Short-form video is becoming an SEO and AI visibility channel for B2B software brands, not just a social media tactic. YouTube Shorts now ranks for more than 1,100 "best + software" searches worldwide, and 92% of those search results also include AI Overviews. Repurpose webinars, podcasts, demos, and customer interviews into short videos that target specific buyer queries, then optimize the title, transcript, captions, and opening hook for search intent. Start with YouTube Shorts because Google cites it far more often than Instagram or TikTok, then publish the same clips on other platforms that fit your audience.
Ramp figured out that its ICP cares about technology adoption data, so it built pages that show trends from its own spend data for individual products and software categories. It updates most pages 1 to 2 times per week and puts key statistics, charts, and dates in HTML so search engines and AI tools can easily read the content. This approach helps its pages appear in Google and AI search results for questions about software adoption and category trends.
AI is raising the floor for PMMs, making safe positioning and surface-level insights easy to replicate. The advantage now comes from judgment. Train teams to explain why decisions are made, not just ship outputs. Take clear, opinionated positions that show who the product is for and who it's not. Go beyond obvious insights by digging into contradictions and talking to customers directly. Most importantly, cut what doesn't matter, since deciding what to leave out is now the highest-leverage skill.
An AI agent built and launched an app by running a tight loop of user interviews, concept testing, building, and iteration. It used Listen Labs to run 2,000 interviews and test 100 ideas in parallel with subagents. Image models handled product builds, while Meta ads drove acquisition. The agent landed on StyleFits, an AI stylist shaped by real user feedback. It pivoted based on low NPS, pricing friction, and trust gaps, then added free trials and shopping links. The result was a scrappy product with hundreds of paying users.
Google's Smart Bidding requires roughly 30 to 50 conversions per campaign per month to exit the learning phase and make reliable bidding decisions. Accounts segmented across too many campaigns spread that signal thin enough to keep every campaign permanently in learning mode. Performance Max compounds the issue when its boundaries with Search campaigns aren't defined through negative keywords and brand exclusions, since PMax will claim high-intent branded queries and inflate costs on terms Search would have won cheaply. No amount of bid adjustments or creative testing resolves these problems when the foundation is wrong.
Specific and verifiable messaging helps buyers more than polished buzzwords. Buyers ignore vague claims like "enterprise grade" and look for proof such as implementation time, migration steps, product demos, customer results, and real user feedback. Marketing copy should describe what a product does, who it is for, and how it solves a real problem instead of relying on vague phrases like "enterprise grade" or "AI powered." AI can speed up research and content creation, but marketers still need human judgment to turn product details into messages that answer real buyer questions.
AI hype creates unrealistic expectations that make it harder for people to recognize the real value AI can deliver today. Social media rewards exaggerated claims, so simple time-saving workflows often get presented as life-changing systems even though many still need significant human oversight. Real expertise comes from building AI solutions that solve business problems instead of talking confidently about AI concepts. Honest examples, measurable outcomes, and continuous improvement build more trust than exaggerated promises.
AI hype creates anxiety and erodes trust. Real progress comes from honest expectations, measurable outcomes, and the sustained work required to make AI reliable.
Execution improves when workflows do not depend on perfect collaboration. Separate learning, debate, and decisions so progress can continue despite unreliable dependencies.
A strong product vision connects the team to a meaningful change for users and guides difficult decisions. It should be collaborative, grounded in customer insight, and tested through real scenarios.
AI is turning security professionals from gatekeepers into builders. Success will require engineering skill, product judgment, and security experiences that developers want to use.
2026.07.03·
franklyspeaking.substack.com·5 min read·TLDR Product
Build AI systems around intelligent routing, not a single model. Send each task to the cheapest reliable option and process nonurgent work asynchronously.
Many product builders blame weak marketing for failure while crediting competitors' success to distribution, all while assuming their own product is better.
SpaceX showed off a prototype for a handset-like device to investors prior to its IPO last month. Designed to run on a proprietary operating system and integrate AI technology from SpaceX's AI, the device would run on a Qualcomm Snapdragon chipset. Elon Musk has previously considered building a smartphone due to frustration over how Apple controls the distribution of third-party apps. A device could help Musk become less reliant on other companies.
OpenAI has proposed handing the US government a 5% stake, worth roughly $42.6 billion, to defuse mounting political pressure. CEO Sam Altman says that giving the public a financial interest in the company is the best way to share the upside of AI. The suggestion is part of a larger proposal in which the government would hold 5% of each of the leading US AI developers via a government vehicle. The proposal was first pitched directly to the Trump administration in early 2025.
Several startups have announced general-purpose robots in the sub-$10,000 range. Nori Robotics is selling a bimanual robot for under $1,400 that's tall enough to reach the top of a table or countertop. The BracketBot is a wheeled robot manipulator with hoverboard-style drive wheels that will sell for under $3,000. Weave's Isaac 1 is a fully capable home mobile manipulator that can put away laundry and tidy a room. It will cost $8,000 or $450 per month.
Biologists packed nonliving components into a cell-like membrane piece by piece, and the bag of molecules started to behave like life. The lab-made cell grew, replicated its DNA, and divided. While the cell is not alive, it is the strongest demonstration yet that it is possible to generate life from nonlife. It can't survive without constant deliveries of food and ribosomes, and it has no defenses or a good waste removal system.
Web Apps for Meta Ray-Ban Display use standard web APIs. This guide shows developers how to build optimized Meta Ray-Ban Display Web Apps. The easiest way to build Web Apps is using AI coding tools and app platforms such as Replit, Claude Code, and Cursor.
ZCode is the official development environment for GLM-5.2. It is available now for download on macOS, Windows, and Linux. GLM Coding Plan subscribers now have a 1.5x usage quota in ZCode.
Meta is developing plans for a cloud infrastructure business that will sell access to AI computing power and models. The business will generate revenue from excess computing power from the data centers Meta built to fuel its AI ambitions. One potential plan is to sell access to various AI models hosted on Meta's infrastructure, similar to AWS Bedrock. Another option may be to sell access to raw computing capacity like CoreWeave.
The debate over the degree to which the US federal government should control access to cutting-edge API tools is now heating up. There is a growing awareness of how powerful these new tools are, but little agreement on how they should be controlled. AI proponents say that bans and implementations would hurt the US in its efforts to stay ahead of China in the geopolitical AI race. Administration officials and advisers are attempting to balance innovation and security while addressing safety risks in the most minimally invasive way possible.
Your models are sophisticated, but they are still wrong when it matters most. Learn how Uber and Domino's are grounding AI with real-world context across their organizations. Download the guide.
Xbox is testing out a disc-to-digital feature that will create a 'digital entitlement' for games that can be moved between Microsoft accounts and Xbox profiles.
The 'usermedia' HTML element reduces boilerplate code, improves security, and provides a seamless recovery path for users who have previously denied access.
Coding agents only move as fast as the inference behind them. FriendliAI is the FRONTIER INFERENCE CLOUD FOR AGENTS — serving frontier open-weight models with the speed, reliability, and cost efficiency agents need in production.
Google is testing a new Gemini Flash upgrade on LM Arena, reportedly offering incremental improvements over the current Flash version. Despite no official confirmation, past Arena tests have often preceded public launches, drawing attention to potential new labels like "Gemini 3.6 Flash" or "Gemini 4 Flash." The Flash tier attracts interest due to its importance for everyday users and cost-conscious developers, handling most free and pay-as-you-go traffic faster than pricier Pro models.
Meta Platforms is developing an internal cloud infrastructure initiative to sell access to its surplus AI computing power and hosted models to external developers. This strategic pivot aims to generate a new revenue stream from the company's massive multi-billion-dollar investments in data centers and chips, directly challenging dominant hyperscalers like Amazon Web Services, Microsoft Azure, and Google Cloud.
Dwarkesh Patel posted some big questions about AI a couple of months ago and received 600 essay submissions for the contest. The three winners of the competition were Jassi Pannu, an Assistant Professor at Johns Hopkins University; Ege Erdil, the co-founder of Mechanize; and Michael Li, a Master of Public Policy candidate at Harvard Kennedy School. The full essays are available in the post.
Autoresearch involves building an 'outer loop' where agents help maintain and improve the primary system using feedback signals, evals, and human input to make progress over time. One of the startups building infrastructure for these self-improving systems is Introspection. This post features an interview with Introspection's co-founder and CEO, Roland Gavrilescu. The discussion covers the shift from agent harnesses to feedback loops, the role of the open-source Pi framework, and why autonomous software factories must first learn from humans.
The ARC-AGI-3 benchmark tests agents' ability to form and update an internal world model through ongoing learning. Continual Harness aims to efficiently enhance agents' self-improvement capabilities on this challenging IQ test.
Frontier models struggle with relatively simple financial tasks. Custom models fine-tuned on high-quality proprietary datasets labeled by expert investors perform better. They are also substantially cheaper. Models of the future will probably feature differentiated intelligence - organizations will have custom models tuned to their specific needs that outperform frontier models.
ZCode is now available on macOS, Windows, and Linux. It combines the best AI agents with existing tools so developers can plan, code, review, and deploy without friction. GLM-5.2 is tuned for ZCode, making agentic coding faster and steadier. GLM Coding Plan subscribers now have 1.5x usage quota in ZCode.
PorTAL is an innovative architecture designed to decouple task fine-tuning from specific base model weights. As the accelerating cadence of foundation model releases creates a compounding financial and engineering burden to re-tune capabilities from scratch, PorTAL offers a way to pay for task adaptation once and amortize it across all future models.
Claude Fable 5 and Mythos 5 have been redeployed. Fable 5 will be included for up to 50% of weekly usage limits until July 7, after which it will be available via usage credits. Mythos 5 has been restored for some organizations in the US. Anthropic is continuing to coordinate with the US government to expand Mythos access to the broader set of domestic and international partners in the Glasswing program.
A joint study by Ramp and Revelio Labs analyzed the relationship between firm-level generative AI investments and employment outcomes across over 21,000 companies in the US. The research reveals that companies with high-intensity AI spending grew their overall headcount by 10.2 percent and entry-level positions by 12 percent over the two years following adoption.