If AI can speed up recruitment, what will distinguish one recruiter from another when everyone has access to similar capabilities? As organisations look towards 2027, Agentic AI could take on larger parts of the hiring workflow, from initial candidate engagement to scheduling and screening.
In a conversation with People Matters, Sameer Penakalapati, Founder and Chief Executive Officer (CEO) of Ceipal, argued that the opportunity lies in what recruiters can do with the time technology gives back to them. As agents take on routine execution, recruiters could focus more on candidate conversations, consequential decisions and relationship-building.
Agents could clear the administrative backlog by 2027
Penakalapati described a recruiter’s day as one often dominated by administration: chasing people, rescheduling meetings, updating records and screening candidates. These tasks leave limited time to understand a candidate’s motivations and expectations.
By 2027, he expects agents to have contacted candidates, coordinated schedules and collected basic information, such as location and preferences for hybrid or on-site work, before the recruiter starts the day. They could also ask preliminary qualifying questions.
The recruiter’s day, he said, could shift from managing “40 small tasks” to handling “a handful of real decision-making, real conversations with the candidates.”
That does not mean agents would run the entire hiring lifecycle independently. Penakalapati described a model in which workflows are predefined in systems, with human involvement at points requiring judgement, negotiation or compliance. As he put it, the work left to people would become “really human, but actually more human, because of this.”
He also pointed to legal requirements in the US and Europe that, in his view, necessitate human involvement in hiring decisions.
Candidate trust remains a human responsibility
Agents could handle scheduling, basic data collection and initial qualifying questions. Building a candidate’s confidence in an opportunity is more complicated.
Penakalapati noted that candidates may want to understand an organisation’s culture, working environment, project ownership, autonomy and learning opportunities. These conversations help candidates assess whether a role is right for them.
“The trust cannot be built by the bots,” he said.
With less time spent on scheduling, resume submissions and paperwork, recruiters could better understand candidates’ needs, address their concerns and explain the opportunity. Penakalapati pointed to a familiar frustration among candidates: they share their information and then find that “nobody calls back and tells us an update.”
More time for recruiter interaction, he argued, could help close this communication gap and improve the candidate experience.
Connected data matters more than a smarter model
For agents to deliver on this promise, the systems in which they operate need to work together. For Penakalapati, the priority is not simply deploying a more capable AI model. “It’s really not about a smarter model; it’s whether your data talks to itself.”
Candidate records, conversation notes, timesheets and other relevant information need to be connected across the recruitment workflow. Deploying multiple agents in silos could create redundant work and technical complexity. Penakalapati favoured a smaller number of interconnected agents aligned with the workflows they are intended to support.
He identified a well-functioning applicant tracking system (ATS), clean data and established processes as important foundations. Organisations need to understand how information is collected and stored, define workflows and determine where human intervention is necessary before assessing the return on investment. Siloed agents from different providers could introduce additional work and, he warned, risks such as data leaks.
For talent acquisition leaders, this preparation also involves data governance and compliance. Penakalapati estimated that working with technology providers to structure data and address these requirements could take three to six months, depending on the state of existing systems.
His advice was to understand the data and define the workflows before subscribing to multiple tools. AI’s ability to reason does not automatically translate into effective action or outcomes; the surrounding systems and processes matter just as much.
If everyone gets faster, how should recruiters be measured?
As AI capabilities become more widely available, Penakalapati argued, speed alone may become less of a differentiator. Comparing the future to a scenario in which everyone is travelling on the same bullet train, he said, “Everybody is the fastest in the room; trust becomes the key.”
This could prompt talent acquisition leaders to reconsider how they measure recruitment performance. Existing key performance indicators (KPIs), such as time to submit and submission-to-hire ratio, may need to be complemented by measures that reflect the division of work between agents and recruiters, as well as the number of recruiter interactions a candidate has before being hired.
Penakalapati observed that recruitment KPIs had remained largely unchanged for 20 years. As workflows evolve, performance measures may also need to reflect the contribution of human engagement to hiring outcomes.
Rethinking recruitment for an AI-driven future
Penakalapati’s vision of recruitment in 2027 is one in which agents handle more routine work while recruiters spend more time on conversations, judgement and decisions that require human involvement.
For talent acquisition leaders, the challenge is to redesign recruitment around that distinction: identify what can be automated, establish the data and workflows needed to support it, and ensure that the time released by technology creates more room for meaningful human engagement.
