Go slow to go fast: Synopsys' data-first transformation approach

Business context: a forty-year-old company growing at AI speed

Synopsys turns forty in 2026, in a market with very few competitors. Its customers are the companies designing the world's most advanced chips and AI changed what those customers needed. Revenue reached $7.05 billion in fiscal 2025, up 15% year over year and roughly double where the company stood five years earlier, and the Ansys acquisition added six to seven thousand people, taking the workforce to approximately 28,000.

That leaves a company carrying both rapid growth and forty years of legacy. Bailie arrived to answer which skills it needs next, which ones it is divesting, and how AI is changing both the products Synopsys builds and the way its engineers build them. He had run the play once before at Cisco, without the technology to support it. One thing carried over: do not make it an HR project.

People problems are everyone's problems, but often we turn up with we're HR and we have the solution, whereas instead I think we turn up and go, let me understand your problem, and then there's some stuff that we can help with as an HR function.”
Ian Bailie, VP, People Operations and Strategy on the TechWolf Podcast

Challenges: the foundation was thinner than the ambition

Synopsys had a job architecture. It had recently been simplified, in a direction that did not serve the work Bailie needed to do.

• A job architecture built for benchmarking, not for work design. Around 3,000 people sit in the application engineering job family under effectively one job definition. It mapped cleanly to market compensation data and told Bailie nothing about who did what.
• No appetite for another platform. Synopsys runs a best of breed stack on purpose: ServiceNow as the front door, Degreed for learning, BetterWorks for performance, Avature for recruiting, Visier for dashboards. A separate skills portal would have been one more login employees had no reason to use.
• Every tool had its own idea of skills. Each platform shipped its own skill list and its own suggestion algorithm, so an employee moving between learning, performance and internal jobs met three versions of themselves.
• Job data decided everything downstream. The pilot showed the output was only ever as good as the job data going in, and that set the sequence for everything after it.

We have three thousand people who are all basically defined as saying they do the same job, but they don't. I don't know which product they support, what technical skill set they have, anything like that.”
Ian Bailie on the TechWolf Podcast

How we helped: one taxonomy, the platforms Synopsys already had

Bailie wanted a data layer rather than an application, and he wanted every other vendor in his stack to run on TechWolf's taxonomy rather than its own.

Trusted data. The pilot ran offline by design. No integrations, just flat files: employee data plus work data from Jira and other ticketing systems, scoped to application engineering. Synopsys spot-checked profiles with employees and managers rather than validating all 3,000, because the question was whether the technology worked. The checks came back accurate enough to move forward, and they pointed at job data as the signal underneath everything else. So the team went back to defining jobs and levels, and pushed for depth: forty to sixty skills with proficiency attached, rather than the ten-skill summaries common in the market.

Executive insights. Skills and task data flow into Visier. The integration was built before go-live, so dashboards populate as each cohort comes online rather than arriving as a report a month later. Synopsys is building out the views leaders ask for: where skills and proficiency gaps sit today, and which future-facing capabilities are missing.

Embedded actions. Skills show up where employees already have a reason to be. Degreed went first and is live for the first cohort, because a profile with proficiency turns straight into a personalised learning path. Avature is next, so a profile will also produce job recommendations in the talent marketplace. BetterWorks follows, carrying skills onto goals and career conversations. Bailie is explicit that skills are there for career growth, not for performance decisions.

Outcomes: what the foundation now makes possible

One profile, one taxonomy, wherever the employee is

Degreed, Avature and BetterWorks are standardising on the TechWolf taxonomy rather than their own. Degreed is live for the first cohort, Avature is due in the coming months, BetterWorks follows. For a company that chose best of breed over consolidation, that translatability is what makes the choice survivable.

Depth a technical workforce recognises

A ten-skill profile earns a shrug from an engineer who could have written it themselves. Forty to sixty skills, each with a proficiency level, describes what the job actually takes. It also makes the ladder visible: between an L1 and an L3 the skills are often the same ones, and only the level moves. A flat list cannot show that difference. A profile with proficiency can.

Where for me this is really powerful, particularly because we have such a technical workforce, is when we get more into forty, fifty, sixty skills. [...] And if I really want to make that job matching piece work, if I want people to understand how they go from an L1 to an L2 to an L3.”
Ian Bailie on the TechWolf Podcast

A rollout that moves cohort by cohort

The people team went live first, with engineering and sales behind it. Because the job data work and the first integrations were done up front, adding a group is now mostly preparation and sponsorship rather than platform work. A company-wide switch was never the plan: the parts of the business with the sharpest need come first.

AI impact as the way into the business

The task data changed the conversation. TechWolf scores each task for how much human involvement it calls for, using the Human Agency Scale developed at Stanford. That lets Synopsys see where work can be augmented, and where it stays human, at the level of the task rather than the job. The first aggregate view for the people team showed clear augmentation potential.

Bailie now opens with the AI impact question, and the skills conversation follows from the business rather than from HR. Inside the people team he is testing this as a loop: show a people partner the tasks with the most augmentation potential, ask whether they agree, then have one of them build the agent for the biggest task and scale it to peers. He is candid that it remains to be seen how it plays out.

Governance: enough trust in the data to act on it

Synopsys treats data quality as the governance question: a profile nobody believes will not be opened twice.

• Job data is validated before inference runs. Managers review and confirm job definitions up front, with proficiency worked into the same step. It is the step Bailie is least willing to compromise on, because job data underpins every profile downstream.
• Employees accept, add and correct in the flow. Inferred skills surface inside Degreed, where employees accept them, add what the system does not know, and get a learning path back for the effort. The first login is treated as the moment that decides adoption.
• Directional beats perfect. Bailie is explicit that the data will never be perfect. What he holds the output to is whether it leads to a decision.

Project challenges: the unglamorous work and the fair questions

• Job architecture is unglamorous and unavoidable. Going one level deeper set the pace for everything else, and it was not on TechWolf's roadmap either. Synopsys is where that gap became impossible to ignore, and it is now a product track: an agent that reviews an existing architecture against a company's own design principles and proposes the next level down.
• The business had a fair challenge. Many groups were already working on AI themselves, so why do they need HR for this? Bailie's answer is that they may not, and that data on tasks, skills and the external market is still worth having once the work has to be redesigned into new jobs.

Next steps: more cohorts, more platforms, sharper questions

The Avature integration comes online next. Once it is live, a skills profile also produces job recommendations in the talent marketplace, which gives employees another reason to keep theirs current. After that it is more groups through the same sequence, and more of the stack carrying skills data.

Synopsys is also testing market intelligence data and a conversational agent that queries internal skills, task and market data together. Bailie's closing point was aimed at his peers in the profession.

A lot of it is around job design and org design. This is HR's moment to really step in.”
Ian Bailie on the TechWolf Podcast