Peak Demand + Retell Partner to Build Healthcare and Transit Voice Agents

About Peak Demand

Peak Demand began as an organization that offered support for website development and basic business automations. As AI capabilities evolved, founder Alex Masters Lecky saw a larger opportunity: organizations increasingly wanted AI to communicate with customers, yet faced a real challenge when it came to ensuring that those agents reliably interacted with the systems businesses already depended on.

Today, Peak Demand designs, integrates and operates production voice AI systems on Retell for organizations with complex workflows, including healthcare providers, manufacturers, and public-sector transit organizations.

Retell provides the best-in-class conversational voice AI infrastructure at the center of these deployments, giving Peak Demand the real-time voice performance, reliability, function calling and development environment required to build production-grade agents.

Peak Demand builds the surrounding integration and orchestration layer that connects Retell to business systems, encodes deterministic rules, handles identity and validation, manages workflow logic, and continuously optimizes each deployment around the client’s real-world operations.

Partner profile

• Partner: Peak Demand
• Location: Toronto, Canada
• Focus: Voice AI implementation and ongoing operations
• Industries: Healthcare, manufacturing, municipal and transit services
• Notable integration: TELUS Collaborative Health Record through custom middleware

Results reported by Peak Demand from its deployment for Creekside Health using Retell and a custom TELUS Collaborative Health Record integration.

Canadian Tech Skepticism

Lecky describes the Canadian market as cautious about adopting AI, particularly in healthcare and the public sector, where privacy, security and operational reliability matter.

Prospective clients may initially think of voice AI as a more sophisticated automated phone menu. Peak Demand’s challenge is demonstrating that an agent can do significantly more: understand a caller, apply the organization’s rules, interact with existing systems and reliably complete useful work.

“It almost goes without saying that Canadians are very skeptical about AI.”

That skepticism has shaped Peak Demand’s implementation philosophy. Instead of asking clients to redesign their operations around AI, the agency builds systems around the workflows organizations already use.

Creekside Health Needed a TELUS Solution

For Peak Demand customer Creekside Health, the need for a voice AI solution was immediate. The clinic struggled to find a capable receptionist who could manage patient conversations while working competently in its electronic medical record system. Doctors sometimes even had to answer the calls themselves. However, answering the phone was only part of the challenge.

A useful agent needed to identify patients appropriately, understand what they were trying to accomplish, follow clinic-specific booking rules, retrieve availability and complete scheduling transactions inside TELUS Collaborative Health Record — all while maintaining a natural conversation with patients who could be anxious, unwell or unfamiliar with the technology.

Using Retell to Connect Patient Conversations to the TELUS CHR

Peak Demand built a custom integration and deterministic orchestration layer between Retell and TELUS CHR. Retell provides the conversational voice AI infrastructure, handling the real-time patient conversation, speech, reasoning and function calling. Peak Demand’s surrounding middleware controls how those function calls interact with the clinical system. It handles areas where the deployment cannot afford improvisation, including identity resolution, booking rules, validation, availability queries, appointment transactions and clinic-specific workflow logic.

Per Lecky, that separation is a deliberate choice.

“Retell gives the agent the ability to understand and communicate naturally with the patient, while Peak Demand’s deterministic software governs the operational actions taking place behind the conversation.”

Getting there, however, required substantial iteration. Lecky recalls reaching roughly the 200th version of the Creekside agent as Peak Demand worked through the edge cases that emerge when conversational AI is connected to a real clinical environment.

Following the launch, Peak Demand continues monitoring and optimizing production agents as new caller behavior and edge cases emerge. Retell’s workspaces, call records and function-call logs give Peak Demand the visibility needed to diagnose unexpected interactions quickly and continuously improve each deployment.

According to Lecky, many production issues can be diagnosed and corrected within minutes or hours rather than requiring a major redevelopment cycle.

“We go through hundreds of versions of these agents to land on something that really works.”

Production Results Change the AI Conversation

Peak Demand reports 74.9% appointment-workflow containment and 94.9% booking-path success at Creekside, with 196 appointments booked during the measured deployment period. Those results come from a broader post-integration analysis covering more than 1,100 connected calls.

For Peak Demand, results like these have changed the nature of conversations with organizations that were initially hesitant about voice AI. Instead of discussing whether the technology can work at all, conversations increasingly focus on which additional workflows it should handle.

The Creekside deployment also reduces the need for clinical staff and physicians to spend time handling routine telephone activity. Peak Demand has applied the same principle elsewhere. For a home healthcare provider, an after-hours Retell agent answers overnight calls, gathers required information, and routes issues to the appropriate human team for follow-up. And for a global manufacturing client, a Retell-based receptionist filters high volumes of unwanted calls while identifying and transferring legitimate sales, parts, technical support and customer inquiries.

“We’re augmenting the operations of high performers by taking away some of the more mundane things, allowing them to do more productive work.”

Retell Has a Knowledge Base Powerful Enough for Canadian Public Transit

Peak Demand originally attempted to build its public transit schedule agent on Synthflow. The challenge there was not simply the amount of information involved, but the structure of that information.

A functional system needed to work across thousands of lines of transit schedule information covering routes, stops, service directions, departure times, schedule variations and the relationships between them. A rider might ask for the next bus without knowing the exact route number, describe a destination rather than a stop, change direction midway through the conversation, or ask a follow-up question that depended on information retrieved earlier in the call. As the dataset and retrieval requirements grew, Peak Demand found that Synthflow could not reliably accommodate the volume and structure of the transit information while maintaining the behavior the project required.

Moving the project to Retell changed what was possible. Retell’s knowledge-base capabilities allowed Peak Demand to organize the large body of transit information and precisely instruct the agent how to search, interpret and use it during a conversation. The Retell agent could then retrieve the appropriate schedule information and communicate it naturally to the caller without Peak Demand having to build a dedicated external API request for every possible schedule query.

Reliability reinforced that decision. During a major global technology outage the previous year — which disrupted numerous platforms and infrastructure providers, including Cloudflare — Peak Demand observed that its Retell agents continued handling calls. Combined with Retell’s performance across increasingly complex deployments, that experience gave Lecky the confidence to consolidate Peak Demand’s voice AI work around the platform.

The Retell Partner Program Is a Force Multiplier

Peak Demand’s healthcare and transit deployments brought the agency to Retell’s attention and ultimately led to an invitation to the Partner Program. For Peak Demand, the relationship works because the two companies solve complementary parts of the production voice AI problem. Retell provides the critical conversational voice AI infrastructure and development environment. This allows Peak Demand to focus on understanding the client’s operations, connecting Retell to existing systems, encoding business rules, testing edge cases, and remaining responsible for the system after it enters production.

“Asking a business owner to take on all of the voice AI operations themselves is often out of the question.”

Lecky works with five to ten other team members and estimates that 70–80% of the agency’s automation and workflow development now happens in Retell. Prior to the Partner Program, the team spent relatively little time participating in the broader technical community because most of its attention was devoted to client work. Through the program, Lecky has increasingly connected with other Retell builders, shared implementation knowledge, and answered questions about complex deployments. Retell, in turn, has helped bring work originally developed for individual organizations to a much broader audience.

Lecky also sees enterprise referrals through the Partner Program as a natural extension of Peak Demand’s model. Many organizations need not just an AI platform but a skilled implementation partner capable of turning that platform into an operational system and continuing to improve it after launch.

What’s Next for Peak Demand?

Peak Demand is currently preparing a group of public transit agents for broader deployment. Alongside schedule retrieval, the project includes an agent designed to handle complex public service requests conversationally.

That system has completed extensive regression testing across multiple request pathways. A custom MCP-based integration layer connects the Retell agent with the transit authority’s systems so that the agent can collect the required information, validate it and submit requests on a rider’s behalf.

The architecture moves Canadian public transit considerably beyond a traditional IVR. Some technical work remains to be completed on the client side, and Lecky says two or three other transit systems are waiting for the project to launch before moving forward with similar deployments. Peak Demand is also in discussions with additional municipal organizations and healthcare providers.

Over time, Lecky expects voice AI to move beyond the telephone itself. Peak Demand is exploring applications where spoken instructions can initiate actions across business software, allowing users to operate complex systems conversationally rather than navigating traditional interfaces. That could make sophisticated technology significantly more accessible to people who are uncomfortable with software or simply prefer speaking to typing and clicking.

“Especially for people who are not tech-y, they could just talk to their technology through a system like Retell.”

Retell’s flexible, always-improving platform enables voice AI to move beyond simply answering a phone call and toward something more significant: conversational systems that can reliably participate in all the day-to-day operation of an organization.

Already using Retell with clients and want to participate? Learn how Retell’s partner program works.