Confluence for 9.15.2024

A way to work between tools. OpenAI model o1.

Midjourney prompt: A painting by Marc Chagall of a person leveraging the combined power of multiple artificial intelligences, capable of deep reasoning. Include a strawberry motif.

Welcome to Confluence. We only have two items this week as we’re doing a deep dive into the new model from OpenAI, o1. But before we get to that, we want to let readers know that, later this month, the authors of Confluence are leading a seminar on generative AI for communication professionals. We’ve done a number of these in the past year, and we cover the fundamentals of how the models work, and how you can use them for editorial assistance, content generation, analysis, thought partnership, and more. They’re also relaxed and a lot of fun. The seminar is two half-days, on Monday, September 30th and Tuesday, October 1st — and it is virtual, so no need to travel. We do have a few spaces open, so if you would like to learn more or join us, you may do so here.

With that said, here’s what has our attention this week at the intersection of generative AI and corporate communication:

• A Way to Work Between Tools

• OpenAI Model o1

A Way to Work Between Tools

How we use the relative strength of different tools to get to better outputs.

We’re often asked which generative AI tool we use most frequently. The simple answer is typically Claude. It’s the best writer, most effective at editing, and produces the best summaries of even complex information. But recently, we’ve found ourselves moving between two tools to achieve the best output. We wanted to share our process, as it might prove helpful in your own work with generative AI.

One area where ChatGPT, specifically its mobile app, excels is in its audio features. We’ve talked about this in the past, and even without the advanced voice capabilities of GPT-4o (which are still slowly rolling out to users), the ability to speak directly to the app and have it create a transcript of what is said — populating the chat history in real-time — is incredibly valuable. So we use it.

When brainstorming an email, memorandum, or even speaking notes for a session, we've found it helpful to talk through our ideas out loud. With the ChatGPT app, we not only end up with a record of our thoughts but can also ask ChatGPT to organize and revise them. This approach has proven particularly effective for capturing the natural flow of ideas that often emerges in verbal brainstorming. Here’s an example of a spoken prompt we recently used to help organize thoughts around a strategic narrative.

I'm going to dictate to you a rough flow of a story, and what I need you to do is create an outline for it. Keep to my language to the extent it makes sense. Let me know if you have suggestions for improving the flow. But really, I need you to make sense of the ideas I'm going to share and turn them into a coherent outline. Does that sound good?

We then spoke with ChatGPT for a solid 10 minutes, working through ideas and thoughts, with plenty of backtracking and self-revision. Doing this allows us to work through ideas in the way that feels most natural to us, making it easier to get to a better output and structure.

Within moments of pressing send on our thoughts, ChatGPT returns a clear and direct outline that accurately reflects our ideas and even voice. This alone is tremendously valuable. Even if we stopped working with AI at this point, it would have both saved us time and improved the quality of the final product.

But knowing that Claude is better at working with text and can write at a higher level of quality, we then took that outline and used it to turn these rough sketches into complete pieces of content. Taking advantage of the most powerful features of the two tools — ChatGPT’s ability to capture and organize spoken thoughts, and Claude's superior writing capabilities — creates a remarkably fast and effective workflow.

The initial output from Claude still isn’t perfect and requires feedback and revision. But even the first draft is of high quality and, most importantly, much closer to the voice and style we’re looking for compared to if we hadn’t followed this process.

The lesson? Be open to moving across tools and models based on their available features and relative strengths. ChatGPT’s audio transcription paired with Claude’s writing prowess creates a sum greater than either of their parts. This flexibility in approach can lead you to better outputs, leveraging the unique capabilities of each tool to get to a better outcome.

OpenAI Model o1

The game may have just changed – again.

For months there’s been speculation that OpenAI had not just a new model, but a new approach to generative AI in the works. It’s been variously called Q* and Strawberry, and this week that model became public when OpenAI announced o1. It’s available in a limited form for preview use for paid users, and there are two versions — o1 and o1 mini. The former is presumably more advanced, the latter, faster. We have access to, and have been getting to know, both.

o1 is important because it represents a shift in how OpenAI’s leading large language model (LLM), ChatGPT, works. As regular readers know, LLMs are prediction machines. They don’t search for and find answers — they create them, a few letters or words at a time, based on an unfathomably complex set of probabilities about what should come next based on what has come before in the sentence and conversation. They are, in essence, operating in a “stream of consciousness” mode. They do not plan, any more than you plan when you are in stream of consciousness mode. As a result, they make mistakes, and their ability to be “thoughtful” is limited (as smart as they seem).

Because of this, users resort to all sorts of tricks to make them more deliberate. We ask them to “think step by step,” and that seems to help, although they aren’t really doing that. We break their tasks into pieces and have them perform them one at a time. We talk to them for a while, then take our output and start a new chat so the models thinking doesn’t get too confounded by what has come before. And so on.

o1 represents a significant shift from this approach. This model goes through a series of preceding steps to understand, deliberate about, plan for, and reflect upon your prompt prior to responding. This is called “chain of thought,” and there’s a good primer on it here. And while OpenAI does not show precisely what this pre-thinking is, it shows you descriptions of the steps and how the model generally did its reasoning (more on this below, and reading these is fascinating in its own right).

At this point an example may be helpful. Here’s a query we gave ChatGPT-4o (the leading current model):

Create a comprehensive list of highly valuable use cases for generative artificial intelligence in the practice of corporate communication. Focus in particular on tasks involved in public relations, media relations, and internal communication. It’s important that this does not focus only on writing tasks but on many other tasks that these professionals engage in or any other basis for which generative AI could be an assistant or productivity aid or thought partner.

This is GPT-4o’s response:

---

Generative Artificial Intelligence (AI) can play a transformative role in corporate communication, especially in public relations (PR), media relations, and internal communication. Below is a comprehensive list of highly valuable use cases for generative AI in these fields, going beyond just writing tasks:

1. Public Relations (PR)

a. Crisis Communication Strategy and Planning

• Scenario Simulation: AI can simulate potential crisis scenarios and suggest appropriate communication responses for each scenario, helping teams prepare crisis management strategies.

• Crisis Response Templates: Generate real-time press releases, talking points, or social media responses in the event of a crisis, tailored to different audience segments and channels.

b. Audience Sentiment and Trend Analysis

• Sentiment Monitoring: Analyze social media, news, and blogs for sentiment around a brand, product, or issue. This can help PR professionals gauge public sentiment before making strategic decisions.

• Trend Identification: Identify emerging topics and trends within an industry that can help PR teams shape proactive communication strategies.

c. Media Pitch Personalization

• Pitch Customization: AI can analyze the interests of specific journalists and suggest personalized pitch angles, ensuring outreach is highly targeted and relevant to the journalist's area of focus.

• Dynamic Press Kits: Automatically generate press kits with multimedia content tailored to specific media outlets or reporters, including data visualizations, videos, or product images.

d. Competitive Analysis

• Competitor Messaging Insights: Generate insights into competitors' PR strategies by analyzing their communications across various channels, helping organizations stay ahead in messaging and positioning.

• PR Benchmarking: Use AI to create benchmarks by analyzing historical PR campaigns, identifying what has worked in terms of outreach and engagement for similar companies or within the industry.

e. Influencer and Stakeholder Identification

• AI-Driven Influencer Mapping: Identify key influencers in a particular industry or region based on audience data and engagement metrics, helping PR teams target the most impactful individuals for collaborations.

• Stakeholder Sentiment Analysis: Monitor and analyze how different stakeholder groups (customers, investors, regulators) feel about a brand, and use this data to tailor communications accordingly.

f. Event and Webinar Planning

• Event Audience Analysis: Use generative AI to analyze previous event feedback and participant demographics to optimize future event planning in terms of content and speakers.

• Event Material Generation: Automatically generate personalized invitations, agendas, and follow-up emails based on audience segmentation, interests, and engagement history.

2. Media Relations

a. Press Release Optimization

• Multi-Channel Press Release Creation: Automatically tailor press releases for different channels (traditional media, social media, blogs) by adjusting tone, style, and length, ensuring maximum engagement across platforms.

• A/B Testing Variations: Create multiple versions of press releases with slight variations in wording or tone, which can then be A/B tested to measure effectiveness with different audiences.

b. Interview Preparation

• Automated Interview Briefs: AI can prepare detailed media briefs for company spokespersons ahead of interviews, including background on the interviewer, likely questions, and suggested talking points based on recent news trends.

• Spokesperson Training Simulations: Use AI to simulate interviews, generating real-time, AI-driven questions to help train company spokespeople under different media scenarios, from friendly to adversarial interviews.

c. Media Coverage Analysis

• Media Impact Reports: Automatically generate comprehensive media coverage reports, tracking sentiment, reach, and engagement levels of stories in which the company is mentioned, and provide data-driven insights for improvement.

• Reputation Monitoring: AI-driven media monitoring tools can continuously scan traditional and digital media to alert PR professionals about negative or positive mentions in real-time, helping inform responsive actions.

d. Journalist Relationship Management

• Media Contact Management: Maintain and update journalist contact databases, and suggest optimal times and methods for outreach based on previous interactions and engagement.

• AI-Powered Newsroom Creation: Generate and maintain a virtual newsroom that creates, categorizes, and updates all media-facing assets (press releases, company news, executive bios) in real time for journalists to access.

3. Internal Communication

a. Employee Engagement Monitoring

• Sentiment and Feedback Analysis: AI can analyze internal communications such as emails, surveys, and chat platforms to detect overall employee sentiment and identify areas where further communication or action is needed.

• Predictive Engagement Tools: Use generative AI to predict patterns in employee behavior and engagement, suggesting interventions to improve morale, reduce turnover, or increase productivity.

b. Internal Messaging Personalization

• Dynamic Content for Intranet: Automatically generate personalized internal communication content (announcements, newsletters) based on employee location, department, and role.

• Personalized Internal Newsletters: Use AI to tailor internal newsletters to different segments within the organization (e.g., leadership vs. entry-level employees), ensuring that relevant information is delivered to the right people.

c. Change Management Support

• Change Impact Analysis: Use AI to analyze the potential impact of organizational changes (e.g., restructuring, mergers) on employee morale and suggest targeted communication strategies to mitigate concerns.

• Automated FAQs and Help Desks: Generate and update FAQs, help desk responses, and internal knowledge base articles based on ongoing employee inquiries and feedback.

d. Employee Onboarding

• Automated Onboarding Material: AI can generate customized onboarding materials, including welcome kits, training schedules, and personalized introductions to key company information based on the new employee's role and department.

• Virtual Training Simulations: Create interactive, AI-driven simulations or role-playing exercises for employee onboarding, helping new hires become familiar with company protocols, culture, and tools.

e. Internal Event Support

• Event Content Generation: Automatically create internal event materials, such as meeting agendas, keynote presentations, and speaker bios, tailored to different internal departments or teams.

• Employee Recognition Programs: Use AI to generate personalized messages or videos for employee recognition programs, drawing on individual achievements and contributions from the past year.

f. Diversity, Equity, and Inclusion (DEI) Communications

• Bias-Free Communication Auditing: AI can analyze internal communications for unconscious bias in messaging and suggest revisions to make language more inclusive and representative of the company’s DEI goals.

• Inclusive Policy Creation: Help HR and internal communication teams generate and refine inclusive policies and materials, ensuring language and structure align with organizational DEI objectives.

4. Strategic Communication and Planning

a. Content Ideation and Campaign Strategy

• Idea Generation: AI can help PR and internal comms teams brainstorm new content ideas for campaigns, employee initiatives, or brand awareness projects by analyzing market trends, audience behavior, and past successful campaigns.

• Long-Term Strategy Optimization: Generate multi-channel communication strategies based on data-driven insights into audience preferences, market trends, and competitor activity.

b. Meeting Summaries and Task Management

• Automated Summaries: AI tools can automatically transcribe, summarize, and extract action points from team meetings, allowing communication professionals to focus more on strategy rather than administrative tasks.

• Follow-Up Automation: AI can generate follow-up emails or task lists based on meeting outcomes, ensuring that nothing slips through the cracks.

5. Legal and Compliance Support

a. Compliance Auditing in Messaging

• Regulatory Alignment: AI can review and audit external and internal communications to ensure that they comply with industry regulations, reducing the risk of legal issues.

• Crisis Communication Legal Review: Automatically flag potentially harmful language in crisis communications that could lead to legal liabilities or regulatory scrutiny.

b. GDPR and Data Privacy Compliance

• Privacy-Safe Content Generation: Ensure that any content created, particularly in PR and internal communications, adheres to data privacy regulations such as GDPR by automatically reviewing and anonymizing sensitive data.

Conclusion

Generative AI has the potential to revolutionize corporate communication by assisting with more than just writing. From strategic planning and real-time crisis response to sentiment analysis and employee engagement, AI can act as a powerful tool for PR, media relations, and internal communication professionals, enhancing productivity, creativity, and precision across a wide range of tasks.

---

And this is o1’s response:

---

Comprehensive Use Cases for Generative AI in Corporate Communication

Generative AI offers a wealth of opportunities to enhance productivity, creativity, and strategic effectiveness in corporate communication. Below is a detailed list of valuable use cases across public relations, media relations, and internal communication, focusing on tasks beyond traditional writing.

Public Relations

1. Media Monitoring and Sentiment Analysis

   • Automated Tracking: AI can monitor news outlets, social media, and blogs to identify mentions of the company, competitors, or industry trends.

   • Sentiment Analysis: Generative AI can analyze the tone of media coverage and public sentiment, providing insights for reputation management.

2. Crisis Communication Simulation

   • Scenario Generation: AI can create hypothetical crisis scenarios to help teams prepare response strategies.

   • Virtual Training: Simulate press conferences and Q&A sessions with AI-generated journalist questions.

3. Event Planning and Coordination

   • Logistics Optimization: AI can suggest optimal dates, venues, and vendors based on historical data and preferences.

   • Agenda Creation: Generate detailed event agendas and schedules tailored to specific audiences.

4. Stakeholder Engagement Strategies

   • Personalized Communication: AI can segment stakeholders and generate tailored messages for each group.

   • Feedback Analysis: Analyze stakeholder feedback to inform engagement strategies.

5. Brand Management

   • Visual Content Generation: Create logos, graphics, and branding materials using AI-powered design tools.

   • Consistency Checks: Ensure all communications align with brand guidelines through AI audits.

6. Influencer Identification and Outreach

   • Data-Driven Selection: AI can identify potential influencers by analyzing engagement metrics and audience demographics.

   • Outreach Personalization: Generate customized messages to initiate collaborations.

7. Strategy Development Support

   • Market Analysis: AI can analyze industry trends and competitor activities to inform strategic planning.

   • Idea Generation: Assist in brainstorming sessions by providing creative concepts and campaign ideas.

Media Relations

1. Journalist Profiling

   • Interest Mapping: AI can analyze journalists' past articles to identify their interests and tailor pitches accordingly.

   • Relationship Management: Keep track of interactions and communications with media professionals.

2. Press Release Optimization

   • Headline Generation: Create compelling headlines that maximize media pickup potential.

   • SEO Enhancement: Optimize press releases for search engines using AI suggestions.

3. Media Kit Development

   • Automated Content Assembly: Compile press releases, company information, and multimedia assets into cohesive media kits.

   • Version Control: Manage different versions tailored for specific media outlets or regions.

4. Interview Preparation

   • Mock Interviews: Simulate interviews with AI-generated questions based on current industry topics.

   • Response Coaching: Provide feedback on spokesperson answers to improve delivery and messaging.

5. Trend Analysis

   • Emerging Topics Identification: AI can spot emerging trends in media coverage to inform proactive communication.

   • Competitor Monitoring: Track competitors' media presence and strategies.

Internal Communication

1. Employee Engagement Initiatives

   • Survey Analysis: AI can analyze employee feedback surveys to identify areas for improvement.

   • Engagement Content Creation: Generate ideas for team-building activities and internal campaigns.

2. Chatbots and Virtual Assistants

   • Employee Support: Develop AI-powered chatbots to answer common HR questions or assist with IT support.

   • Onboarding Assistance: Guide new employees through orientation processes.

3. Intranet Content Management

   • Automated Updates: AI can suggest content updates based on employee interactions and popular topics.

   • Personalized Content Delivery: Tailor intranet news feeds to individual employee interests.

4. Change Management Communication

   • Impact Analysis: Assess how organizational changes affect different departments and generate targeted communication plans.

   • Scenario Planning: Simulate employee reactions to proposed changes to refine messaging strategies.

5. Training and Development

   • Customized Learning Paths: AI can recommend training modules based on employee roles and career goals.

   • Interactive Simulations: Develop AI-driven simulations for skill development and scenario-based learning.

6. Meeting Efficiency

   • Scheduling Optimization: Find optimal meeting times across global teams using AI scheduling tools.

   • Automated Summaries: Transcribe meetings and generate action item lists for participants.

Cross-functional Applications

1. Data Visualization and Reporting

   • Automated Reports: Generate comprehensive reports on communication metrics, media coverage, and campaign performance.

   • Interactive Dashboards: Create real-time dashboards for stakeholders to monitor key performance indicators.

2. Multilingual Communication

   • Real-Time Translation: Facilitate global communication by translating content across multiple languages instantly.

   • Cultural Adaptation: Adjust messaging to fit cultural nuances in different regions.

3. Compliance and Legal Review

   • Regulatory Analysis: AI can scan communications to ensure compliance with industry regulations and legal requirements.

   • Policy Drafting Assistance: Help draft internal policies by analyzing legal documents and organizational guidelines.

4. Accessibility Enhancement

   • Content Adaptation: Automatically generate alternative text for images, captions for videos, and simplify language for better accessibility.

   • Inclusive Design Suggestions: Provide recommendations to make communications more inclusive.

5. Audience Segmentation and Personalization

   • Behavioral Analysis: Segment audiences based on interaction patterns and preferences.

   • Customized Messaging: Generate personalized content for different audience segments to increase engagement.

6. Predictive Analytics

   • Trend Forecasting: Predict future industry trends and public interest topics to stay ahead in communication planning.

   • Crisis Prediction: Identify potential risks by analyzing data patterns and suggest preemptive actions.

7. Social Media Management

   • Content Scheduling: Optimize posting times and frequency based on audience engagement data.

   • Automated Responses: Use AI to generate immediate responses to common inquiries or comments.

8. Visual and Interactive Content Creation

   • Graphic Design Assistance: Generate design concepts for visuals, infographics, and promotional materials.

   • Interactive Tools: Create AI-powered quizzes, polls, and interactive experiences for audiences.

9. Collaboration Facilitation

   • Document Management: AI can manage versions and edits in collaborative documents.

   • Task Assignment: Suggest team roles and responsibilities based on project requirements and individual skill sets.

10. Time and Resource Management

    • Task Prioritization: AI can help prioritize tasks based on deadlines, impact, and resource availability.

    • Budget Optimization: Analyze spending patterns and suggest cost-saving measures.

Conclusion

Generative AI serves as a versatile tool in corporate communication, extending far beyond content creation. By automating routine tasks, providing deep analytical insights, and fostering creativity, AI empowers communication professionals to focus on strategic initiatives and meaningful engagement with their audiences.

This comprehensive list highlights how AI can act as an assistant, productivity aid, and thought partner across various facets of public relations, media relations, and internal communication, ultimately enhancing the effectiveness and efficiency of corporate communication efforts.

---

There is a fair amount of overlap, but we find the o1 response cleverer, with some use cases that we would think of but that GPT-4o does not (like journalist profiling). Vellum AI has a detailed 4o / o1 comparison here, and here are some charts that show how o1 stacks up.

Above o1’s response in the previous example, though, is a note that the model “Thought for 16 seconds”:

When you click on that note, you see this:

Highlighting AI’s assistance

I’m focusing on how generative AI aids corporate communication tasks, such as public relations, media relations, and internal communication, beyond just writing.

Identifying key tasks

I’m mapping out corporate communication tasks, focusing on public relations, media relations, and internal communication, to identify areas where generative AI can be helpful.

Investigating generative AI possibilities

I’m highlighting how generative AI can streamline public relations and media tasks like crafting press releases, media monitoring, crisis communication planning, and media training, among others.

Charting potentials

I’m mapping out generative AI applications, including media analysis, crisis communication, event planning, stakeholder engagement, and strategic development. Each application explores diverse scenarios.

Analyzing media impacts

I’m taking a closer look at how AI tools can revolutionize media relations. From simulating interviews to monitoring media coverage, these technologies streamline tasks and uncover media trends.

Enhancing internal communication

AI aids in crafting chatbots for employee inquiries and generating policies or FAQs. It also personalizes training, fine-tunes event planning, and provides data insights.

Breaking down applications

I'm identifying how AI, especially generative AI, enhances virtual assistants by managing schedules, organizing tasks, creating creative concepts, generating visual and audio content, summarizing meetings, analyzing competitors, and managing social media.

Enhancing compliance

AI is at the forefront of ensuring communications adhere to standards, while generative AI crafts compliant language and aids in drafting policy documents. Ideal candidates possess adaptability and a commitment to meeting compliance requirements.

Enhancing communication

Sources: AI's scheduling prowess optimizes time and task prioritization, while generative AI actively aids in planning initiatives.

Harnessing AI for productivity

AI and generative AI offer a plethora of productivity solutions. These range from real-time translation and ethical communication analysis to crisis simulation and interactive content creation.

That’s interesting in its own right. Also interesting is OpenAI’s official guidance on prompt design for o1, which is very different from what has become common practice for most of us:

These models perform best with straightforward prompts. Some prompt engineering techniques, like few-shot prompting or instructing the model to “think step by step,” may not enhance performance and can sometimes hinder it. Here are some best practices:

• Keep prompts simple and direct: The models excel at understanding and responding to brief, clear instructions without the need for extensive guidance.

• Avoid chain-of-thought prompts: Since these models perform reasoning internally, prompting them to "think step by step" or "explain your reasoning" is unnecessary.

• Use delimiters for clarity: Use delimiters like triple quotation marks, XML tags, or section titles to clearly indicate distinct parts of the input, helping the model interpret different sections appropriately.

• Limit additional context in retrieval-augmented generation (RAG): When providing additional context or documents, include only the most relevant information to prevent the model from overcomplicating its response.

That’s it. In essence, “Don’t hold its hand — it doesn’t seem to need it.”

The conversation online among experts about o1 has also been interesting. In many tasks it does not seem demonstrably better. It doesn’t seem to be a better writer, for example (and we think Claude Opus is the best writer regardless). We haven’t seen many great prompts or “use cases” for o1 yet, except from scientists, programmers, and mathematicians, who occasionally post an “Oh my God!” experience. In a way, o1 is like a tool you find in the garage that you don’t recognize — you know it has to be great at something, but you’re not quite sure what that something is.

Our view, though, is that the immediate utility of o1 is not the point. The point — and we think this point is important — is that this is probably the start of a new phase of evolution in generative AI. OpenAI has admitted that this is just a preview, and that it’s a very thin version of what they intend to build. We expect improvement monthly. But there is something much deeper going on here, and that’s in how OpenAI built o1. To this point, the ability to train a generative AI model was the main scaling factor in how good that model could be. Train it on more text, with more variables, and the better it is. With o1, OpenAI has also used generative AI to test the model, which is the process of improving its ability after training. As the graph from OpenAI below shows, both training and testing improve on a log scale with computing power:

What this means is that we seem to now have a double improvement loop with generative AI, at least from OpenAI. As computing power continues to increase, generative AI will likely benefit twice in the developmental cycle (once for testing and once for training). If so, this should create a substantial acceleration in model capability. (And this is further proof of our friend and Wharton professor Dan Rock’s assertion that LLMs are general purpose technologies).

We’ve been writing Confluence for just over a year.  That first issue talked about how Claude 2 was a good writer and how ChatGPT now had custom instructions so you could give it some response preferences. Since then we have seen Gemini Advanced, custom GPTs, GPT-4o, Claude 3 Haiku / Sonnet / Opus, Claude Sonnet 3.5, Claude Projects, Claude Artifacts, Gemini Gems, interactive charting with GPT-4o, the ability to have text in DALL-E and Midjourney, the release of Midjourney 6.1, GPT voice and (for some) live voice interaction, high quality LLM-generated video, voice replication, and real-time voice transposition into a different language. Just to name a few.

There is a possibly apocryphal attribution to Stanford computer scientist Roy Amara, with him saying in the 1960s that “we overestimate the impact of technology in the short-term and underestimate the effect in the long run.” Reading about and using o1 this week we feel a bit like that. GPT-3 and the models that followed built so much hype that some are considering the whole thing a bubble. o1 and the methods that have created it, on the other hand, seem to be creating a more subtle splash in the first week of release. But we think there is a reason OpenAI chose not to name the model GPT-4.5. They started over, calling it o1. They see it as a new beginning. And whatever GPT-5 will be, the methods behind o1 will be part of the process of creating it. OpenAI seems to see this moment as a fundamental shift in the trajectory we’ve been on. If so, you might make yourself ready for what the next 12 months might bring.

---

We’ll leave you with something cool: Google’s NotebookLM now lets you generate audio discussions of materials you upload (think: papers, reports, articles, etc.). We were blown away by this and will share more about our experience in a future edition of Confluence.

---

AI Disclosure: We used generative AI in creating imagery for this post. We also used it selectively as a creator and summarizer of content and as an editor and proofreader.