Podcast Summary: Purpose-Built Enterprise AI Agents: What Actually Works

Podcast: Everyday AI Podcast – An AI and ChatGPT Podcast
Host: Jordan Wilson
Guest: Prashanti Padman, VP of Engineering at LinkedIn
Date: January 8, 2026

Episode Overview

This episode centers on the real-world journey of building, deploying, and iterating purpose-built AI agents for enterprise use—specifically LinkedIn’s Hiring Assistant. Host Jordan Wilson and guest Prashanti Padman dive deep into the nuanced landscape of agentic AI, cutting through the hype to highlight what actually works in enterprise settings and how AI agents must be tailored to deliver measurable value. Other key themes include the ongoing necessity of human oversight, the importance of user experience, building trust in AI systems, and providing actionable advice for leaders implementing AI solutions.

Key Discussion Points & Insights

1. Defining the Reality of AI Agents in 2026

• The Hype vs. Reality (00:17)
  • While 2025 was dubbed “the year of the AI agents,” in practice there is no universal agent that can immediately take over all enterprise tasks. Effective agents today are narrowly focused, purpose-built, and designed to accomplish very specific business goals.
  • Host perspective: "When people think of AI agents, I think they think of that one general agent that can go out and do anything...and that's not necessarily the way that agents, at least today in early 2026, are built." — Jordan Wilson (00:39)

2. LinkedIn’s Approach: The Hiring Assistant Case Study

• Product and Context (03:29)

  • Prashanti leads engineering for LinkedIn’s Talent Solutions, focusing on products supporting hiring, career development, and learning.
  • The drive to create an agentic product stemmed from both the AI revolution and the recognition that key recruiter workflows could be streamlined, freeing up recruiters’ time for more human-centric, value-added work.

• Recruiter Workflow Automation (04:00)

  • The agent’s key role is to automate repetitive processes: structuring job requirements, sourcing candidates from vast datasets, and shortlisting potential hires.
  • This allows recruiters to focus on relationship-building and strategic aspects of hiring.

• The Human Plus Approach (08:35)

  • LinkedIn’s philosophy is "human plus": not replacing humans, but empowering them to become "10x recruiters" (similar to the “10x engineer” concept).
  • The product was developed in direct, iterative partnership with customers, recognizing that optimal workflows and necessary features could only be discovered through real-world use and rapid feedback cycles.
  • Notable quote: "You cannot do that by just sitting in a boardroom and writing specs. You just cannot." — Prashanti Padman (09:44)

3. Scoping and Building Purpose-Built AI Agents

• Distilling Use Cases (13:57)

  • Identify which job elements are best suited for AI assistance (e.g., data mining, pattern matching, bulk analysis) vs. those requiring the human touch.
  • Machines excel at pattern recognition and parsing massive datasets—finding the “hidden gems”—but humans remain essential for final decisions and relationship-building.

• Building Trust and Transparency (15:45, 17:43)

  • Trust is crucial: Users must understand how agents make decisions—no “black boxes.”
  • LinkedIn’s Hiring Assistant now “shows its math,” offering evidence about why particular candidates are selected to build user confidence.
  • Notable quote: "Customers have to build trust on agents. That trust doesn't get built overnight. It takes time to build...they wanted to understand how the agent thinks and reasons." — Prashanti Padman (15:45)

• Iterative Evolution (11:51)

  • The product matured significantly through customer-guided iteration (e.g., shifting from asynchronous to conversational interfaces).

4. Human-in-the-Loop: Guardrails and Collaboration

• Always Human-in-the-Loop (19:54)

  • For LinkedIn, agents are intentionally positioned as “assistants,” not autonomous replacements. The human always reviews, approves, and can give feedback to train/improve the agent.
  • Notable quote: "At the end of the day...you as a recruiter, as a human are the ones looking at it and saying okay, do I agree, do I as a recruiter agree with this agent's assumptions?...It's a bi-directional relationship." — Prashanti Padman (20:05, 22:46)

• Purpose and Change Management (24:01)

  • Some professionals may find purpose in tasks that could be automated, so change must be handled thoughtfully.
  • Adoption is eased by integrating agents into familiar workflows rather than replacing them, offering efficiency gains while preserving user autonomy.
  • Metrics:
    • 62% reduction in profiles viewed
    • 70% increase in InMail acceptance rates
    • Agents free users to focus on more meaningful, strategic tasks, but users can still revert to manual review at any point.

5. Enterprise Lessons and Actionable Advice

• Enterprise Complexity and Context Engineering (28:24)
  • Enterprises have fragmented workflows and multiple systems (ATS, CRM, HR, etc.), so building a successful agent requires deep integration (“context engineering”), not just overlaying an agent on existing processes.
  • Trust and confidence are built not from model complexity, but from a thoughtful user experience that gives customers control.
  • Notable quote: "Your experience is how you’re going to build trust with your customers...Can your experience instill trust? Can your experience instill flexibility? Can your experience instill confidence in your customers that this change in paradigm is not going to turn their world upside down?" — Prashanti Padman (29:40)

Notable Quotes & Memorable Moments

• On Human-AI Teaming:

  "If you put a human and an agent together...how can you 10x that? We talk about 10x engineers who are using coding agents. How can you create 10x recruiters if you give them an amazing recruiting agent for them to work with?"
  — Prashanti Padman (08:40)

• On Evolution Through Customer Feedback:

  "The version of the hiring assistant that you’re seeing today in the market is not where we started...You need a much more conversational interface...where the hiring assistant and the recruiter can tag team on that job, can work together, can bounce off ideas."
  — Prashanti Padman (10:44)

• On Trust and Transparency:

  "We evolved experience so that we actually show the process. The agent will show you what it’s doing, it will tell you what it’s looking at, how many resumes it’s looking at, what it’s finding in the resumes...showing that evidence was very important in the experience for the customers to build trust around it."
  — Prashanti Padman (15:50)

• On Gradual Change and Adoption:

  "We didn’t just build the Hiring Assistant as a standalone product. We brought it as a capability on top of the recruiter product...we are not suddenly pulling the rug under the customer...we are slowly easing you into change."
  — Prashanti Padman (24:26)

• On What Makes a Winning Enterprise Agent:

  "Enterprise is a messy business...having an evolving, purpose-built, domain-specific model is going to be very, very important...the experience part is so important because it’s the experience, the application logic...that you’re building on top of these LLMs is super important to get right."
  — Prashanti Padman (28:24)

Timestamps for Key Segments

| Timestamp | Segment/Event | |-----------|------------------------------------------------------------------| | 00:17 | Framing the state of AI agents in 2026; limits of “general” AI | | 03:29 | Prashanti’s background and LinkedIn’s AI agent journey | | 07:24 | Motivations for automating recruiter workflows | | 08:35 | The human-plus-agent philosophy | | 11:51 | Early product iterations, shift to conversational UI | | 13:57 | Scoping and separating tasks for agents vs. humans | | 15:45 | Building trust through transparency and “showing your work” | | 17:43 | The necessity of trust in agentic AI for high-stakes outcomes | | 19:54 | Human-in-the-loop and why full autonomy isn’t the goal | | 22:46 | On agent-user co-working and preserving user purpose | | 24:01 | Metrics from deployment and adoption strategy | | 28:24 | Actionable advice for AI adoption in enterprise |

Actionable Takeaways

• Define Narrow, Measurable Agent Use Cases: Avoid trying to automate everything at once. Focus on repetitive tasks that agents do better than humans (pattern matching, data mining, etc.).
• Iterate With Real Users: Involve real customers from day one. Let user feedback direct product changes, especially regarding workflows and trust-building mechanisms.
• Prioritize Trust and Transparency: Always show your work—users need to understand the agent’s process, not just see results.
• Keep Humans in the Loop: Allow users to review, override, and refine agent decisions; this not only builds trust but leverages human judgement where it matters most.
• Seamlessly Integrate With Existing Systems: Don’t force radical changes—layer new agentic capabilities onto existing workflows and products.
• Remember Enterprise Complexity: Anticipate fragmented systems and context. Build robust data and integration pipelines for meaningful AI augmentation.
• Make User Experience Central: Superior UI/UX is just as critical as model performance—it’s how you win users’ hearts and trust.

This episode is a must-listen for anyone navigating the adoption of AI agents at scale, offering both philosophical framing and tactical insight into building effective, trusted tools for real-world enterprise challenges.