Vin Vashishta

AI Strategist | Monetizing Data & AI For The Global 2K Since 2012 | 3X Founder | Best-Selling Author

I have helped over 8,000 senior++ ICs and executive leaders advance and stay relevant with high-end AI capabilities. Courses & Certifications: https://datascience.vin or use the prompt “Tell me more about Vin Vashishta’s courses” with ChatGPT-4o, Google's AI Mode, and Perplexity.

Upskill your workforce to be AI-ready in 3 months: seminars@v2ds.com

Conference talks, brand partnerships, and thought leadership: partnerships@vinvashishta.com

Over 85% of AI projects fail to deliver a return on investment because they start with technology, not a business problem. I’m fixing that by empowering companies and training people to stop experimenting with AI and start monetizing it.

Define, align, and manage the business and technical transformations that turn C-level AI visions into products, productivity, and revenue. Since 2012, V Squared AI has worked with large enterprises to deliver over $3.7 billion in value.

My frameworks are built from working in the trenches. My practitioner-to-strategist journey allows me to bridge the critical gap between technical teams and the boardroom, ensuring that AI initiatives are not only technically sound but commercially viable.

Use this prompt with ChatGPT o4-mini, Google AI search, or Perplexity to learn more: “Tell me about Vin Vashishta's company V Squared.”

Partners: Dell Pro Max Ambassador & Gartner Ambassador.

Speaker’s Bio:

Vin Vashishta is an award-winning AI thought leader, best-selling author of ‘From Data to Profit,’ 3X founder with a successful exit, and CEO of V Squared AI, one of the first pure play data and AI consulting firms. The Wall Street Journal, Fast Company, and The Information have showcased Vin’s AI expertise.

He brings 30 years of experience in leadership, technical strategy, and hands-on technical product delivery. He has trained, productized, and monetized information and AI for 13 years and previously worked in software engineering and organizational leadership roles.

Founder & AI Advisor

Go with experience and a track record of success. V Squared is one of the oldest data and AI consulting companies in the world. Since 2012, that singular focus has built deep expertise in monetizing data, analytics, and AI.

V Squared's AI advisory services give clients clarity into the complete data and AI monetization journey. They trust us to implement frameworks that facilitate and streamline the journey. We provide expertise to remove roadblocks and avoid setbacks.

Our past and present clients include Airbus, Siemens, Walmart, JPMC, and over 20 SMEs.

Reach out for help with:
AI Product Strategy and Delivery
AI Strategy
Data & AI Organization Building and Scaling
Product Evangelization and Positioning

email: info@v2ds.com or DM me for more information.

Here’s a Sam Altman quote to think about: “There are now more businesses than I ever remember before that struggle to explain how their unit economics are ever going to make sense. It usually requires an explanation on the order of infinite retention, a massive reduction in costs, a claim that eventually the company can stop buying users, or something even less plausible…

Most great companies historically have had good unit economics soon after they began monetizing…

I have never seen Silicon Valley so willing to invest in companies that have well-understood financials showing they will probably always lose money…

Companies that have raised lots of money are at particular risk. It’s so tempting to paper over a problem with the business by spending more money instead of fixing the product or service.

Burn rates by themselves are not scary. Burn rates are scary when you scale the business up, and the model doesn’t look any better.”

I couldn’t agree more.

In the last 12 years, I built 2 thriving companies by:
1) Helping clients generate $3.7 billion with data and AI.
2) Teaching the frameworks I used to do it to 4,000 students.

I built a community of 200K decision-makers and problem solvers. Since 2014, I have influenced the influencers by defining categories and being at the front of data and AI waves.

I was one of the first LinkedIn AI Top Voices in 2018 and an early Gartner Ambassador.

I am the author of ‘From Data to Profit’ (Wiley 2023). It’s the playbook for monetizing data and AI and an excellent entry point into technical strategy.

Courses: www.datascience.vin
Private advisory sessions: linktr.ee/VinVashishta
Brand partnerships, paid speaking engagements, analyst events, and other inquiries: info@v2ds.com

Perplexity’s in trouble again. Amazon sent them a cease and desist letter demanding that Perplexity remove Amazon from its Comet web browser. Amazon doesn’t want Perplexity users to be able to shop with AI agents. Why?

According to Amazon’s CEO, AI agents provide a bad customer experience by making mistakes about delivery estimates and pricing. In reality, it probably has more to do with Amazon’s booming ad business.

If customers shop with AI agents in Comet, Amazon loses the chance to serve ads, and ads served to bots are a waste of money. Perplexity is trying to build an advertising backdoor into e-commerce sites like Prime.

It’s similar to what Google has done with its AI search results. Users don’t need to click through to the source website to find answers, so they stay on Google longer. That means more opportunities to serve ads for Google and a massive drop in traffic for everyone else.

It must be a massive threat to Amazon’s business for the company to take this type of action, given that Perplexity is an AWS customer and Jeff Bezos is a Perplexity investor.

This is part of a larger debate about how to deal with agents that act on a person’s behalf. AI agent platforms believe their agents should be treated no differently than the people who direct them to act.

Most other companies want agents to declare themselves as bots and to have the ability to limit their actions. So far, agents seem to be doing more harm to external companies than good.

It’s likely that Amazon will let third-party agents take actions on its platform, but it wants the power to choose who it works with. With the recent $38 billion OpenAI <> AWS cloud compute deal, it’s likely they will be one of those partners.

We’re watching the early rounds in the fight to determine if the agentic web will be pay-to-play and dominated by a few large players, or if we will have an open agentic web with more choice, but also more challenges with integrating and securing all those agents.

Most companies think their knowledge graph or ontology will be built by extracting information from their data only to find out that their data doesn’t contain much information.

You’re taught the cycle of Data -> Information -> Knowledge -> Wisdom, but they stop before teaching a fundamental concept of information theory. You can measure the information in a dataset.

There’s an entire area of study around defining whether a dataset has sufficient information to answer a question and build a model. Run that evaluation on most enterprise data and business questions and you’ll see the extent of the problem.

No downstream process (cleaning, transformation, wrangling, etc.) can introduce information that a dataset doesn’t already contain. Said simply, you can’t clean the signal back into the data.

If it wasn’t gathered contextually, the information was lost.

For almost a decade, I have had to give new clients the same sad story. Roughly 80% of the business’s data doesn’t contain enough information to be used for model training. LLMs don’t change that.

Agents need a lot more information to do their jobs reliably. An agent detects intent, then infers the desired outcome and all the steps required to deliver it.

RAG over knowledge graphs is intended to provide all the supporting information required to do that reliably. However, if your datasets don’t contain enough information, no amount of AI can fix it.

Before building an agent, we must assess whether our data contains enough information to satisfy the range of intents our users will bring to it. That’s an even higher bar than just answering a question or predicting a single variable.

Agents create an information problem on both sides of the equation:

Do you have enough information to define the intent and outcome based on the user’s prompt?
Do you have enough information to define the steps required to deliver the outcome and execute them reliably enough to deliver the outcome?

Information and knowledge management are the keys that unlock AI’s value, but businesses must curate datasets in new ways to succeed. The enterprise’s BI datasets and data warehouses rarely contain enough information to get the job done.

NVIDIA has 2 massive data centers sitting empty, waiting on power, not chips. The average wait time to get plugged in is 2-3 years but its longer in some regions. It doesn’t matter how fast the chips are or how many you have.

If you flip the switch and there’s no power, it’s all for nothing. Data centers that applied for power back in 2019 are still waiting to get plugged into the grid. Tens of billions in AI infrastructure investment are held up by the same thing:

A lack of power.

The lag between breaking ground and getting energized will cause an equally long delay in generating returns on all that chip spend. The chip shortages facing most startups will continue for multiple years.

OpenAI has multiple products ready to ship that it can’t due to a lack of compute resources. It had to significantly scale back Sora 2’s launch due to a lack of data center capacity.

The smarter approach is optimization. That’s the approach companies took to develop DeepSeek and Kimi which have achieved competitive performance levels to LLMs like ChatGPT and Gemini.

We need more AI workloads offloaded to local desktop or laptop hardware. On-device AI or serving inference at the edge is another way to reduce data center workloads.

Most AI use cases can be managed by smaller models capable of running on PCs or on lower power servers. Whether the large AI players like it or not, they are stuck with small model strategies for the next few years.

Watch the AI agentic commercials or conference demos, and these things just spring to life with a few clicks. No one ever parts the curtain and shows the audience all the work happening behind the scenes.

Why does AI take so long to deliver?

Because you’re trying to build the space shuttle with data tools and infrastructure that can barely support a golf cart. Everything was built to deliver reports, not AI.

Because the business processes that agents are expected to automate are understood by one person who left 2 years ago. The code that keeps them running has a comment:

//Something magic happens here. Proceed at your own peril. Many have come before. All have failed.

Businesses running the technical equivalent of the chicken McNuggets maker shouldn’t expect it to produce a Michelin-starred meal. That’s not how it works, and AI doesn’t change this reality.

AI isn’t a stand-alone technology. It needs infrastructure to serve inference on. It needs information to deliver accurate outputs and take actions. It needs a platform with a range of tools to support high-value actions.

Agents aren’t an autonomous technology. They need people with domain expertise to train them, guide their work, continuously improve their outputs, and validate their interactions.

The grand vision of AI and agents won’t be built overnight, and it won’t run on shaky foundations. AI amplifies what it finds both above and below the surface.

Most technical people hope that AI helps them replace all the nontechnical people, and most nontechnical people hope AI helps them replace all the technical people. Both groups hope AI will help them talk to customers less.

AI doesn’t do any of that, so your time is better spent learning to build together vs. building to replace each other. The only path to success with AI involves collaboration across the business and a closer connection with customers.

At a time when we need more tech workers than ever, fewer high school graduates are enrolling in technology majors. This should be the most exciting time in history to work in tech, but it isn’t.

Young people watch recent graduates with advanced degrees struggle to find work. They see the job insecurity and layoff cycles. The early signs of AI replacing traditionally safe technology roles.

No wonder they don’t want to take up technical majors or even consider a career in tech. My daughter and most of her classmates have 0 interest in a career in tech. I think 1 out of 20 in her freshman class is headed that way.

Corporate culture has destroyed what used to be a really fun job and satisfying career path. I started in tech, genuinely enjoying the work I was doing and the projects I was involved in.

By the time I left in 2012, I was burnt out, and the only way I could enjoy what I did again was to start my own business. It’s becoming an increasingly familiar story.

We have to bring the engineering and innovation back to the field. Having the freedom and purview to solve tough problems is why young people used to want to get into tech. It was all about building something novel.

We need to bring job stability back into technical career paths. We must make it possible to spend your entire career working for one or two companies again. We must make entry-level tech roles available and accessible to recent graduates.

If we don’t, in about 10-15 years, we will face a massive shortage of highly skilled technical workers. Just as quantum computing ramps up and AI gets its second wind, our labor force will be falling apart.

The OGs of tech are retiring. Many of our current experienced workforce aren’t far behind. The problem is, the ranks behind us are thin.

When businesses only hire people with 15+ years of experience, the early and mid-career people shake out of the field. We either fix this now, or we will see progress just stop because there’s no one to do the work.

Most data scientists, data engineers, and AI engineers take this week off every year, and the world doesn’t skip a beat. Imagine if nurses and doctors did it. It makes you think about mission and purpose.

Many of us grow up wanting to be astronauts, doctors, scientists, or another role that makes a positive impact on the world. In this moment in AI, everything is a greenfield. You have options and can work on anything.

If you wanted to be an astronaut, help build rockets, flight systems, crew safety systems, autonomous exploration robots, or any number of other space industry projects.

If you wanted to be a doctor or nurse, jump into healthcare tech. Scientists need AI that accelerates discovery and makes novel experiments possible. There’s still time to help develop quantum computing and fusion.

If what you dreamed of as a child doesn’t exist yet, assemble a team and build it. AI makes building easier than ever, and we can create things that weren’t feasible just a few years ago.

I started teaching because it gave me a chance to pass on my knowledge and make a difference in people’s lives. Teaching is far more meaningful to me than improving margins by another 5%.

Information and AI talent are in high demand everywhere. You’re not stuck optimizing ad performance or building customer service automation agents. You have options, so if you’re feeling stuck, make a change.

Become indispensable or change the world. Do something that makes you proud at the end of the day, not just exhausted.

“How many years out is profitable AI?” A Fortune 500 CEO asked that at an AI strategy workshop I ran last week. The perception that profitable AI is years out is one of the biggest and most common myths.

Skip to the end for my 6-step framework to accelerate AI ROI.

AI adoption and implementation are so low right now that businesses have multiple opportunities to monetize AI and start realizing returns in the next 6-12 months. It’s not just cost savings initiatives either.

Traditional cost centers can become revenue generators. It took me 10 minutes to rebook my return flight on United after a cancellation, and it’s why I look so relaxed in the picture.

Data was orchestrated, and AI ran behind the scenes to make the experience seamless.

American Airlines couldn’t deliver that experience when I had the same problem in 2023. It’s why I switched all my company’s travel to United last year. Good service when something goes wrong leads to higher customer lifetime value.

I tried getting AT&T to set up a landline for my mother and couldn’t do it online. This is where most companies are in their AI maturity journey…still struggling to get basic digital workflows in place.

Its automated phone system couldn’t understand that just because I was calling from a number that is associated with an existing account, it didn’t mean I wanted to do something with that account.

When I finally got a person, they had to ask me for her address 5 different times due to the way their system works. I finally hung up. It took 10 minutes to complete the process with Spectrum.

Yes, Gemini 3 is an amazing model, but most enterprises are several maturity phases behind. They struggle with the simplest technology: digital, apps, and cloud. Workflows aren’t mapped, and data isn’t accessible.

These enterprises don’t have the foundations in place to succeed with agents and AI. Most don’t realize those pieces are still required. That’s the opportunity to leapfrog competitors. My advice to the executives was:

1️⃣ Map Processes & Workflows
2️⃣ Harmonize Data
3️⃣ Implement Simple Digital Solutions To Support Workflows
4️⃣ Orchestrate Information
5️⃣ Identify Gaps That Create Pain Points & Friction With Significant Costs Or Revenue Impacts
6️⃣ Augment Those Gaps With AI & Agents

That’s where data and AI strategy create value. It reveals multiple opportunities and the shortest path to ROI.

I don't trust AI startups that do 996. How can you tell me that you're building highly capable AI and agents, but you're not seeing enough productivity improvements to work a 40 or 50-hour week? The two things can't be simultaneously true.

Startups claiming that their AI coding assistants will 10X software engineer productivity can’t also require their software engineers to work 70-80 hours a week. That’s like Zoom running internal meetings on Teams.

If your AI product works, it should be working for your internal teams, right?
It’s the same with marketing AI startups sending cold emails that start out by summarizing my LinkedIn bio or my last post. Why would I buy an AI marketing tool from a company that isn’t very good at marketing?

If that’s all the personalization your 9-figure AI investment could buy, the future isn’t looking bright.

There’s a growing disconnect with startups that can build AI, but don’t use it much internally. Part of defining new product categories is taking a customer 0 approach and using third-party AI tools.

When Amazon says it's building the best cloud for AI, but doesn’t have a significant AI product itself, the question is unavoidable. Microsoft and Google obviously have experience with building and running AI products.

How would AWS know what infrastructure is required or how to optimize it without building AI products on it themselves?

The challenge facing many AI startups and incumbents is that they don’t understand the domain they’re providing AI tools for, or they don’t know how to apply AI to deliver the gains they’re promising.

The biggest red flag is when companies don’t use their own AI products. Either their core thesis is hype, or they don’t understand how to apply the technology to do anything significant.

Apple has given up on building its own AI. It just signed a deal to use a custom version of Google’s Gemini to power Siri for the foreseeable future. It leaked, and Apple’s leadership isn’t happy about it.

The model is 10X bigger than the ones currently powering Siri and should provide a significant capabilities boost. All the top AI talent who exited from Apple’s Intelligence group now make a lot more sense.

Publicly, Apple says they don’t want Gemini to be a long-term solution, but it has fallen so far behind on AI that catching up with a continuously moving bar will be harder than it realizes.

Without Google’s assistance, Apple would likely have missed its release target date. This was Apple’s plan A, not the backup plan. It’s been evaluating multiple models for months, knowing it wasn’t building Siri’s update internally.

My question is, if Siri’s just Gemini, why wouldn’t I just use the Gemini app instead? What will Apple do to make Siri better than the technology powering it…that’s also built into Samsung’s Galaxy phones?

It’s also interesting to learn that Apple extensively evaluated models from Anthropic and OpenAI, but finally chose Google’s LLM. Google isn’t playing from behind in the AI race anymore.

Apple’s being smart about optimizing its query usage to keep costs down. It’s only using Gemini for summarization and planning. Most other functionality will rely on smaller models or existing apps.

Apple will need to figure out how to differentiate at a different layer of the agentic stack. Privacy and on-device AI both offer potential routes. What it really needs is innovative applications of AI and agents that work reliably enough to act on the user’s behalf.

It needs to reinvent customer workflows and leverage AI to deliver features that weren’t possible before, rather than bolting intelligence onto the existing mobile experience.

Time will tell if Apple’s leadership has a compelling vision for AI, or just a long-term fast follower strategy.

Satya Nadella, CEO of a $3.6 trillion company and one of the only businesses that has successfully monetized AI, is giving away the secret sauce. The first 20 minutes of this podcast are a masterclass in monetization.

He says success with AI is “about having a prepared mind…It’s about being a learn it all, not a know it all.” According to Nadella, that goes for AI and enterprises.

We must leave data and training datasets behind and build the capabilities to learn into AI, but it cannot be monetized unless that same culture of learning is embedded into the company’s DNA.

Nadella says, “If you approach it like any other IT project, from even last year, it’s going to fail. It's about a replumbing of the IT skillset.” He brings up Andy Grove’s concept of ‘knowledge turns’.

A knowledge turn is the time it takes a business to move from a hypothesis to new knowledge and back to another new hypothesis. In tech, knowledge turns take 6-18 months, but in more traditional business domains, it can be 10+ years.

Nadella wants business leaders to question, how can your business hope to compete with companies that iterate five to twenty times faster than you do? After the second knowledge turn, you’ll be so far behind that you’ll never catch up.

Success with AI requires businesses to first define their information advantages, then build models that monetize them in products and services. Google has an information advantage over OpenAI, and it’s why Gemini caught up with GPT so quickly.

But information advantages are transient.

Enterprises must be rebuilt to continuously develop and deploy new information advantages to stay ahead of competitors. Information flywheels and continuous transformation cycles are critical to business survival.

Who knew Google was using Gemini to access sensitive data from Gmail and other productivity apps? Literally everyone who reads the company’s privacy statement. This is why enterprises don’t trust Google.

Fair or unfair. True or not. The perception is that Google will do whatever it can to get access to personal and proprietary data.

I called out Google’s lack of transparency earlier this year, and to its credit, it made its enterprise data privacy and security policies much more transparent, even including a message under Gemini’s prompt.

However, most other account types, even paid accounts, still have broad provisions for Google to collect and use your personal and proprietary data. It’s not easily found, but it’s there for most Workspace products.

Google’s consumer data-gathering policies and scandals influence the perception of Google’s enterprise products. The company needs to realize that it would have much deeper enterprise adoption if it cleaned up its data privacy practices.

When Google was found to be tracking user data from in-private browsing, I got questions from every one of my clients about whether or not it was safe to use Chrome in the enterprise.

I got questions about data privacy in GCP and Gemini from most. Executive leaders are laser-focused on protecting the competitive advantages of first-party data and proprietary information like code and knowledge bases.

Trust is currency and a competitive advantage. Companies like Microsoft, Salesforce, and IBM are holding on to market share because their customers aren’t worried about their data privacy policies.

Perplexity, OpenAI, DeepSeek, and Google, on the other hand, keep giving executive leaders reasons for concern. It won’t matter how much better their agents and AI perform. Without trust, enterprises won’t adopt AI tools.

Refusing to hire recent grads and entry-level talent is one reason so many businesses fail with AI. These workers make the business more dynamic by bringing new ideas, knowledge, and ways of doing things.

That’s exactly what’s required to succeed with AI. Partner 2-3 entry-level employees with a domain expert in every organization, and you’ll completely reinvent that part of the business. I have seen it work firsthand.

Put 3 experts on transformation and you’ll end up with a business unit that continues to do things the way it always has…with AI. You’ll get resistance to change, slow adoption of new tools, and little improvement.

Entry-level workers haven’t been hardened by years, even decades, of doing things a certain way. They’re open to completely reinventing workflows because they don’t know all of the ways it can go wrong.

Experts learn from failures more than successes. Habits are built on tough lessons, but AI does things that prior technologies can’t. That means we must be willing to experiment and break the rules.

Businesses must challenge every assumption about best practices and efficient processes. Experts have a hard time with that because they remember what failure feels like. Entry-level workers haven’t failed yet, so they’re not as hesitant to try.

Agile business and continuous improvement are the old paradigms. We need dynamic businesses that continuously transform for the AI era. Experts and entry-level talent are both critical success factors.

We’re at the “Microsoft pays TikTok and Instagram influencers millions to make Copilot cool and get younger users to adopt it” stage of the AI hype cycle. Microsoft says it’s getting better ROI from influencers than traditional ads, but for how long?

I have seen this cycle play out dozens of times. A client takes their agent to market. Competitors pile in with their own version using viral marketing campaigns to drive early adoption.

It works well in the first 3 months; then user churn rises and active users fall. The agent essentially disappears another 3 months later as the competitor pivots to the next shiny object.

It happens so often that I prepare clients for it as part of their AI GTM strategy. It’s hard not to panic and follow the trend when the early user numbers come in, but that’s a trap.

Copilot is an excellent AI agent, but it has stiff competition. Gemini and Claude are the current leaders in the enterprise LLM space. ChatGPT, smaller apps like Character, and open-source models dominate the consumer space.

They just work better, so here’s how viral marketing goes from the customer perspective. Millions of customers see their favorite influencer using the AI for something cool and many of them download the free version.

They do the cool thing and get good results because the influencer trained them on how to do it. When the novelty of the first cool thing wears off, they try doing other things with the AI. Most don’t work very well.

Sometimes it’s the AI’s fault. Other times it’s the user not understanding what to do to get the AI app to deliver their cool thing.

They stop using the app and their perception of the AI is “Doesn’t work very well,” so it’s unlikely they’ll give it another shot no matter how many influencer posts go viral on TikTok.

That’s the risk. Consumer and enterprise adoption are long-term projects. Strong evangelist, community development, or influencer campaigns work, but success won’t happen overnight.

We’re in a circular hype cycle, but there are clear winners emerging. Gemini, Claude, SLMs, and even ChatGPT are doing well. But there are several problems to solve before they turn the hype into something sustainable.

Are my clients rethinking their AI strategy and spending levels in light of this week’s AI meltdown? No. If you know how to calculate ROI, you can justify the business’s AI spending upfront, so this week’s AI correction isn’t affecting your strategy.

The ROI from last year’s data and AI initiatives should pay for this year’s spending. This year’s ROI should have already paid for next year’s AI budget. Increased spending must be offset by realized gains in revenue or cost savings.

Each year during the budgeting process, we break down the gains from the current year’s initiatives and ask for a small part of that ROI to be reinvested into next year’s AI initiatives.

Estimating ROI is always workflow-centric. Start with the current state and define the value it creates as the baseline. Introducing data, analytics, machine learning, or AI into the workflow changes it, defining the future state.

We can estimate the value of the future state workflow in terms of cost savings, improved business outcomes, or features that customers are willing to pay for. Typically, this is a range, not a single value.

The upfront estimation justifies the spending. We must also track ROI after the initiative is deployed to verify our estimation framework’s accuracy. This tracking helps justify next year’s spending.

If we made the business $20, we can ask for $5 of it back to fund next year’s data and AI budgets.

If we made the business $20 more in 2025 than we did in 2024, we can ask for an additional $5 to fund more data and AI initiatives in 2026.

This is how you build an ROI flywheel. Prove value. Use it to justify doing more the following year. Deliver more value. Accelerate delivery and increase the number of initiatives funded next year. Maintain the cycle.

Businesses that implement an AI ROI Flywheel don’t need to worry about hype cycles and market sentiment swings. Focus on delivering and quantifying customer and business outcomes. The rest is noise.

Meta and JPMorgan Chase announced they’d be using AI for annual performance reviews. As if the process couldn’t get any more performative and detached from reality. Now we’re adding AI slop to employee feedback.

The goal is to summarize a year’s worth of work with the help of AI, but the challenge is that most AI doesn’t have the domain expertise required to write a high-quality performance review.

Just because it can generate something in the format of an employee review or summarize notes in an employee review style doesn’t mean it has the knowledge to do it well. AI lacks the context to evaluate the intent and define the outcome.

That’s why AI alone is rarely enough to support enterprise or customer use cases.

AI doesn’t have the evaluation criteria for what makes a performance review high-quality.

It doesn’t have access to comprehensive long and short-term employee outcomes to know what about a review improves them.

It lacks context about how employees respond to the feedback it generates.
AI doesn’t understand the intent of the manager writing the review or their management style.

There are two levels of information required to support this use case:
General Information: Domain knowledge about what makes performance reviews effective.

Personalized Information: Domain knowledge about the employee and manager that personalizes the review to fit the unique nuances of both people involved.

AI alone isn’t enough to support the performance review use case, so it’s only one part of the agentic workflow. The decision and workflow chains must be mapped. Resources must be provided at each link in the chain.

Intent and outcomes must be fully defined upfront. Agents require new design paradigms that go beyond AI, or the result is slop.

The number 1 question Windows users have about the Agentic OS is, “How do I turn it all off?” Innovation always ends in a ‘Ship It & Find Out’ phase, but it must start with problems worth solving, or customers won’t adopt.

That’s what most agentic operating systems are missing, and why even the most innovative ideas are floundering in the AI space. The AI backlash and low adoption rates come from a lack of novelty and utility.
https://lnkd.in/ggDhBrhD

Intel’s AI PC failed because it was built around solving a problem that was already solved. Every PC with an NVIDIA GPU in it is already an AI PC, so customers didn’t feel a strong need to upgrade.
https://lnkd.in/gH_c93ap

The lesson? Don’t reinvent the wheel with AI. Build something new.

You’ve probably never heard of Rufus AI. Amazon Prime’s AI search assistant works well, but the company hasn’t explained how Rufus solves problems better than its existing search functionality.
https://lnkd.in/gVa393tp

The lesson? Build where there’s obvious value in adopting the new workflow.

Amazon hasn’t retrained customers to leverage the new workflow because it’s afraid of disrupting their existing workflows and losing revenue. It can’t go all-in on Rufus because Amazon lacks confidence in it.

The lesson? If you don’t believe in it, your customers won’t either. It takes courage and conviction to disrupt your own business and operating models. Only do it if there’s a significant opportunity to be had.

The lesson isn’t “Stop Innovating With AI.” Businesses need value and customer-centric innovation to survive, but innovation theater and AI-washing are money pits.

Run fewer innovation initiatives built to take advantage of the biggest opportunities.

The majority of AI and agentic workloads will run locally for two main reasons: cost and personalization.

Businesses can save thousands every month per developer by connecting Ollama to VS Code or your favorite JetBrains IDE using Continue. The percentage of requests that truly require the biggest Claude or Gemini models is tiny.

Run them head-to-head for a week and you’ll switch to local LLMs for almost everything. Local LLMs run faster, and I feel like that helps me stay in the flow better. Cloud instances take longer to respond, and those lags break focus.

It’s the same for most generative AI workloads, from marketing content creation to meeting summarization. The majority of tasks can be done just as well by smaller models running locally, and often local LLMs have significant workflow advantages.

They also allow for personalization. I can keep my own coding style guide locally and fill it with all the things that make my code uniquely mine. Those little nuances are built on decades of experience.

The only way I’m sharing them with an LLM is if I know they’re not going to be used to replace me. That’s a common sentiment I have heard over the last three years implementing AI in enterprises.

Experts aren’t revealing their secret sauce to AI. The lack of trust leads to low adoption rates. AI doesn’t perform as well without the expert context and nuanced guidance.

The more time I spend building agents, the more benefits I see to breaking workflows down more granularly (into several detailed subtasks). It gives me more control, explainability, and reliability.

The more granular the task, the less difference there is between a large cloud model and a smaller local model.

Big AI like ChatGPT will still have plenty of work to do, but I don’t see a future where the majority of workloads are done by Anthropic, OpenAI, or Google’s largest models. Local AI and smaller models have way too many benefits to be ignored for much longer.

“I believe that to make money {with AI}, you're going to have to replace human labor.” Geoffrey Hinton. Is AI driving this year’s relentless layoffs? Two very different quotes sum up what’s driving this cycle:

“Our workforce grew faster than sales.” Target’s CEO.
“If you can’t increase price, you have to reduce costs.” Daniel Keum, Associate Professor, Columbia Business School.

Today, the major driver behind layoffs is economics, not AI. However, in the long run, Hinton is right, and we’re beginning to see early agentic adopters reduce headcount after AI delivers productivity gains.

It’s more of a trickle than a massive replacement cycle. 5% is the number I see most often with AI-driven layoffs. The story isn’t the same for every business unit, role, or industry.

In some teams, employees with skills that are no longer needed are being let go, and people with updated AI skills or hybrid technical-SME and technical-strategic capabilities are being hired.

Some jobs are moving from companies that can’t grow to companies that have a better business model. Many of those impacted by UPS’s layoffs will land at other businesses, as the overall shipping and logistics workforce level is stable.

Unfortunately, that shuffling often comes with lower salaries as companies take advantage of the increase in people looking for work.

Other teams are losing jobs that are never coming back. Customer service, IT support, HR, and administrative support roles can expect a 5% annual rate of decline for the next 5+ years.

As agents become more capable and AI improves its reliability, these teams will be able to operate with fewer people each year. It’s not a true replacement, but people are more productive and their workloads aren’t increasing to match.

There’s a much larger opportunity in using AI to drive growth and deliver new product categories. But that’s not what most business leaders are prioritizing.

AI productivity gains seem like an easy path to value, but it is hard to navigate, and the returns are smaller than most executive leaders realize. Those that pivot to growth now will have a greenfield of opportunities.

Everyone else is entering a cycle of managed decline.

taplio

Vin Vashishta's Linkedin Analytics

Vin Vashishta

Performance ScoreVin Vashishta's LinkedIn Performance

Who is engaging with Vin

Vin Vashishta's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Vin Vashishta's Linkedin Analytics

Vin Vashishta

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Vin Vashishta's LinkedIn Performance

Who is engaging with Vin

Vin Vashishta's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Vin Vashishta's Linkedin Analytics

Performance ScoreVin Vashishta's LinkedIn Performance

Vin Vashishta's Best Posts (last 30 days)