Issue #60: GPT-5

Issue #60: GPT-5

Author

Upside Team
August 08, 2025

In partnership with:

Motivate Venture Capital

Good morning.

Thanks for joining Upside.org’s ~47,000 founders, limited partners, venture investors, and other subscribers for this week’s startup overview.

We spent a few weeks away, going off-grid to recharge but boy are we back into it!

In this week’s issue: OpenAI unveils GPT-5 for all ChatGPT users, reshaping AI interaction; Meanwhile, Figma's IPO stirs up industry chatter.

But before we dig in, a note to founders from the co-founder of one of our partners, Motivate VC, David Wieland:

After years on both sides of the table, I've come to believe that successful founders and VCs share a peculiar trait: we're professionally dissatisfied people.

I don't mean unhappy—quite the opposite. But there's this constant low-level irritation when you see inefficiencies, broken processes, or unmet needs everywhere you look. Your brain never stops asking "why isn't this better?" and "what if we could..."

This restless curiosity is rocket fuel for starting companies. It's what drives you to believe that yes, massive problems still exist unsolved, and yes, you're as qualified as anyone to tackle them. That combination of pattern recognition and audacious self-confidence is rare and precious.

But here's what I learned the hard way: the very mindset that makes you a great founder can become a liability as your company scales. The constant problem-hunting that serves you in 0-1 mode can distract from the execution discipline needed in 1-100 mode.

I regularly see it with founders who create chaos just so they have hair-on-fire problems to solve instead of doubling down on what's working.

As a startup founder, I observed myself getting bored when things got "easy"—when the core problems were solved and it was about optimization and scaling.

That's when I realized my sweet spot was in the earliest stages, which ultimately led me to seed and pre-seed stage venture capital. Now I get to channel that restless energy into pattern matching across hundreds of companies, staying perpetually curious about emerging problems and the brilliant founders solving them.

To all the founders out there still in the thick of it: that restless dissatisfaction you feel? It's not a bug, it's a feature. The world needs people who can't help but see what's broken and believe they can fix it.

Keep building. Humanity depends on your beautiful inability to accept the status quo.

Well said, David. Now let’s dive in!

Spotlight

The AI giant's latest upgrade aims to set new benchmarks for usability, coding accuracy, and safety in AI completions.

In a bold move, OpenAI has launched GPT-5, a cutting-edge upgrade now available freely to all ChatGPT users. Leveraging significant advancements, the new model promises fewer inaccuracies, better coding skills, and an enhanced focus on 'safe completions'—ensuring user interactions remain productive and secure. This comes at a pivotal time when generative AI is defining workflows across industries, from education to software development. As OpenAI doubles down on democratizing advanced AI, it potentially widens accessibility to state-of-the-art tools that were once exclusive to premium users. GPT-5's rollout is bound to spark curiosity and varied reactions in the tech world. Learn more.

Key Highlights:

  • GPT-5 is now available to all ChatGPT users, making state-of-the-art AI more accessible.

  • The model is designed for 'safe completions,' minimizing the chances of generating harmful or risky outputs.

  • OpenAI claims significant improvements in coding-assisted tasks, catering to the increasing demand for developer-friendly AI tools.

  • Fewer factual inaccuracies and hallucinations make GPT-5 a step forward in making the model more reliable and trusted.

  • The move aligns with OpenAI's mission to expand free access to cutting-edge AI innovation across the globe.

This launch signifies another leap forward in the AI arms race, positioning OpenAI as a frontrunner in the democratization of generative AI. With growing competition from rivals like Anthropic and Google DeepMind, OpenAI's broader accessibility strategy could set a new industry standard. As businesses and individuals explore the boundaries of what GPT-5 can achieve, the focus now shifts to understanding its real-world benefits—and limitations—in critical fields like programming, content creation, and education. While the buzz builds, the broader story is one of an industry racing to make hyper-intelligent AI both mainstream and responsible.

Quick Bites Overview

Here are some interesting quick news items from the tech world:

Lava Sizzles with $5.8M to Power the Agent-Native Economy

Digital payment platform Lava just secured $5.8 million in funding to revolutionize the 'agent-native economy.' The startup aims to empower merchants by providing digital wallets that leverage credits to streamline transactions. This innovation could be a game-changer for businesses looking to simplify payments and reduce friction at checkout. As the gig economy evolves, Lava’s solution offers a compelling way to handle financial exchanges with ease. Learn more.

Rune Raises $24M to Leave Spreadsheets in the Dust

Excel spreadsheets for military logistics? Rune Technologies says 'not anymore.' This Anduril alumni-led startup has raised $24M in Series A funding to launch TyrOS, an AI-powered predictive logistics tool that operates even in internet-deprived environments. By modernizing the logistics backbone with advanced tech, Rune is setting out to make military operations more efficient—and possibly less reliant on last-minute guesswork. The future of defense logistics is looking smarter. Learn more.

Figma IPO: Meme Stock or Masterstroke?

Against all odds, Figma went public after dodging a failed Adobe acquisition—and now it’s making waves. According to Jai Das of Sapphire Ventures, its performance is “a little bit of a meme stock,” adding intrigue to the art of startup exits in 2025. The journey from market uncertainty to IPO success positions Figma as a case study in resilience—and a subtle reminder that tech drama can have happy endings. Learn more.

Hadrian’s Wall of Funding: $260M for Space and Defense Automation

Hadrian Manufacturing just landed a staggering $260M to automate factories for space and defense industries. The company is on a mission to retool American manufacturing with cutting-edge automation systems, ensuring the U.S. stays ahead in producing high-precision parts. By focusing on space and defense, Hadrian is tackling two sectors that demand top-tier quality and speed. Elon Musk, are you paying attention? Learn more.

Confident Security Raises $4.2M to Wrap AI in Privacy Protection

Meet Confident Security, the San Francisco-based startup that wants to be the 'Signal for AI.' Out of stealth with $4.2M, the company has built a tool that ensures data privacy while working alongside AI models. In an era where privacy breaches headline the news, Confident Security’s solution feels less like a product and more like an essential upgrade for AI operations. Move over, insecure algorithms—you’ve been replaced. Learn more.

Startup Shutdown Of The Day :(

Tesla shuts down Dojo, its AI supercomputer gamble

Tesla has officially pulled the plug on Dojo, its ambitious AI training supercomputer project. Initially touted as a cornerstone for achieving full self-driving capabilities, the decision follows the exodus of 20 team members who have since launched DensityAI, a startup focused on data center services. The move casts doubt on Tesla’s timeline for AI-driven autonomy and marks the end of a bold in-house innovation experiment.

Key Points:

  • Dojo was first announced in 2021 as Tesla's proprietary supercomputer for training self-driving AI models.

  • The project faced hurdles including high costs, delayed timelines, and key team departures.

  • Around 20 Tesla employees transitioned to form DensityAI, a new company in the data center sector.

  • Dojo was positioned as a key piece of Tesla’s roadmap to full self-driving, which still hasn't materialized.

  • The shutdown highlights the risks of proprietary R&D in cutting-edge technologies.

Lessons for Founders:

The decision to sunset Dojo underscores the high stakes of developing customized technology in-house, especially in areas requiring enormous resource investment, such as AI and machine learning. Tesla’s pivot serves as a reminder to startups and larger companies alike: when key talent departs and milestones are repeatedly missed, it may be time to reassess strategy. Founders should weigh the benefits of focusing on their core strengths versus reinventing the wheel with proprietary systems. For more details on the Dojo project’s rise and fall, Learn more.

What did you think of today's email?

Your feedback helps us create better emails for you!

Login or Subscribe to participate in polls.