In partnership with
BREAKING STORY
Britain is investigating Meta's plan to crawl business websites for AI
Why it's interesting: Meta planned new terms for businesses that advertise or run Pages on Facebook and Instagram. According to PPC Land, they would have let Meta send "scrapers and crawlers" to your website and use what they collect in its products. You would have agreed simply by carrying on, with no partial opt-out. On 30 October the UK Competition and Markets Authority opened an investigation into whether Meta is using its dominant position to impose unfair terms, including using business content for AI features without giving businesses a choice. Meta paused the change on 28 October to consult industry groups. The CMA has not decided whether any law was broken.
What it means for you: if you sell or post through Facebook or Instagram, your product pages, photos and articles are exactly what an AI model learns from. The new terms are paused, not withdrawn, so this will come back.
How to apply it: read any terms notice from Meta before you click through. Open yoursite.com/robots.txt and check it names Meta's crawlers: Meta says Meta-ExternalAgent crawls for uses such as training AI models, while FacebookExternalHit builds the link previews your posts rely on. Block the first if you wish; keep the second. The draft terms made no promise to honour robots.txt, so also decide, in one sentence, what you are willing to share.
MARKET TRENDS
DoorDash took 993 million orders. AI agents placed a few hundred.
Why it's interesting: DoorDash, the US food and grocery delivery app, reported 993 million orders for July to September, up 28% on a year earlier. It has built its own assistants, Ask DoorDash and Text DoorDash, and an MCP connection (a standard plug that lets AI assistants use a service) so outside chatbots can order for you. Yet chief executive Tony Xu told analysts that third-party agents had placed "maybe several hundred orders in total", TechCrunch reported. The ability exists; almost nobody uses it yet.
What it means for you: if you run an online shop, AI placing orders on its own is a later bet. What is happening first is "AI recommends, human buys": someone asks a chatbot what to get, then clicks through and pays themselves.
How to apply it: don't build a custom AI checkout. If you are on Shopify, TechCrunch reports agent checkout is rolling out to "all eligible Shopify merchants", so let your platform carry it. Spend the hours on being the shop an assistant recommends: complete product details (sizes, materials, prices, shipping), an honest comparison page against the alternatives, and reviews on every product page.
GROWTH HACK / DIY
The "Made with" badge: how Tally passed $6M a year with ten people
Why it's interesting: Gamma, the AI tool that turns a prompt into a finished slide deck, told SiliconANGLE it has had about 110 million sign-ups for its free version. Decks shared by free users carry a small "Made with Gamma" badge that only paid plans remove. Tally, the bootstrapped form builder, shows what this does for a small team: its founders wrote in September that it passed $6M in annual recurring revenue with 10 people, and that for five years its growth was word of mouth plus the "Made with Tally" badge.
What it means for you: if other people see what your users make, every shared piece introduces you to a warm viewer who has seen your product work. As an illustration only, 2,000 pieces a month seen by 50 people each, with 0.5% clicking and a fifth of those signing up, is 100 new users a month without ad spend. It does little if the output stays private or viewers will never become makers.
How to apply it: add a small "Made with [Product] · Make your own free" line to free users' output, in a footer or on a thank-you screen. Send clicks to a template page with no sign-up wall, tag the link with ?ref=badge so you can count it, charge to remove the badge, and review the numbers after 30 days. Our estimate is a day or two of building.
It's time to set up your holiday streaming campaign
Are you ready to reach holiday shoppers on CTV? For many streamers, the season is already underway.
Last year, according to Roku internal data, holiday content searches on Roku saw an 86% increase during October. And in November, those searches rise an average of 39% weekly. This surging holiday spirit is mirrored by consumer activity, as about half of Roku streamers shopped Black Friday and Cyber Monday deals in 2025.
Meeting those shoppers is easier than you might think. Begin with the seasonal outcome you hope to drive: awareness, online sales, leads, store visits, or another measurable goal. From there, define your audience, upload your creative, set the budget and campaign dates, and establish measurement before launching. You can even create a shoppable ad unit that lets streamers buy with a click of their remote.
Our holiday playbook offers a month-by-month guide to reaching shoppers during this all-important time of year.
First-time spenders get a $5,000 match on the first $5,000 spent with code BEEHIIV5K. Terms apply.
AGENT LAB
Self-scheduling agents: which chores to hand over and which to keep in chat
Why it's interesting: most AI agents, assistants that can take actions as well as answer, wait to be asked. LangChain, maker of a popular agent toolkit, says more than 70% of runs on its Managed Deep Agents platform now start from a schedule rather than a person typing, according to its post on proactive, always-on agents. The agent wakes on a timer and can even book its own follow-ups.
What it means for you: recurring chores stop depending on your memory: inbox triage, chasing unpaid invoices, a Monday look at competitors' pricing pages, turning each new post into social drafts. By our rough estimate that could be half a day a week back for a busy solo operator. The risk is an agent that acts too early or too hard; the HALO study found those two faults caused 61.20% of rejected proactive responses.
How to apply it: let a job run on its own only if it recurs, its inputs arrive without you, you can review the result later and mistakes are cheap to undo. Keep it in chat if it needs your taste, spends money or contacts people. This weekend, set up one read-and-report job on a schedule in n8n, Zapier or GitHub Actions, set it to notify you only, and keep payments and client messages waiting for your approval.
TRENDING AI RESEARCH
When people love an image, they assume a human made it
Why it's interesting: a new paper, Beauty Is Human, by researchers at Stevens Institute of Technology and UMass Boston, reanalysed a public test in which about 11,000 people guessed whether each of 50 pictures was made by a person or by AI, and picked a favourite; the authors analysed the 9,152 who named and judged one. On average they got 60.6% right, only a little better than a coin toss. They called their own favourite human-made 76.6% of the time, about three in four, against 54.6% when others judged the same pictures. The gap was almost identical whether the favourite was really made by AI or by a person, so liking, not skill, was doing the judging.
What it means for you: if you sell art, design or content, quality alone will not prove you made it by hand. If you use AI images in marketing, customers may assume a designer made them and feel misled later. Say nothing and your audience fills the gap.
How to apply it: show eight of your images, unlabelled, to five customers and ask which they like and who made each. Write one plain disclosure line, such as "sketched by hand, coloured with AI assistance", and use it everywhere. Then test a listing with and without it. One caveat: the test drew a self-selected blog audience and deliberately hard pictures.
TOOL REVIEW
Riverside Agents: four AI helpers that prep, edit, clip and promote your episodes
Why it's interesting: Riverside, the remote podcast and video recording tool creators have used since 2020, has added four AI agents that split a weekly episode between them. Riverside says they "prep your questions, script and studio"; the Editor tightens and cleans the cut, the Clipper picks social clips and the Marketer writes captions and schedules posts. Per the launch press release, "nothing goes live without the creator's approval."
What it means for you: for a one-person weekly interview show, our estimate is three to four hours a week back, if the output needs only light edits. As of 2 Nov 2026, Riverside's pricing page lists Pro at $29 a month, or $24 billed annually; extras such as AI B-Roll still use credits.
How to apply it: run an old episode through the 14-day trial, read every caption before approving, and log your hours against your usual week.
Verdict: Try it if you already record in Riverside; skip if your footage lives elsewhere.
WHAT GOT FUNDED
Percent Labs (percent-labs.com) · Food and beverage · Syria. No founders are named publicly. It runs several delivery-only restaurant brands, from Nashville hot chicken to breakfast sandwiches, out of shared cloud kitchens. It says it handles more than 10,000 orders a month after 14 months and uses AI to manage orders. It closed an undisclosed seed round valuing it above $1 million, backed by Syrian angel investors abroad.
NanoForge AI (nanoforgeai.com) · AI for materials R&D · South Korea (announced 30 Oct). Led by CEO Kim Dong-hyun, it designs industrial materials with atomic-level AI and tests them in robotic labs. It says lab results feed back into its models, and it aims to cut materials development from more than ten years to about a month. It raised a ₩14.6 billion seed round led by Mirae Asset Venture Investment, with Futureplay, Mashup Ventures, DSC Investment, Korea Development Bank and EcoPro Partners.
AirMembrane (airmembrane.co.jp) · Semiconductor materials · Japan (announced 30 Oct). Led by president Masataka Hasegawa, it makes graphene pellicles that shield photomasks during EUV chipmaking. It says its AIST-born process grows graphene over 100 times faster than usual methods, with pilot production planned for 2027. It secured about ¥850 million in total, combining equity led by Global Brain (with Mitsubishi UFJ Capital, KSP, aSTART and others) with an SMBC loan and a NEDO subsidy.
Setara Biosciences (setarabiosciences.com) · Biotech · United States (announced 30 Oct). Co-founded by David Goldstein and David Breckenridge, it develops pills for severe rare genetic diseases, starting with the nerve disorder CMT2C. It says it targets the overactive protein behind a disease whatever a patient's mutation, then extends to commoner conditions. It raised a $90 million Series A from Deerfield, Canaan, Droia, EcoR1, Euclidean Capital and Regeneron Ventures.
Finfra (finfra.io) · Fintech · Indonesia (announced 30 Oct). Co-founded by Markus Prommik and Reinis Simanovskis, it sells lending software that handles loan origination, underwriting, servicing and collection in one system. It says lenders can operate in Indonesia through its API without building local teams, and apps can embed credit. It raised an undisclosed Series A extension from ADB Ventures' NextFin Asia fund, with Cento Ventures and Accion Ventures.
QUICK HITS
Wabi raised $55M at a $1.25B valuation, led by IVP, for Eugenia Kuyda's "personal agent" that builds mini apps. Investors are still backing consumer assistants, so expect more of them standing between you and your customers.
Anthropic's Usage and Cost Analytics API is in beta, breaking Claude spend down by model, workspace, API key and user. If you build on Claude, you can now see which customer or feature costs you most.
Deepgram is buying Daily, maker of Pipecat, a toolkit for building voice agents. Pipecat stays open source, so phone or voice assistants built on it keep working.
That's today. Hit reply and tell me what you're building. I read every email.
Dev
