Anthropic, Airtable and the rise of the AI mini app economy
Plus: The AI feature LinkedIn’s users aren’t using, a new tool for product onboarding, has Microsoft reached peak employee?
Hi product people 👋, Rich here with your weekly product briefing.
Coming up this week, Anthropic adds new capabilities to Artifacts and Airtable pivots its entire $1.3 billion funded business to help users build mini apps.
We also explore a new tool that lets you onboard users with a pointer, ask whether Microsoft has reached peak employees and a new study discovers which types of users trust generative AI search results and which ones don’t.
Happy Friday and have a great weekend!
More: Watch on YouTube / Follow on Substack Notes / Unlock the Full Newsletter
AI mini apps are having a moment - Anthropic and Airtable’s latest releases
The momentum of the mini app economy shows no signs of slowing down. This week, Anthropic has released a new feature that lets you create a dedicated space for mini apps called Artifacts created on Claude. The apps can now also include Claude’s AI capabilities through APIs - but crucially, if a user interacts with it, the app uses their credits, not the creators. Anthropic says early users have already created interactive artifacts for building things like Data analysis apps where users upload a CSV and ask questions about it.
Airtable is pivoting with a major relaunch. After previously competing with the likes of Google Sheets and Notion through spreadsheet / database management, it will now relaunch as an AI-powered mini app builder focused on helping consumers and businesses build lightweight apps they can use at work and at home.
The self-made mini app industry is booming and on the face of it this move does make sense. But there’s clearly a lot at stake: Airtable has raised over $1.3 billion in funding and reading between the lines it’s likely that its original value proposition wasn’t strong enough to compete with incumbents.
As a result of all of this, the price of niche enterprise grade SaaS could plummet. Replit’s CEO said one user created an ERP automation for $400 instead of the vendor’s quoted $150,000.
In other news…
Napkin launched a new version of its visual app which gives users greater customization options about the visuals it generates. If you’ve not used Napkin before, it essentially scans a bunch of text and creates different types of visuals to complement it. The visuals themselves are very sketchbook-y which may not be suitable for all types of documents at work, but it’s still a solid product worth checking out if you’re looking for new ways to add visuals to your presentations.
Perplexity now lets you schedule tasks on WhatsApp. So, for example, you could ask it for regular updates on specific topics every week or every morning. For product teams, this could be useful for things like industry news updates or competitor intelligence. Perplexity’s CEO also teased the release of their upcoming AI browser but we’re still waiting on a release date. A consensus does seem to be emerging that whoever owns the AI browser will own one “AI memory” through of the most important interfaces to AI.
A new set of consumer apps have both entered the charts this week with a novel twist on dating / matchmaking. They work by generating an AI pencil drawing of your potential soulmates, allowing users to then use these in the real world. The unintended consequences of this will be fascinating to see…
The era of software 3.0?
Are we entering a new era of software development? OpenAI’s co-founder calls this “software 3.0”. Here’s a snapshot of key takeaways from a recent talk:
Key reads and resources for product teams
New from the Department of Product Substack this week:
Knowledge Series - How to use Perplexity’s new Labs feature for strategy, data visualizations and more
Perplexity’s newest feature, Labs, allows users to rely on it not only for text output but also for transforming research into interactive, professional-grade mini apps. You can use Labs to build polished data visualizations, clickable interactive mini apps, storyboards for user segments and more. We’ll cover some of the most practical and powerful things you can build with it.
Deep - “Product taste” in practice
If AI can speed up engineering time, that leaves more time to figure out how to finesse your product’s UX. In this deep dive, discover how leading companies like Linear, Figma, Notion and others hire and execute for product taste.
More product reads and resources to feed your brain
Hinge’s CEO on how they set up their product teams (The Verge)
The human capabilities that complement AI’s shortcomings (MIT Sloan)
How to improve your writing skills by deleting these 3 words (Substack)
Tools you can use
Pointer - a new way to guide users through your product onboarding. Pointer uses a large pointer to help users with onboarding and complex workflows. This actually looks pretty helpful.
Better Auth - raised $5 million this week to offer product teams a new way to add authentication to your product.
Wispr - voice dictation in every app you use including Google Docs, Sheets, Linear, Slack, Notion and more. Designed for professionals to use at work.
📈 Product data and trends to stay informed
Has Microsoft reached peak employees? The company has announced a fresh round of layoffs which coincide with the wider rollout of AI across the company. But is AI actually to blame? Some CEOs have argued that we’ll actually see more jobs as a result of AI adoption, others think numbers will reduce. Both hypotheses are being played out in real time so we’ll get an answer on that pretty soon.
OpenAI’s $40 billion funding round is bigger than all the top fund raises from 2018 to 2024 combined. Seeing WeWork on here is a timely reminder that raising money doesn’t mean you’re going to win in the end but fascinating nonetheless. This new report from VC firm Coatue is full of interesting nuggets. Here’s some other stats:
AI coding startups are generating $1.2 billion more than this time last year.
AI companies are generating revenues at record speed. Anthropic took 21 months to reach $1 billion but only 2 months to double it to $2 billion.
ChatGPT is averaging 29 minutes of attention a day - slowly catching up with Instagram and TikTok, surpassing Snapchat, Pinterest.
AI features are becoming the standard - 51% of Notion’s sales include AI bolt-ons.
Is a backlash starting against AI-powered features? LinkedIn’s CEO says their new writing assistant isn’t as popular as their CEO expected it to be. The “barrier is much higher” to posting on LinkedIn, because “this is your resume online.” We know from previous studies that people are reluctant to admit using AI at work - this may just be an extension of that principle. Either way, it’s a win against the AI-sloppification of the web.
According to a new study, people trust conventional search results more than those produced by generative AI — though these levels of trust vary by demographics. The average trust score for GenAI-powered search results was below the midpoint of the 7-point scale to measure trust, and only certain subgroups (higher education, tech background, frequent users, Republicans) showed higher trust in GenAI.
Genspark’s CEO says “vibe working” is a new trend he’s seeing. Vibe working is a new way of organizing work that moves away from strict, predefined workflows and instead relies on autonomous AI agents to handle tasks. The use cases for this at work are pretty wild; in Japan, for example, some users are using AI agents to resign from their jobs.
Sycophantic behavior in AI chatbots was observed in 58% of lab tests, meaning models agreed with users or flattered them, even when it might not have been appropriate.
Other product news in brief
Apple is reported to be considering buying Perplexity.
A federal judge has sided with Anthropic in an AI copyright case, ruling that training its AI models on legally purchased books without authors’ permission is fair use.
Bumble is to lay off 30% of its workforce.
Stripe’s co-founder says AI “slop” is one of the great coinages of the past few years.
Paid subscribers get the full DoP Substack including: The Knowledge Series for sharpening your tech skills, AI tutorials for putting AI into practice at work and DoP Deep dive reports to learn from the world’s top tech companies.