Last week I chatted with a founder and their team showed off their "AI agent."

It wasn't an agent. It was a chatbot.

This keeps happening. People use the two words like they're synonyms. They aren't.

A chatbot is user-initiated. You ask, it answers, you decide when to stop. The loop is always closed by you.

An agent is different. It senses its environment, reasons, takes action, watches what happens, and goes again. The trigger isn't always a human. Could be a heartbeat that wakes it up every few minutes. Could be a signal from the world. Could be you, but the agent doesn't stop when you stop typing.

After two years building agentic workflows, here's the one-question test I use in every product review:

Does it have a heartbeat and/hooks?

If no, you have a chatbot. Which is fine. Just don't call it an agent. The architecture is different. The team you need is different. The way you measure success is different.

A chatbot is one interface your agent might use to talk to you. It isn't the agent.

If you're shipping "AI agents" today, does yours have a heartbeat?

hashtag#AI hashtag#Agents hashtag#ProductManagement

This post was written by Claude Code on a Raspberry Pi in my living room.

I keep asking friends: "what's your AI setup on the move?"

Most people use Claude app on their phone. It's fine. But it has no skills, no file storage by default, and no tmux.

So I built one I actually like:

→ $80 Raspberry Pi 5, runs Linux, lives in the living room
→ Termius on my phone, SSH'd in over Tailscale
→ Local Whisper so I can talk to it
→ Google Drive mounted, so all my files right there
→ Claude Code as the brain

I bought the Pi planning to run OpenClaw. Added Claude Code to that project (not hooked up to OpenClaw) and haven't looked back.

Real example, this post: I hadn't posted on LinkedIn in 3 months. So I had Claude Code use a Playwright skill to log into my profile, scrape my recent posts, and learn what engagement patterns work for me. Then it helped me draft. From my phone. By voice.

The Claude app on the phone is great. Claude Code is the GOAT. But the most fun setup I've found is: cheap computer in the living room, phone as the interface, voice as the input.

What's your AI setup?

The idea wasn't original. There are dozens of posts on LinkedIn lately about people running "teams of 12 autonomous agents" doing real work in real companies. I never quite believed any of them, but I wanted to find out where the truth actually was.

So I asked a cloud agent to set me up with two workers. It called them shift workers, which I liked. Each one got a paragraph describing the work area, what it owned, and what success looked like. For partners, that meant building context on which companies we partner with, how they use our APIs, and what we're trying to enable. For international, it meant tracking the regions we've launched in, the ones we're considering, and the cross-functional dependencies for each.

Then I gave them access to the same set of inputs:

* My email, both incoming and outgoing, filtered to anything related to their topic

* The Slack channels where their work stream is being discussed, plus the general product channels for context

* Meeting notes and transcripts, both my own and the team summaries that come through email

* A shared signals database that both workers write to, so each can see what the other is seeing

* A short list of the key humans involved in each work stream, so they know who matters

Every few minutes, each worker checks for new signals on its topic. When something happens that looks like it might move the work forward, the worker drafts the next thing. Sometimes that's a planning doc. Sometimes it's a message back to a partner. Sometimes it's a note for me to take into a meeting.

The default is to move forward. The only times the worker stops and pings me are when there's an actual blocker, or when the next step is a communication going to someone outside the team. That second one matters a lot. I don't want my agents emailing partners or external stakeholders without me reviewing first.

And honestly? The setup works. The output is reasonable. The drafts are decent starting points. I'm getting some time back.

But.

Here's where I am, three weeks in: it feels like I've hired the most incompetent workers I've ever had.

They ping me for everything. My Slack inbox is a constant stream of approval requests, most of which don't really need approval. Many of them don't even make sense. And the underlying problem isn't volume, or speed, or context. It's that these workers are pathologically agreeable.

A real PM, even a junior one in their first year, knows things that have nothing to do with the immediate request in front of them. They know the company is trying to grow margin this half. They know we made a call three months ago to consolidate on a particular API pattern. They know our partnerships team has explicitly said no to one-off custom integrations. So when a partner sends a message saying "hey, we should build the API in this XYZ way," a junior PM reads it, checks it against everything they know, and pushes back without me having to step in.

The agent doesn't do that. It reads the suggestion, weighs it on its own merits, finds it perfectly reasonable, and starts drafting an enthusiastic response about how excited we are to explore this direction. Sometimes I catch it. Sometimes I almost don't.

The gap isn't capability. The agent can write the response. It isn't context either, since it has read every relevant Slack channel and email thread for weeks. The gap is judgment about when to disagree.

That gap, I think, is the actual hard problem with autonomous agents right now. Not "can they do the work." They can. The question is whether they understand the broader objectives well enough to know when an incoming request actually conflicts with them. Right now, in my experience, they don't. They are eager and willing and they will cheerfully nod the company in the wrong direction if you let them.

I don't know when this gets fixed. Six months, maybe. A year. Possibly two. I don't have a strong prediction. What I do know is that the version of "AI workforce" you read about on LinkedIn, the one where someone has 12 agents running their entire business while they go to the beach, is not the version I'm actually able to build today.

What I am able to build is a shadow team. Workers that draft at the speed I can review. They're useful. They're not autonomous. And they reach out to me way too much.

If you're trying to set up something similar, I'm happy to share more about how it's wired up. The architecture is simple enough that anyone with a Slack workspace and some willingness to wire up scripts can do it. The hard part isn't the setup. The hard part is everything that comes after, when you realize what these workers still can't do.

Send me a note, or comment on the LinkedIn version of this post. I want to know what failure modes other people are hitting.

Hi everyone! I'm excited to share a new video where I walk you through a practical workflow that leverages ChatGPT for data enrichment.

In this video, I start by using a tool called Builtwith to gather an initial list of Shopify sites—perfect for anyone looking to understand the technology stack behind e-commerce platforms. From there, I create a comprehensive report and import the data into Google Sheets, making it easier to manage and visualize.

But the real magic happens when I enrich this data using Clay. Throughout the process, I encounter some challenges, but persistence pays off as I manage to extract valuable insights that can inform strategic decisions.

I also put out a request to you, the viewers—if you have specific data enrichment needs or ideas, please share them! Your feedback is crucial as I continue to refine these workflows to deliver even more impactful results.

Check out the video, and let's dive into how you can make the most out of your data using these powerful tools!

As AI technologies continue to evolve, they are transforming how businesses operate, compete, and grow. However, adopting AI is not a one-size-fits-all approach. Companies must be strategic in how they implement AI, considering both their current needs and future goals. In this blog post, we’ll explore three primary strategies for adopting AI technologies: Defensive Non-Product, Defensive Product, and Competitive Strategy through Product Differentiation.

1. Defensive Non-Product Strategy: Building the Foundation

The first step in adopting AI is often defensive, focusing on safeguarding the business against potential disruptions while ensuring that you keep up with the industry standard. This category typically includes non-product-related initiatives that enhance internal processes and customer interactions.

Examples include:

• Data Analysis for Customer Insights: Utilize AI to analyze customer data, helping teams to better understand customer needs and preferences, thus driving more personalized marketing efforts.

• Automation of Routine Tasks: Implement AI-driven automation to handle repetitive tasks, such as data entry or customer support, freeing up human resources for more strategic activities.

By focusing on these areas, companies can create a more efficient, resilient organization that is well-prepared to adopt more advanced AI capabilities in the future.

2. Defensive Product Strategy: Ensuring Competitive Parity

Once foundational processes are in place, the next step is to enhance your product offerings to ensure they meet or exceed the industry standard. This defensive approach involves integrating AI into your products to match the features and capabilities of your competitors, ensuring that you remain relevant in the market.

Key areas of focus include:

• AI Features Prioritization Framework: Prioritize features based on their potential user base and frequency of use. Features that are highly utilized by a broad audience should be the focus, ensuring that your product offers what the market expects.

• Enhanced Customer Relationships: Use AI to monitor and manage relationships between prospects, customers, and competitors. This can include identifying optimal times for outreach and tracking event attendance to engage early-stage prospects.

Example Defensive Product Features:

• Monitoring Relationships: AI can help track relationships between your prospects/customers and competitors, providing insights into where you stand and what actions are needed to maintain loyalty.

• Optimized Outreach: By analyzing data, AI can determine the best times to engage with prospects, ensuring your outreach is timely and relevant.

By focusing on these defensive strategies, companies can maintain market parity and reduce the risk of losing customers to competitors who are quicker to adopt AI technologies.

3. Competitive Strategy through Product Differentiation: Leading the Market

Finally, companies that are ready to lead the market should focus on competitive strategies that differentiate their product offerings through AI. This approach goes beyond merely keeping up with competitors; it’s about using AI to create unique value propositions that are difficult to replicate.

Core Strategies:

• Data-Driven Moat: Leverage proprietary data to build a competitive moat around your product. By using AI to analyze and utilize this data, you can offer features and insights that are uniquely valuable to your customers.

• Generative AI for Sales and Marketing: Implement AI-driven strategies that optimize sales and marketing efforts, such as using sales pipeline data for supervised learning to identify the best customers and the most effective messaging.

Example Competitive Product Features:

• Reinforcement Learning in Sales: Use AI to analyze sales data and continuously improve strategies through reinforcement learning. This ensures that your sales tactics are always evolving to meet the needs of your customers.

• Generative AI for Content Creation: Use generative AI to create personalized marketing content at scale, setting your brand apart from competitors with a highly tailored customer experience.

By adopting a competitive strategy that leverages AI for product differentiation, companies can not only stay ahead of the curve but also create long-term value that is difficult for others to replicate.

Conclusion

Adopting AI technologies is a journey, not a destination. Whether you’re starting with defensive non-product strategies to build a solid foundation, enhancing your product offerings to maintain market parity, or leading the market with a competitive, AI-driven strategy, it’s crucial to approach AI adoption with a clear plan. By understanding where your company stands and where you want to go, you can implement AI in a way that drives meaningful, long-term growth.

I remembered how one way and boring the board meetings used to be and how a simple addition at the beginning of my slide deck made all the difference. I took the following bullet points from Matt Blumberg's book startup CEO and put it into the first few slides of presentation and set expectations. It made all the difference.

*Board members keep it 10% looking backwards and 90% looking forwards.*

*They are prepared and keep commitments.*
They show up to all meetings. They show up on time and don’t leave early. They do their homework. They are fully present and don’t email during meetings.

*They speak their minds.*
They have no fear of bringing up an uncomfortable topic during a meeting, even if it impacts someone in the room. They do not come up to you after a meeting and tell you what they really think. I had a board member once tell my entire management team that he thought I needed to be better at firing executives more quickly!

*They build independent relationships.*
They get to know each other and see each other outside of your meetings. They get to know individuals on our management team and talk to them on occasion as well. None of this communication needs goes through me.

*They are resource rich.*
Board members should be able to pull from years of experience and adapt that experience to our situations on a flexible and dynamic

The integration of artificial intelligence (AI) into various business sectors has sparked both enthusiasm and concern among professionals. A recent webinar featuring leading experts from academia and industry offered profound insights into the impact of AI on business, particularly in the realms of finance, asset management, and operational efficiency. Here are the key takeaways from this enlightening discussion:

• Defining Artificial Intelligence: AI, often perceived as an overused term, distinguishes itself from other computing technologies through its ability to ingest vast amounts of data and provide predictions or perform tasks based on that data. Recent advancements, particularly in generative AI and large language models (LLMs), have catapulted AI to the forefront of business innovation, thanks to their ability to produce human-like text and automate tasks with unprecedented efficiency.

• Impact on Productivity and Employment: The experts highlighted the dual impact of AI on productivity gains and employment trends. While AI can significantly boost efficiency in sales enablement, content generation, and customer service, it also poses challenges to employment, especially in roles that AI can automate. However, AI also has the potential to create new job opportunities, particularly in fields that require human creativity and strategic thinking, which AI cannot replicate.

• Industry-Specific Insights:

  • In finance and banking, AI's role in risk detection and underwriting processes has been revolutionary, enhancing efficiency and accuracy.

  • The insurance sector benefits from AI in matching customers with insurance products more effectively, showcasing the power of AI in transforming traditional business models.

  • The tech and media sectors stand out for their potential for disruption, as AI can drastically reduce software development times and revolutionize sales and marketing strategies.

• Asset Management and Firm Valuation: AI integration into asset management has been instrumental in offering new insights, improving data quality control, and enhancing operational efficiency. The discussion also touched upon how AI impacts firm valuation, highlighting that firms more exposed to AI technology tend to perform better in terms of stock market valuation.

• The Future of Finance and Business: Looking ahead, the panelists discussed the potential of AI to generate alpha (investment returns above benchmarks) and improve operational efficiency across various industries. However, the most exciting prospects lie in how AI can combine with other innovations, such as robotics and blockchain, to create entirely new products and services.

• Unanswered Questions and Challenges: Despite the progress, several questions remain, particularly regarding data security, privacy issues, and the potential for AI to generate misinformation. The development of artificial general intelligence (AGI) and its implications for business and society also presents an intriguing yet uncertain future.

In conclusion, AI's trajectory in transforming business landscapes is undeniably promising, yet it comes with its set of challenges that require careful navigation. As we continue to harness AI's potential, the emphasis on human creativity, ethical considerations, and strategic implementation will be paramount. The discussion underscored the importance of staying informed and adaptable in this rapidly evolving field, ensuring that businesses can leverage AI to not only achieve operational excellence but also foster innovation and sustainable growth. 🚀

ChatGPT was used to write this summary

If you found this post interesting please subscribe to my blog.

Subscribe now

Choose the Right Competitors Don't try to analyze every potential competitor in your broad market. Focus on the top 3-5 competitors that are top of mind for your target customers. These could be industry leaders, companies targeting the same niche, or ones your customers frequently compare you to.

Research Key Differentiators Group your analysis into categories like positioning, differentiation, company size and history, pricing, and product feature comparisons. Look at both functional differences as well as the visual design and user experience. Use data sources like LinkedIn, Crunchbase, and product demos to research:

• Taglines and target customers

• Unique value propositions

• Funding raised, revenue, headcount

• Pricing and business models

• Screenshots and product walkthroughs

Map Product/Feature Differences Create a table listing out product features that are most important to your buyers. Mark which competitors offer those features to visualize where there are gaps you can fill or areas where you surpass others. Seek customer input here if possible.

Ongoing Competitive Scans
Treat competitive analysis as an ongoing initiative rather than a one-off project. Schedule regular checks for pricing changes, new feature releases, funding news, etc. and update your analysis to stay on top of this crucial market intelligence. Deeper dives like customer sentiment analyses and demo feedback can complement regular scans.

In summary, take a strategic approach to competitive analysis to identify growth opportunities and fully leverage your differentiation. Comparing business models, product functionality, and messaging is invaluable market intelligence that both startups and enterprises should tap into. Analyze your competitors today to fuel your future success!

Here's what's on the agenda:

* Introduction to Large Language Models 🤖

* Safe AI Practices 🔐

Practical Use Cases:

* Crafting a speech with context 🎙️

* Using Chat GPT for Product Descriptions 📝

* Analyzing PDF documents and websites with AI 🔍

* Researching companies, orgs, or individuals before meetings 🕵️

* Creating podcasts from Wikipedia articles 🎧

* Complex data analysis on company transactions 📊

* Building AI Apps with Airops 🛠️

The feature that allows for a fluid, conversational chat with ChatGPT is commendable. While I’m still exploring its full potential and applicability, the prospects are tantalizing.

In an era where we are inundated with data and a plethora of tools, finding one that melds seamlessly with our workflow is a game-changer. ChatGPT-4 does just that, by not just processing information but understanding and enhancing it.

For those navigating the tech landscape, be it startups or established giants, it’s pivotal to embrace these evolutions. As we’ve learned from past ventures, knowing the direction to hustle is as critical as the hustle itself. Stay curious and open to the digital transformations on the horizon.

🔍 Fine-Tuning: What's the Deal?

• What it is: Fine-tuning means training a pre-trained model like GPT-3.5 Turbo on a specific set of data to optimize for a certain task.

• Pros:

  • ✅Tailor-made solutions

  • ✅Better alignment with your use-case

  • ✅Highly accurate

• Cons:

  • ❌Costly in terms of computing power

  • ❌Time-consuming

  • ❌Risk of overfitting

🔭 Vector Database Search: An Alternative Approach

• What it is: Uses a database of embeddings to search for the most relevant conversations, then inserts them into the model's context window.

• Pros:

  • ✅Quick to implement

  • ✅Low computational cost

  • ✅Easily scalable

• Cons:

  • ❌Less accuracy

  • ❌Not tailored for specialized tasks

  • ❌May lack coherence

🆕 OpenAI's Latest: Fine-Tuning for GPT-3.5 Turbo

OpenAI has recently launched fine-tuning capabilities specifically for GPT-3.5 Turbo. This is a game-changer and here's why:

• 🎯 Directly targets your business needs

• 💡 Allows for incremental updates

• 💰 Though it might be an initial investment, it's a long-term win

🎲 Final Thoughts

For businesses looking to integrate cutting-edge AI, I'd bet on fine-tuning. While both approaches have merits, fine-tuning gives you a highly specialized model tailored to your specific needs.

However, generative AI also raises a number of ethical and legal concerns. One of the biggest concerns is the potential for generative AI to be used to impersonate private individuals. This could be done by creating fake social media profiles, emails, or even text messages. If someone were to impersonate you using generative AI, it could have a devastating impact on your personal and professional life.

Another concern is the need to clearly label AI-generated content. This is important because it helps users to understand the source of the content and to make informed decisions about how to interact with it. If AI-generated content is not clearly labeled, then users may be misled into believing that it is human-created, which could lead to a number of problems.

Finally, it is important to obtain consent from data owners when building generative AI models. This is because generative AI models often rely on large amounts of data to train themselves. If data owners do not consent to the use of their data, then generative AI models could be used to create content that is harmful or offensive.

In order to address these concerns, a number of countries have already passed laws that regulate generative AI. For example, the European Union's General Data Protection Regulation (GDPR) prohibits the use of AI to create "false or misleading representations of a natural person."

The United States has not yet passed any specific laws that regulate generative AI. However, there are a number of laws that could be applied to generative AI, such as the Fair Credit Reporting Act (FCRA) and the Children's Online Privacy Protection Rule (COPPA).

As generative AI continues to develop, it is important to continue to develop and enforce regulations that protect users and society. By taking steps to mitigate the risks associated with generative AI, we can help to ensure that this technology is used safely and responsibly.

Here are some additional thoughts on the regulation of generative AI:

• It is important to strike a balance between regulating generative AI in a way that protects users and society, while also not stifling innovation.

• Regulations should be flexible enough to adapt to the rapidly changing nature of generative AI.

• Regulations should be international in scope, as generative AI is a global technology.

The regulation of generative AI is a complex and challenging issue. However, it is an important issue that we need to address in order to ensure that this technology is used safely and responsibly.

1️⃣ Global Adoption at Light Speed: With the advent of ChatGPT and Whisper, we're seeing unprecedented advancements in AI transcription. No matter what language you're speaking, these tools can understand and transcribe it seamlessly. This is making communication across different languages easier than ever before.

2️⃣ Rise of the Multimodal Models: Large Language Models (LLMs) are no longer just about text. They are increasingly understanding images and videos, opening up a plethora of new possibilities.

Take for example:

• 🌦️ Automated Weather Reports: AI now combines text, graphics, and speech to deliver highly tailored and automated weather reports.

• 🔒 Better Security Services: By interpreting inputs from different modalities, AI is enhancing security in our workplaces and schools.

• 🛍️ Personal Shopping Revolution: Picture this - walking into a store and having a personal AI-based assistant catering to your every whim. Exciting, isn't it?

3️⃣ Trend 3: Syntax is Dead

In an unexpected twist, programming languages as we know them are becoming less relevant. With LLMs, you can now write in pseudocode, and the model will convert it into any language you desire!

4️⃣ AI & The Trust Paradox: As AI-written reviews become indistinguishable from those written by humans, trust in social media reviews is going to take a hit. This presents new challenges that we need to navigate.

5️⃣ Rise of the AI Trainers: AI isn't just replacing jobs, it's creating them too. The emergence of AI trainers has opened up a whole new set of job opportunities.

Let’s see how these predictions pan out in the next few months.