162: Rich Waldron: How to build and manage AI agents from a single, composable platform without coding

What’s up everyone, today we have the pleasure of sitting down with Rich Waldron, Co-founder and CEO at Tray.ai.

Summary: Marketing ops folks stand at a crossroads where iPaaS platforms and AI agents are colliding in crazy ways. Rich pulls back the curtain on what happens when workflows become agent "skills": Imagine your carefully built automations transformed into autonomous assistants that diagnose tech issues, provision applications, and manage complex Salesforce campaigns without manual intervention. Your marketing stack could suddenly act like a "junior admin" on demand, while you focus on strategy. The explosion of AI features has turned martech leaders into "AI referees" juggling competing vendor tools, yet those who master both fundamentals and experimental curiosity become "10X automation heroes" - the first teammates that are called when problems need solving. As Rich explains, career security comes from momentum, not stability.

About Rich

• After University, Rich spent several years building different projects in the UK which included a web agency, a media company and a mobile app for social gatherings
• Tray was officially founded in 2013, bootstrapped by selling Wellington boots on eBay – the early product idea was email automation but pivoted to enabling less technical people to utilize APIs to integrate their tech stack
• Alongside his 2 co-founders, they spent the better part of 4 years building the product and raising a seed round in 2015. Between 2018 and 2020, Tray grew from $500k to $20M ARR
• Today, Tray processes Billions of transactions across the platform every month and they’ve gone all in on the composable AI integration and automation movement

The Rise iPaaS and AI Orchestration

iPaaS exploded because enterprise suites were too slow to open up their integration capabilities. CDPs made similar mistakes with rigid architectures, birthing today's composable alternatives. Every software system eventually faces the same primal challenge: intercommunication. Rich recounts how this pattern also repeats throughout computing history with startling consistency. Monolithic ERPs dominated early landscapes, where engineers cobbled together custom connections between internal components. These hand-built bridges crumbled easily, leaving teams scrambling for standardized frameworks that could withstand daily operational stress.

As specialized software proliferated around these central systems, integration pressure mounted. "We're still not that far through on adopting the cloud," Rich points out, puncturing the tech bubble many of us live in. While cloud technologies feel omnipresent to industry veterans, countless organizations remain firmly planted on physical servers. This reality created distinct evolutionary phases for iPaaS:

On-premise to on-premise connections (the original integration challenge)
On-premise to cloud bridges (MuleSoft's territory) 
Cloud-to-cloud orchestration (where Tray focused)

Each phase demanded fundamentally different architecture. Cloud applications introduced unique payload structures, execution patterns, and API designs that rendered previous integration approaches obsolete. "Every application now has an API," Rich explains, describing how this technical shift triggered organizational transformation. Marketing departments grew increasingly technical, with marketing ops professionals discovering they could craft custom experiences by tapping into these newly accessible APIs.

> "iPaaS has to evolve because if your iPaaS was built purely for an era when AI wasn't a consideration and your customers are now suddenly saying, 'We're looking at how we infuse AI in these processes,' the requirements have changed again."

You've likely witnessed this evolution in your own organization. Remember when connecting two systems required an IT ticket and weeks of waiting? Now your marketing team builds automations while the sales team creates their own customer journey orchestrations. Technical power diffused across departments, democratizing integration capabilities previously locked behind developer expertise.

Today's iPaaS platforms face their greatest evolutionary pressure yet: AI integration. Rich describes how existing processes built on traditional platforms now crumble under AI's weight. Semantic analysis, novel reasoning models, and entirely new integration approaches have rewritten the rules. iPaaS vendors who built for the pre-AI era now race to adapt as customers demand intelligent workflows. The platforms that flourish will embrace AI as a core architectural principle rather than a bolted-on feature.

Key takeaway: Evaluate your integration platform based on whether it was (re)designed for today's AI-centric landscape or simply patched to accommodate it. The most effective iPaaS solutions evolve alongside major architectural shifts rather than struggling to catch up after they've occurred.

What Makes an Agent Truly "Agentic" Beyond the Marketing Hype

The AI agent landscape is blurring with contradictions and wild claims and it’s only going to get crazier. While vendors plaster "agent" labels on everything with an algorithm, Rich isn’t worried about definition. The terminology matters far less than what these systems actually do.

> "The AI isn't just reasoning over a set of data, but it's actually going and taking action on a user's behalf... I've done the response for you and I've handled the follow up and I've gone and filed this over here, and it's actually carrying out a series of actions based on the reasoning that occurred in the first place."

AI agents take autonomous action. They handle support tickets end-to-end. They file documents. They complete multi-step processes without human intervention. They execute rather than suggest.

Tray's team experienced genuine goosebump moments when they combined their connector infrastructure with LLM reasoning. You could almost hear the click as puzzle pieces fell into place. Their ten-year vision suddenly materialized before their eyes:

Semi-technical staff performing complex cross-organizational tasks
Teams breaking free from application limitations
Workers escaping data accessibility problems
AI executing the best next steps, not just recommending them

This capability triggered an immediate "holy shit" reaction during internal testing. Everything changed in that moment. The strategic implications struck like lightning: adapt or die. Many category leaders fail exactly here, at this precipice of change, clinging to outdated paradigms while disruptive innovation rewrites the rules.

The adoption curve is also likely to be shockingly steep. Century-old enterprises with conservative DNA are already running AI workloads in production using Tray. Some skipped entire technological generations, leapfrogging directly into AI implementation. They've dumped their data into databases, layered AI analysis on top, and built reactive systems around the outputs. The comfort level with these technologies has accelerated across industries at a pace that defies conventional adoption timelines.

When Tray rebranded from tray.io to tray.ai, they acknowledged that connection alone provides insufficient value in this new world. The platforms that enable autonomous action through AI will dominate the future landscape. The rest will fade into technological obscurity, remembered only as stepping stones.

Key takeaway: The future competitive advantage in your martech stack is going to come from AI that acts on your behalf, not just analyzes and recommends. When you implement systems where AI executes complex workflows based on reasoning, you empower your teams to achieve broader impact with fewer technic...