TOMORROW 🚨 60% of AI projects fail because of data. Join our Tray First Look to see how teams are fixing the AI data supply chain bottleneck: bit.ly/4slq9tP

Behind every AI system is someone who decided a problem was worth fixing. For #InternationalWomensDay, we’re highlighting one of those builders. Tray sales engineer Maura St Clair on building useful AI agents.

Daylight saving stole an hour. Your IT team was already short on time. Password resets. Ticket triage. Access requests. An ITSM agent handles it in seconds. If the clock jumps forward, your service desk capacity should too. Run your numbers here: bit.ly/3NnrJwf

More bots wasn’t the answer. @Life360 moved to an “agent-of-agents” model. One orchestrator agent routing to specialized sub-agents behind the scenes. One entry point. No bot confusion. Built-in governance. See how they scaled AI without sprawl.

There’s a gap between MCP scope and enterprise governance. That gap is where over-permissioned agents and hidden risk live. MCP needs a governed layer between agents and systems. Here’s what that looks like: bit.ly/493hRjR

MCP doesn’t sprawl because teams are careless, but because it’s easy. @jwpepper moved from scattered MCP experiments to a governed, centralized model with Tray Agent Gateway. Real IT agent. Real guardrails. Production scale. Watch the full session: bit.ly/4rqnfE5