The Enterprise Is Quietly Onboarding a Second Workforce, and It Is Not Human
Every large company runs on software that remembers. For twenty years, customer relationship management (CRM) systems have logged every customer, every call, every deal, building a permanent memory of the business. What they've never done is act on that memory by themselves. A person still had to read the record and decide what to do next.
That's starting to change. Across staffing, financial services and government, a new layer of digital workers is appearing inside enterprise systems, quietly taking on tasks that used to sit in someone's inbox. These systems don't just store information. They read it, reason about it, and in a growing number of cases, act on it, escalating to a human only when real judgment is required.
Among those building this layer is Amit Kumar Tiwari, Vice President of Product Engineering at Asymbl, a company developing a workforce orchestration platform for the staffing and recruiting industry. "Software has started to do the work rather than just record it," he says, describing the shift plainly rather than as a sales pitch. "I've been building on the Salesforce platform for nineteen years, and I haven't seen a change of this size before."
The most visible version of this trend so far is in complaint and case management, where digital workers combine data, CRM and artificial intelligence (AI) to resolve issues end to end rather than routing them to a human at every step. One worker classifies the complaint, another retrieves the customer's history, another works out the likely cause, and another proposes or executes the fix. "What used to be a queue of tickets becomes a handoff between digital workers," is how Tiwari describes it, adding that the harder achievement isn't any single clever step but a system that "runs reliably inside a real enterprise platform, on a Tuesday, without breaking anything."
That caution is common among people who've watched earlier waves of enterprise technology stall. In Tiwari's experience, most AI pilots still don't survive contact with a real organization, and the reasons rarely have much to do with the underlying model. "The hard part is everything around it: permissions, audit trails, compliance, and ten years of process nobody wrote down," he says, a lesson he traces back to leading a large-scale rollout of Salesforce's Financial Services Cloud in wealth management, where he also set up a centre of excellence to govern the platform after launch, and where architecture and governance decided outcomes far more than the technology itself.
Security sits close behind reliability as the industry's biggest open question. Digital workers increasingly touch sensitive customer data, and the tools built to protect that data are becoming as important as the workers themselves. Tiwari's own answer to that problem is iDataMasker, a data-masking tool he built that is listed on the Salesforce AppExchange and lets development teams test against realistic data without exposing real customer records. Data masking, audit trails and security built directly into development pipelines rather than checked afterward are becoming standard requirements rather than nice-to-haves across the industry. "A company that can't prove its data discipline hasn't earned the right to automate," Tiwari says.
What the shift doesn't appear to mean is fewer people. In recruiting, digital workers can source, screen and schedule candidates faster than any human could. What they still can't do is build the kind of trust that makes a strong candidate say yes to an offer. The more common pattern emerging across the industry is human effort moving toward judgment, empathy and relationship-building, while software absorbs the repetitive middle of the work.
Whether this second workforce becomes as ordinary as the CRM systems it's built on will likely be decided less by how advanced the models get and more by questions of governance, trust and reliability that have very little to do with artificial intelligence at all.
