A software developer prepares a pull request and pauses for a moment. Much of the code in the update came from an AI assistant.
The system has already flagged a potential issue and suggested a fix before anyone else on the team reviews the change.
Anyone who has worked around software teams knows pull requests can trigger long review threads and days of back-and-forth. Now an AI assistant often joins the discussion, generating code and flagging problems before another developer even opens the thread.
According to new global research from Snowflake and Omdia, nearly half of the code written inside organizations today is generated with AI assistance.
Technical teams are already reorganizing around that reality, often discovering the same tools creating new roles are eliminating others. The research shows engineering groups both hiring and eliminating roles as AI tools become embedded in everyday workflows.
In Canada, organizations are prioritizing these shifts at the front door, moving faster than global peers to target customer-facing experiences.
For many organizations, the technology shaping how software gets built is now beginning to form how customers interact with the business as well.
Engineering teams are reorganizing around AI systems
The report surveyed more than 2,000 business and technology leaders across 10 countries. Among those organizations, 77% say AI adoption has created jobs somewhere in their workforce, while 46% say roles have been eliminated.
Many companies report both outcomes at once.
This paradox is most visible in technical departments. IT operations, cybersecurity, and software development report some of the largest job gains tied to AI adoption (at 56%, 46%, and 38% respectively). Those same functions also report some of the biggest reductions.
In practice, AI is taking over certain tasks while creating new work around building and operating AI systems.
“Engineering teams are increasingly focused on making AI work in production by deploying AI agents at scale and ensuring they operate reliably and safely in real-world environments,” says Qaiser Habib, head of Canada engineering at Snowflake.
The pattern is already visible in large technology companies.
Salesforce, for example, has publicly discussed “rebalancing” its workforce as it invests in artificial intelligence. CEO Marc Benioff has described cutting some traditional roles while hiring aggressively for AI-focused engineers and teams building the company’s Agentforce platform.
The change mirrors what many organizations in the Snowflake research describe, which is fewer roles tied to routine development tasks, and more demand for engineers who can build, monitor, and scale AI systems.
That focus is changing how teams are structured. Habib says organizations are introducing new responsibilities around system design, evaluation, and control. Companies are also building expertise around the infrastructure that supports AI agents, including integrations, security, and performance.
Anyone who watched the cloud boom of the 2010s will recognize the pattern.
When cloud computing became widespread, companies created new teams responsible for infrastructure, automation, and security. AI is prompting a similar kind of reorganization within engineering groups.
Software development itself is changing
Ask a developer what slows down a release and the answer usually involves code reviews, testing, and a few bugs that appear at the worst possible moment. The modern developer’s workflow is becoming an exercise in “AI orchestration.”
Developers are using AI systems to help generate code, review pull requests, flag potential issues, and monitor performance once software is running. The research suggests the shift is already well underway, with 48% of code now generated with AI assistance, the human role is pivoting toward oversight.
“AI is becoming embedded across the entire software development lifecycle,” says Habib.
Currently, the technology’s strongest foothold is in analytics (71%), code reviews (66%), and generation (65%). Many organizations say the tools are speeding up the work developers already do.
Eighty per cent report faster development velocity, while 76% say AI-assisted development has reduced costs.
The technology is also changing how teams manage quality.
More than eight in 10 organizations report improvements in testing and bug detection when AI coding tools are used, and 80% also say those systems help improve overall code quality.
That changes where developers spend their time. Instead of writing every line manually, many teams now rely on AI systems to generate or suggest code while developers focus on architecture decisions, business logic, and oversight.
“Developer roles are evolving from primarily writing code to defining architecture, business logic, and guardrails for AI-assisted systems,” says Habib.
Testing is moving earlier in the development process as well. AI tools can scan code continuously, flagging issues while software is still being written instead of after a release candidate is built.
AI becomes another coworker, reviewing code, surfacing problems, and helping teams move faster from idea to production.
As engineering teams stabilize these internal tools, the focus is shifting from the back-end to the front office.
Canadian companies are moving faster on customer-facing AI
Once companies begin using AI across engineering teams, the next question is where else could this work?
Attention typically shifts to the customers.
The research suggests Canadian companies are moving quickly in this direction, with 45% of Canadian respondents saying their organizations are prioritizing generative AI tools in front of customers. Globally, that figure sits at 36%.
AI systems can answer routine questions, surface account details, or guide customers through common tasks while human agents step in for more complex issues.
“These areas offer clear, immediate returns,” says Shannon Katschilo, country manager for Canada at Snowflake. “Particularly through service automation and improved customer support, making them a practical entry point for many organizations adopting AI.”
But some companies are already pushing further.
The research shows that 31% of Canadian respondents say their organizations already use agentic AI in production, while another 32% say they’re interested but “early in the adoption process.”
Those systems can complete more complex tasks. Instead of answering a single question, they can analyze data, generate recommendations, and trigger follow-up actions across different systems.
Scaling AI still depends on the foundations
As AI tools move deeper into everyday work, many organizations are discovering that deploying them is one thing. Making them reliable across an entire company is the hard part.
The report found that 96% of organizations face obstacles when trying to expand AI initiatives. In other words, basically everyone.
Most of those challenges have little to do with the algorithms themselves. The difficulty sits within the data companies already own.
Systems hold information in different formats. Quality varies from one dataset to another. Much of the data organizations collect has never been prepared in a way that AI systems can easily use.
Governance quickly enters the conversation as well. When AI systems generate insights, respond to customers, or trigger automated actions, companies need to know exactly what information those systems can access and how decisions are being made.
“Organizations that successfully scale AI combine ambition with strong data foundations,” says Katschilo. “The Canadian companies seeing the most value, especially with autonomous agents, are those investing in high-quality data and the workforce skills needed to turn AI capabilities into measurable business impact.”
Even with the rapid pace of AI development, those fundamentals still determine which organizations see real results.
A developer opens that pull request. An AI assistant suggests a block of code while another system checks the logic and flags a potential bug before anyone else on the team sees it.
These are now routine situations.
The organizations turning those moments into real value are the ones that prepared the groundwork long before the AI arrived.
Final shots
- AI is already reshaping engineering teams. Across the organizations surveyed, 77% report job creation tied to AI while 46% report role reductions, often inside the same technical groups.
- Canadian companies are pushing AI into customer experiences quickly. Forty-five per cent are prioritizing the use of AI in customer-facing tools, higher than the 35% global average.
- Scaling AI still comes down to fundamentals. Ninety-six per cent of organizations report obstacles when expanding AI initiatives, most tied to data quality, governance, and skills.
