OpenClaw & Claude Cowork: AI Agents That Actually Take Action- Tech Tuesday Feb 11

AI Technologies in 2026: Focus on Claude

Overview of Claude's Features

• The presentation highlights advancements in AI technologies, specifically focusing on Claude and its features compared to other LLMs like ChatGPT.

• Claude offers unique capabilities such as co-working functionalities, which allow users to collaborate seamlessly within a single platform, akin to Zapier.

• The "Claude Code" feature is beneficial for both developers and non-developers, enabling the generation of code for various applications through simple prompts.

Browsing Capabilities

• Claude includes a browsing feature that allows it to navigate websites like Zillow for real-time data collection, enhancing its utility beyond traditional coding tasks.

• Users can initiate plans where Claude performs actions such as searching for properties and taking screenshots during the process, simulating manual research tasks.

Market Analysis with Claude

• The browsing function enables users to gather insights about property listings and market trends efficiently without needing extensive programming knowledge.

• Key market insights include price ranges and popular property features, making it easier for users to analyze rental markets or set competitive pricing.

Skills Feature

• The "Skills" feature allows users to automate routine tasks by creating specific workflows that can be executed repeatedly, similar to delegating tasks to virtual assistants.

• Users can input instructions for recurring tasks (e.g., collecting foreclosure information), which Claude can then execute autonomously.

Questions About Skills Implementation

• A question arises regarding the programming language used when creating skills; this leads into a demonstration of how skills are generated within the app interface.

• The discussion emphasizes user-friendliness in skill creation while addressing concerns about visibility into the underlying code structure.

Tax Preparation and Document Processing Using Whisper Flow

Overview of Tax Document Processing

• The speaker discusses the need to summarize financial documents for tax purposes, specifically focusing on income and expenses for the year 2025. They plan to use a tool called Whisper Flow to facilitate this process by converting speech into text.

Utilizing Whisper Flow for Tax Summarization

• The speaker expresses their intention to seek help with taxes, mentioning they have documents from their property manager that cover short-term rental income and expenses from January to December. They aim to create a tax-friendly summary of these documents.

• Before processing all months, the speaker decides to start with January's data first, ensuring accuracy before scaling up the summarization process for subsequent months. This approach allows for review and adjustments as needed.

Technical Challenges and Solutions

• The speaker encounters technical difficulties while using the app version of Whisper Flow but resolves it by restarting the application. They emphasize that having multiple applications running can lead to freezing issues on their machine.

• After successfully attaching files in the app, they note that Whisper Flow is versatile and works across various platforms like Gmail and WhatsApp, enhancing productivity through voice-to-text capabilities.

Generating Tax Summary Files

• As the tool processes data, it creates an Excel file summarizing January's financial information, which will be easier for accountants to handle compared to raw PDF documents. The speaker mentions that itemized bookings may not be necessary; a gross summary might suffice instead.

• The generated skill in Python is designed primarily for creating Excel files but does not yet include functionality for parsing PDF documents directly; thus, additional steps are required for full automation in future workflows.

Future Enhancements and Workflow Automation

• To make document processing repeatable each year, there’s a discussion about adding a parsing layer that would allow automatic extraction of data from PDFs into structured formats like Excel workbooks with monthly tabs or individual files per month. This would streamline future tax preparations significantly.

• A participant raises a question about extracting specific details (like drywall plans) from construction drawings using similar technology; the response suggests experimenting with language models or tools available to see if they can provide useful insights without needing extensive manual review of large documents first.

Understanding Document Analysis and AI Tools

Initial Discussion on Document Types

• The conversation begins with a request for more information regarding the types of documents being discussed, specifically mentioning sketches and written documents like PDFs.

• There is a focus on wanting to extract specific cost information from construction plans without reviewing the entire document, highlighting the need for efficiency in data retrieval.

Utilizing AI for Information Extraction

• It is noted that all necessary information can be found within the document itself, emphasizing regional considerations in construction costs.

• A suggestion is made to use GPT-4 for vision analysis, indicating its potential utility in extracting relevant insights from visual data.

Project Settings in AI Tools

• The speaker mentions having a paid plan for ChatGPT and expresses intent to utilize it effectively by asking specific questions about the document.

• Both ChatGPT and Claude offer project settings where users can upload files once and ask multiple questions over different sessions, enhancing user experience.

Caution Regarding Sensitive Information

• A participant identifies themselves as a Cisco network engineer with prior NSA experience, warning against inputting sensitive information into public LLM (Large Language Model).

• They stress that any data entered could potentially be reproduced or accessed publicly, advising against sharing personal or financial details.

Cybersecurity Concerns with AI Usage

• The discussion shifts to cybersecurity incidents associated with AI tools, citing concerns over vulnerabilities and misuse of sensitive data.

• An example is provided regarding Salesforce's handling of client documents through OpenAI, raising alarms about privacy breaches due to automation replacing human jobs.

Final Warnings on Data Privacy

• The speaker warns that even if users opt out of training models on their data, there are no guarantees that their information will remain private.

• Emphasizing caution again, they reiterate not to input any sensitive information into LLM platforms unless one is comfortable with potential public exposure.

Real Estate and AI Tools Discussion

Importance of Property Management

• The speaker warns about the risks of having one's name on all properties, suggesting that it could lead to lawsuits from various parties including lawyers and tenants.

• A recommendation is made to use a Florida land trust for property management, emphasizing the need for separation of assets.

Overview of Claude's Features

• The discussion shifts to summarizing features of Claude, highlighting its browsing and chat functionalities which are familiar to users of large language models (LLMs).

• It is noted that much real estate data is publicly available, advising against purchasing generic market reports unless they are highly specific.

Automation with Co-work

• Co-work is introduced as a tool similar to Zapier that connects different applications for task automation.

• Users can sync Co-work with Google Drive and Gmail, allowing it to optimize schedules based on past meetings and upcoming events.

Task Automation Strategies

• The speaker suggests identifying manual tasks currently being performed regularly as prime candidates for automation.

• Emphasis is placed on using text commands instead of hiring freelancers for routine tasks, making automation more accessible.

Utilizing Claude Code

• Claude code is discussed as a tool beneficial for those looking to build applications or dashboards without needing extensive programming knowledge.

• While basic coding can be generated by AI tools like Claude code, complex projects still require some programming skills to ensure quality output.

Cursor: An Effective Programming Tool

• Cursor is highlighted as an easy-to-use application for integrating AI tools in programming, suitable even for those with minimal coding experience.

• The free version offers substantial functionality compared to other tools; however, there are paid tiers available for increased usage.

Limitations and Considerations in AI Programming

• There’s no strict time limit on Cursor usage; rather it's based on how much context you build during interactions.

• A user shares their experience with Claude 4.6 regarding limitations in weekly usage caps when asking complex questions.

Planning Phase in Application Development

• The importance of thorough planning before building applications with AI tools is emphasized. This includes creating markdown files detailing desired features and tests.

Understanding AI Code Generation and Security Concerns

Effective AI Code Generation Strategies

• Emphasizes the importance of refining input for better code generation outcomes, noting that generic prompts can lead to subpar results.

• Suggests creating a markdown file to outline tasks, ensuring approval at each step before proceeding to maintain quality control.

• Mentions Cole Medine as a leading figure in AI frameworks, highlighting his contributions on platforms like GitHub.

Following Influential Figures in AI

• Recommends following experts who specialize in tools like Claude or OpenAI for practical insights and frameworks.

• Lists Greg Eisenberg and Nate B. Jones as valuable sources for daily updates and strategic advice on AI developments.

Insights into Daily AI News

• Discusses Nate B. Jones's aggressive posting schedule, providing significant insights into the future implications of AI on jobs and industry trends.

• Acknowledges the popularity of Jones’s content due to its relevance and actionable advice.

Security Concerns with AI Tools

• Raises questions about security risks associated with tools like Claude accessing personal data across devices.

• A cybersecurity engineer expresses strong concerns regarding potential invasiveness of such tools on personal data.

Best Practices for Using AI Securely

• The engineer shares their cautious approach by limiting access to sensitive information while using local LLM models for safer interactions.

• Advocates using multiple smaller models instead of one large model to reduce errors and improve accuracy through internal checks among models.

Future Discussions on Security Measures

• Proposes a follow-up discussion focused on best practices for securing personal data when using AI tools, including email management strategies.

• Highlights IBM Technology as another resource for understanding concepts related to prompting and secure agents in the context of autonomous systems.

Exploring Agentic AI Systems

• Introduces the concept of agentic AI systems that not only process information but also consider privacy implications in their operations.

Discussion on Risk Management and Open Claw

Understanding Risk in Property Management

• The speaker notes that risk is relatively low for their current operations but acknowledges that with a larger portfolio (20+ properties), they would be more vigilant about identifying and mitigating risks.

Introduction to Open Claw

• The discussion shifts towards "Open Claw," highlighting its significant impact and the high security concerns associated with it, suggesting that this topic may require further exploration in future discussions.

Technology Utilization and Automation

• Participants are encouraged to share thoughts on various technologies, including potential automation opportunities or questions they might have regarding these tools.

Cost Considerations for AI Tools

• A participant expresses concern over the cumulative costs of multiple AI tools, indicating a budget cap as they subscribe to several premium services.

• They inquire about strategies others use to manage costs while optimizing usage of AI tools, questioning whether people opt for multiple subscriptions or select only a few essential ones.

Exploring Free LLM Options

• Mention of "Hugging Face" and "Open Router" as platforms providing access to both free and paid LLM options. Users can utilize API keys for various models available on these platforms.

Limitations of Current LLM Models

• Discussion around the limitations of certain LLM models, such as censorship issues where specific queries are not answered due to programmed restrictions. Examples include health-related inquiries being redirected away from direct answers.

• An anecdote illustrates how users can sometimes bypass restrictions by framing questions differently, showcasing the flexibility in querying models like ChatGPT.

Local Model Installation Insights

• A participant mentions local installations of models like Kimmy from Hugging Face gaining popularity among programmers, emphasizing its effectiveness based on download statistics.

Subscription Strategies for Programmers

• A programmer shares their approach: maintaining a subscription solely to Claude while using APIs for other models when needed. This strategy allows them to optimize tool usage without excessive subscriptions.

Recommendations for Tool Usage

• It’s suggested that users identify their primary tasks before selecting AI tools. Trying out different temporary plans can help determine which model best suits individual needs over time.

Document Management Techniques

• Discussion includes methods for managing documentation across different LLM platforms without needing retraining each time. Suggestions include creating comprehensive documents uploaded into systems designed for retrieval augmented generation (RAG).

Learning Resources

• Reference made to Andrew Ng's beginner-friendly courses on deep learning available online, encouraging participants to explore educational resources related to AI technology.

Project Summary and Future Applications

Project Overview

• The speaker requests a markdown file summarizing the project, including its goals, tasks, and outcomes for future reference or uploading to Learning Management Systems (LMS).

Tools for Markdown Exporting

• A free web browser tool is mentioned that allows users to export chat content into a markdown file. It works with Chrome-based browsers like Edge and Firefox.

• The specific name of the tool is not recalled but is suggested to be searched as "free markdown exporter."

Insights on AI Market Trends

• Discussion about the current state of AI tools indicates they are heavily subsidized, with ChatGPT reportedly subsidizing $60 billion last year.

• The speaker warns that as more people use AI tools, prices may increase contrary to mass production principles; examples include rising subscription plans for various AI services.

Investment Opportunities in Data Infrastructure

• Corning Glass is highlighted as a stock worth considering due to their production of fiber optic cables essential for data centers.

• Concerns are raised about restrictions on building new data centers in certain states due to environmental impacts.

Exploring Tool Capabilities and Community Engagement

Summarization Techniques

• The speaker suggests summarizing discussions at the end of meetings to create documents that can be transferred to other applications later.

Community Involvement

• An invitation is extended for community members to share ongoing projects, regardless of completion status, allowing for feedback and spotlight opportunities during future tech discussions.

Lead Generation Strategies Using APIs

Transitioning Employee Roles

• A participant discusses replacing an employee responsible for lead generation by training an open cloud system on this task using a Mac Mini setup.

API Utilization for Real Estate Leads

• Inquiry about finding an API that can efficiently locate "for sale by owner" listings through Zillow instead of relying on web scraping methods.

Recommended APIs

• A specific scraper created by a community member is recommended; it allows users to input URLs from Zillow searches directly.

Additional Resources

• Another API option available via RapidAPI offers extensive search parameters tailored for real estate data collection.

How to Effectively Search for For Sale By Owner Properties

Setting Parameters for FSBO Searches

• The discussion begins with the importance of setting parameters when searching for "For Sale By Owner" (FSBO) properties, emphasizing that this differs from other searches which may require a URL input.

• A suggestion is made to utilize skip tracing techniques, which were found helpful through YouTube videos, indicating the value of online resources in real estate research.

Utilizing Real Estate Agents and MLS Access

• If you have connections with a real estate agent who has MLS access, it's recommended to request daily emails listing new FSBO properties. This can streamline the search process by providing regular updates directly to your inbox.

Community Engagement and Support

• Participants are encouraged to engage with the community for additional support or questions regarding FSBO searches. Both the speaker and other community members are available to assist.

Addressing Membership Issues

• A participant raises a concern about accessing community membership due to credit card issues. They mention finding a charge related to an early sign-up bonus but are unsure about its validity.

• The speaker advises reaching out directly to resolve any membership-related issues, highlighting the importance of communication in such situations.