# Your AI Isn't the Advantage. Your Judgment Is

Sid Pednekar | AI Engineer NYC 2026 | 18:13

Source: https://www.youtube.com/watch?v=C2-v6wkHXhY
Channel: AI Engineer (https://www.youtube.com/@aiDotEngineer). Summarised by AIE Talks.
Page: https://aietalks.com/talks/your-ai-isnt-the-advantage-your-judgment-is
Published: 2026-11-06
Tags: product-strategy, workflows

## TL;DR
- Companies get better AI results when they choose a small number of high-value problems instead of adding tools to every existing workflow.
- AI leaders redesign work around judgment, with people and systems gathering evidence, investigating exceptions, and focusing attention where it matters.
- An advantage compounds when shared capabilities preserve evidence from successes and failures, so the organization learns instead of leaving knowledge with one person.

## Summary
Sid Pednekar argues that access to AI is now widespread, while business results remain uneven. He says the scarce resource is judgment, which means choosing what matters and deciding how to act when there is no obvious answer. His framework has three parts. First, companies should define the right problem, invest deeply in a few valuable workflows, and redesign how work gets done. Second, they should build shared capabilities by bringing technology and business people together, standardizing common infrastructure, and adapting the edges to different teams. Third, they should make learning compound by feeding successes, failures, corrections, and supporting evidence back into the organization. Two hypothetical asset management firms illustrate the difference. One adds summarization and other tools to existing work. The other redesigns research around investment theses, evidence, exceptions, and focused investigation. Pednekar's argument is practical: AI creates more answers, but judgment determines which work deserves attention and how the organization changes around it.

## Key ideas
### Judgment is scarce when intelligence is abundant
[03:32](https://www.youtube.com/watch?v=C2-v6wkHXhY&t=212s)
Pednekar says AI is making software, analytics, recommendations, and reports abundant. More answers do not automatically create more value. The scarce resource is judgment, which he defines as choosing what matters and how to act when there is no single obvious answer. It includes creativity about which possibilities are worth exploring, taste about what good looks like in a particular context, and decisions about which trade-offs are worth making. Judgment appears throughout an organization, from the C-suite and software engineers to AI leaders, portfolio managers, and data scientists. The practical question is where the organization's limited judgment is worth spending.

### The right AI problem starts with the bottleneck
[05:28](https://www.youtube.com/watch?v=C2-v6wkHXhY&t=328s)
Pednekar contrasts two hypothetical asset management firms. Company B starts with a co-pilot and asks where AI can be added to an analyst's existing work. It summarizes earnings transcripts faster, but the summary can remove context. Company A asks why material signals are arriving too late and where attention, AI capacity, and bandwidth should be focused. That question leads to a redesign or reevaluation of the workflow. The same approach can apply to research capacity, decision speed, data quality, latency, or reliability. Pednekar says leading organizations spend their attention on fewer, higher-quality decisions instead of spreading it across every possible use case.

### Deep investment beats scattered pilots
[08:16](https://www.youtube.com/watch?v=C2-v6wkHXhY&t=496s)
Once a company identifies the right problem, Pednekar says it should concentrate investment in the highest-value workflows. Experimentation can help, but he warns that it can turn into innovation theater or productivity theater without actual value. Company B launches separate tools for document summarization, data retrieval, and co-piloting. Company A aligns engineering, analytics, AI capacity, and judgment around an end-to-end research workflow. Pednekar says research shows leaders spend more than 80% of their investments on high-value use cases that transform business functions or create new offerings. The distinction is depth and alignment rather than the number of pilots launched.

### AI leaders redesign work around judgment
[09:59](https://www.youtube.com/watch?v=C2-v6wkHXhY&t=599s)
Adding AI to an existing process is different from changing the work itself. Company B makes earnings transcripts, research reports, financial models, and recommendations move faster. Company A asks an analyst to define an investment thesis, then uses systems to gather evidence that supports or challenges it, monitor developments, investigate exceptions, and examine selected cases. Pednekar says the stronger approach reorganizes roles and capacity around judgment. His example is not just automation of individual tasks. It changes what the analyst does and where human attention goes. He cites a comparison in which high performers are 75% more likely to redesign workflows, while 25% of the other cohort does so.

### Shared capabilities require business and technology to work together
[11:53](https://www.youtube.com/watch?v=C2-v6wkHXhY&t=713s)
Pednekar says companies need shared capabilities after choosing the right problem. The boundary between technology and business is moving. Engineers are working closer to business outcomes, while business professionals are building tools and automations. Organizations can place engineers or product managers near business teams, and can also place portfolio managers in technology-oriented settings. Breaking down silos helps bring the right people into the same room. Pednekar says large enterprises increasingly view technology as a business growth function. He connects that shift to the ability to build capabilities that serve more than one isolated project.

### Standardize the core and personalize the edges
[13:04](https://www.youtube.com/watch?v=C2-v6wkHXhY&t=784s)
Pednekar recommends building common foundations while allowing teams to adapt their outer workflows. In a data science example, the shared core could contain data sets, semantic layers, lessons about trade-offs, records of what worked, and known gaps. A new person can then build on earlier work instead of recreating it. The question he leaves the audience with is what the next person should not have to rebuild. This approach applies across teams with analysts, clients, software engineers, or data scientists. The common foundation preserves organizational knowledge, while the edges can differ according to the team's work.

### Agent infrastructure has to fit cross-workflow work
[14:37](https://www.youtube.com/watch?v=C2-v6wkHXhY&t=877s)
Pednekar says traditional enterprises were designed around a person or group interacting with one application. Agents work differently. They move across workflows and applications, take context, deliberate, and take actions. Leaders therefore need to modernize core infrastructure and understand where the organization is in that process. He identifies data quality and data reuse as common challenges. He also describes several possible agent roles: subject-matter expert agents, workflow agents that can open and close tickets, and chief-of-staff-style agents that track the organization's state and what needs to happen next.

### The system compounds when it preserves learning
[16:34](https://www.youtube.com/watch?v=C2-v6wkHXhY&t=994s)
The final pillar asks what happens after AI is deployed. Pednekar says leading companies examine whether a failure can provide an advantage through learning and whether a success can become part of the system. An individual may learn from a correction, but the organization must also learn. That knowledge can feed back through a new tool, added context, or another change to the workflow. He says the evidence should be preserved with the learning. AI may free capacity, but the organization needs to decide where that capacity goes, monitor the relevant measures, and build a feedback loop around them. Company A compounds capabilities because its work, infrastructure, and learning are connected.

## Notable quotes
- "What is judgment? Judgment is choosing what matters and how to act when there's no single obvious answer." (04:27)
- "The winners just don't just automate, they redesign." (11:11)
- "So, what should the next person not have to rebuild?" (14:31)
- "The individual learned, but did the system?" (17:00)
- "Everyone has AI. Who has the advantage?" (17:46)

## Tools & references mentioned
- Slack

## Who should watch
- You are running many AI pilots but cannot explain which workflows create business value.
- Your team is adding copilots to existing processes and needs a way to decide when the work itself should be redesigned.
- You are building enterprise agents and need shared data, infrastructure, and feedback loops that preserve learning across teams.

## Related talks

- [Why Waiting for the Next AI Model Can Pay Off](https://aietalks.com/talks/why-waiting-for-the-next-ai-model-can-pay-off) (John Stecher, Blackstone & Scott Worland, Norm AI, 20:45)
- [Small AI Teams with Huge Impact](https://aietalks.com/talks/small-ai-teams-with-huge-impact) (Vikas Paruchuri, Datalab, 17:36)
- [Leadership in AI Assisted Engineering](https://aietalks.com/talks/leadership-in-ai-assisted-engineering) (Justin Reock, DX, 18:11)
- [The AI Pivot](https://aietalks.com/talks/the-ai-pivot) (Brittany Walker, CRV & Chris White, Prefect & Bryan Bischof, Hex, 35:16)
- [How to Fail at AI Strategy](https://aietalks.com/talks/how-to-fail-at-ai-strategy) (Hamel Husain, Independent consultant & Greg Ceccarelli, 17:03)
