Senior Product Engineer 1

10-Bagger potential on the TSX: two small-cap stocks to watch now. For long-term wealth builders, turning an initial $20,000 into $200,000 means achieving a 10-fold return. While that may seem out of reach in the short term, over a 20-year horizon, it is mathematically entirely feasible at a compound annual growth rate of approximately 12.2%. On the Toronto Stock Exchange, two small- to mid-cap growth companies with clear business models and long-term expansion potential in their respective sectors - Docebo (TSX:DCBO) and WELL Health Technologies (TSX:WELL) - are worth investors' attention. Suggested opportunity. Docebo: An AI-Powered Enterprise Learning Platform Docebo provides a cloud-based training and learning platform for large organizations to train employees, customers, and partners across onboarding, compliance, and professional skills development. Once deeply embedded in daily operations, this business model generates recurring subscription revenue. Artificial intelligence is becoming a key variable in enhancing Docebo's platform value. In January 2026, the company acquired 365Talents, an AI-driven skills intelligence and workforce analytics firm, gaining technology to identify employee skills, uncover capability gaps, and proactively recommend training or internal mobility opportunities - tightly binding skills management with learning execution. Financially, Docebo reported first-quarter annual recurring revenue (ARR) of US$248.9 million, up 10.6% year-over-year, while free cash flow climbed to US$27.6 million, demonstrating its ability to improve profitability through operational efficiency even as it expands. Despite recent valuation adjustments in the software sector, Docebo's market capitalization of roughly US$650 million (approximately US$480 million) leaves considerable room for growth if its AI-powered platform continues to win large enterprise customers. WELL Health: A Hybrid of Physical Clinics and Digital Healthcare WELL Health is a unique player in Canada's digital health space, operating a dual business model that combines "physical" and "digital" - running a network of brick-and-mortar medical clinics while supplying digital tools that help healthcare providers manage patients, electronic records, billing, cybersecurity, and virtual care. This combination allows WELL Health to benefit simultaneously from direct healthcare delivery and back-end system efficiency gains. Against the structural challenges of Canada's physician shortages, aging population, and overburdened healthcare system, demand for WELL Health's services remains robust. In the first quarter of 2026, the company recorded 1.9 million patient visits, while revenue rose 25% year-over-year to C$368.3 million. More notably, its Canadian operations (including clinics and the WellStar software platform) achieved a C$100 million annualized adjusted EBITDA (earnings before interest, taxes, depreciation, and amortization) run rate three quarters ahead of schedule - and on lower revenue than originally projected - proving that operational efficiency is improving as the business scales. With WELL Health's current market capitalization at approximately C$1.1 billion, if management continues to deliver on growth and efficiency improvements over the next several years, the long-term return potential is worth watching. Risks and Investment Strategy Of course, these investments are not without risks. Docebo faces intense software competition, lengthy enterprise sales cycles, and integration uncertainties from acquisitions. WELL Health carries debt, relies in part on acquisitions for growth, and operates within highly regulated healthcare systems. Both stocks are highly volatile, and sharp price corrections can occur even when the long-term thesis remains intact. Investors may consider a gradual position-building approach, spreading capital across both companies while maintaining patience measured in years. Docebo offers recurring software revenue tied to AI and workforce development, while WELL Health combines digital healthcare with a rapidly expanding clinic network. A $10,000 allocation to each does not guarantee a 10-fold return, but the market capitalizations of both companies - and the vast size of the markets they address - make that long-term goal worthy of serious consideration.

Docebo's MCP Server signals the MCP era - 6 musts for secure, production-ready LMS integration. The LMS is no longer just a destination interface. In July 2026, it became a callable service layer inside enterprise AI assistants, and that shift has immediate consequences for L&D, IT, and security teams. Docebo's general availability release of its MCP Server on July 22, 2026 marks a practical turning point: learning operations can now happen in ChatGPT, Copilot, Claude, and Gemini instead of only inside the LMS UI. That sounds like a UX improvement, but the deeper change is architectural. When an LMS becomes a toolset inside assistant workflows, operating models, permission boundaries, and governance controls all need to evolve. Why this release matters now. What makes this moment important is not just feature velocity, but implementation timing and enterprise readiness pressure. From Docebo's release and community rollout details, teams are moving from beta patterns into production architecture now. This includes a required move away from temporary beta endpoints to permanent server URLs and a re-authentication cycle for users. In practical terms, this is not future planning - it is active cutover work. For enterprise learning leaders, this creates a narrow window to answer three business questions: * Where should learning workflows execute - in LMS screens, AI assistants, or both? * Which actions are safe to expose to natural-language tool invocation? * How will you audit and control assistant-mediated access at scale? From LMS destination to LMS tools: the new operating model. The strongest signal in Docebo's MCP direction is the move from passive retrieval to operational actions. Current capabilities position MCP for both learner and admin use cases, including: * Learner-side access to training information, progress, and certifications * Admin-side workflows for enrollments, course and learning plan management, and content curation * AI-assisted content curation actions such as discovering third-party content, comparing options, and importing into folders This changes day-to-day L&D work in a few meaningful ways: * Fewer context switches for admins and managers * Faster completion of repetitive operations through natural language * More pressure to design role-specific tool exposure, because assistants can now execute, not just answer The key takeaway: MCP is not just another integration. It turns the LMS into a governed capability layer that can be invoked from multiple assistant surfaces. What implementation teams must get right first. The configuration flow described in Docebo's developer documentation is straightforward, but it has important constraints that affect rollout quality. Core implementation realities. * Setup is a shared responsibility between Docebo Superadmins and assistant-platform admins * OAuth app setup and MCP server setup are distinct steps * Users complete one-time authorization, but SSO-related flow limitations still require careful sequencing * Custom server architecture enables persona-based exposure of tools * The built-in learner server is fixed, while custom servers can be created, enabled, disabled, edited, or deleted Immediate rollout checklist. * Create separate OAuth apps per assistant client when needed * Standardize MCP server naming and path conventions by persona/use case * Validate production endpoint changes before user enablement * Segment assistant connectors by audience (learner vs admin) to reduce permission confusion * Run pilot tests with realistic enrollment and content-volume scenarios A notable lesson from community feedback is that error messaging clarity matters. Permission failures can present as ambiguous resource errors in assistant conversations, which creates avoidable support load if role boundaries are not clearly designed. Security and governance: what admins should demand next. MCP adoption in enterprise learning is viable only if security controls mature alongside usability. Two external signals are especially relevant: * Microsoft's enterprise MCP model emphasizes read-focused tool boundaries, delegated permissions, rate limits, and activity logging through existing governance surfaces. * The MCP enterprise-managed authorization extension formalizes centralized IdP control to reduce per-user token sprawl and improve revocation governance. * NSA security guidance highlights broader MCP risks - including weak access control patterns, token/session handling gaps, tool misuse paths, and insufficient auditability. For enterprise teams, this translates into concrete control requirements: * Centralized authorization policy via enterprise IdP * Least-privilege tool exposure by persona and environment * Strong token lifecycle controls (rotation, revocation, scope discipline) * Comprehensive audit logging tied to user identity and tool invocation context * Execution guardrails for high-impact actions and abnormal request patterns * Clear change control for tool catalogs and connector updates In short, if LMS workflows are becoming assistant-executable, then governance must shift from UI permissions alone to assistant-era policy enforcement. What comes next in the MCP era of enterprise learning. Docebo's July 2026 release shows where the market is heading: enterprise learning platforms are becoming part of a broader AI tool fabric. The winners will not be organizations that simply connect an LMS to an assistant. They will be the ones that operationalize this model with disciplined architecture, role design, reliability testing, and security-by-default controls. The strategic opportunity is clear: make learning workflows faster and more embedded in work without sacrificing control. The teams that treat MCP as both a productivity layer and a governance challenge will set the new standard for enterprise L&D execution.

Docebo's fair value estimate remains at CA$35.97, unchanged despite recent model adjustments. Analysts note that updated assumptions are reshaping the risk and valuation narrative around the learning platform company, with differing interpretations emerging across Street research. The company raised its full-year 2026 revenue guidance to US$271 million to US$275 million, up from previous guidance of US$267.5 million to US$269.5 million. First-quarter 2026 revenue is expected between US$65.4 million and US$65.6 million, compared with US$57.3 million in the same period last year. Bullish analysts highlight the company's clear story, whilst bearish voices caution that narrow research coverage leaves gaps in understanding execution and growth prospects. Docebo recently unveiled a next-generation learning platform at its Inspire 2026 event.