8DIGITS news
News & Opinions.
Timely takes on technology and systems that matter for small and mid-sized businesses.
Google Ads Just Gave Every Small Business an AI PPC Analyst
Google Ads’ new AI Dashboards turn plain-English questions into performance reports and explanations, making deeper PPC analysis far more accessible to SMBs and small agencies.
Read →
ChatGPT Ads Are Coming for Google and Meta’s PPC Budgets
ChatGPT Ads are becoming easier for European SMBs to access, manage and measure, creating a new high-intent paid acquisition channel alongside Google and Meta.
Read →
n8n Assistant: describe, build, run and debug on the same canvas
n8n’s Assistant preview turns a plain-English outcome into a normal, editable workflow on your canvas, with execution logs and human gates on credentials and activation. Useful for EU agencies wiring HubSpot, Shopify and form stacks, if you still review the graph before it goes live.
Read →
DeepSeek V4.1 Flash is cheap enough to run real workflows. That is the point. It is not a drop-in Claude.
DeepSeek shipped V4.1 Flash on 10 September: a cheap multimodal model that undercuts Gemini Flash on token price and sits close to Claude, GPT and Grok on some agent benches. For SMBs the win is workflow volume, not a one-shot intelligence swap. Hosted DeepSeek is still the wrong place for client data.
Read →
ChatGPT Work Data Agent Makes SMB Life Just A Little Better
ChatGPT Work Data agent can investigate business data and build interactive dashboards from natural-language questions, potentially giving small businesses analytical capabilities that previously required dedicated BI teams.
Read →
Klaviyo “headless”: 400+ tools via MCP. Start with scopes, not the chatbot slide
Klaviyo’s K:BOS 2026 message: Klaviyo headless, is easy to misread as “we shipped a chatbot.” The Community write-up is clearer: more of Klaviyo’s data and supported actions can be called from Claude, ChatGPT or a custom agent through MCP and APIs (they cite 400+ tools alongside nearly 500 APIs).
Read →
Google Ads AI labels: the other half of Meta’s Article 50 story
EU AI Act Article 50 is not only a Meta Ads checkbox. Google Ads now has its own disclosure path for creatives that were created or edited with AI, and Spanish / EU advertisers need both platforms on the same checklist.
Read →
Google hotel search and flight search is remade for the EU. The rate left the first screen
If you run a hotel, restaurant or local travel site, EU Google Search for hotels, flights and restaurants now looks different. Google is splitting results into two units: one for OTAs and aggregators, one for direct suppliers.
Read →
Juspay’s Breeze Universal plugs Shopify and Woo into ChatGPT (pilot). Compliance is still your job
Juspay announced Breeze Universal around 2 September: a pilot so brands can be discoverable across AI surfaces, including ChatGPT and Claude, plus first-party merchant or bank experiences. Shoppers can invoke @breeze inside ChatGPT and get recommendations across participating brands without leaving the chat.
Read →
Claude commerce-agent blueprint is a demo. Klaviyo’s post is about the quarters after
Anthropic published an open reference implementation for building commerce agents on Claude: a shopping-agent starting point, sector examples (retail, travel and others), and an engineering guide. Klaviyo’s 8 September write-up is useful because it separates the repo from production.
Read →
Agentic Commerce Just Got Real, Meta’s AI Can Now Buy Things for You.
Meta’s new Muse AI agent can research products and complete purchases through Stripe. For ecommerce businesses, AI agents are starting to become actual customers.
Read →
Website Building Isn’t Finished After AI Builds It Anymore
AI coding tools made websites dramatically faster to build. Fimo is taking the next step, using autonomous agents to continuously maintain, optimize and improve them.
Read →
Visa and Revolut run a live AI-triggered card payment in France
Revolut and Visa completed a live AI-triggered card payment in France, authenticated with a Payment Passkey on existing card rails. It is still a pilot, but shops that take EU/UK cards should ask their PSP how agent orders will show up in settlement, refunds, and fraud.
Read →
ChatGPT Work learns your writing style from Gmail and Slack
ChatGPT Work can now learn how you write from Gmail, Slack, Drive, and SharePoint, then reuse that voice in new drafts. Connect the apps, skip the “write like me” prompt, and test it on one mailbox before you let it touch a shared inbox.
Read →
X starts paying creators for originality instead of engagement farming
X is replacing creator Revenue Sharing with Original Content Rewards, shifting payouts toward original expertise, analysis and creative work instead of pure engagement.
Read →
Half of web traffic is now bots, and marketers are paying for it
More than half of web requests now come from bots and AI agents, and marketers are seeing the effects in retargeting audiences, measurement and rising CPMs.
Read →
Shopify WebMCP and UCP actually change the way your store is AI agent readable.
Shopify WebMCP is now exposed in their liquid stores with tools that enable browser agents to search, add to cart and check out on the shopper’s live session, and storefronts advertise UCP version 2026-08-25 with no merchant config change. What that means for Spanish Shopify shops, and what to fix before you care about agents.
Read →
Google Shopping Ads now arrive directly inside AI answers
Google is putting Shopping ads directly inside AI Mode answers. For ecommerce SMBs, AI search is quickly turning from a discovery tool into a new storefront.
Read →
Enterprise MCP by Workato is making connecting AI agents to business software surprisingly easy.
MCP makes it easy for AI agents to use business software. Workato’s Enterprise MCP shows the next challenge: controlling what those agents are actually allowed to do.
Read →
AI agents compressed a two-week cyberattack into less than 10 hours
AI agents are making business workflows faster. Unfortunately, that applies to cyberattacks too.
Read →
Webflow MCP 2.1 lets AI agents build much more of your website
Webflow is giving AI agents considerably more control over websites with MCP 2.1, an upgrade announced at Webflow Conf 2026 that expands what tools like Claude, ChatGPT, Codex and Cursor can actually do inside Webflow. The biggest addition is native GSAP interaction support.
Read →
ChatGPT Is Slowly Becoming the Front End for Your Business Software
ChatGPT adds Zendesk and OneNote plugins, bringing AI deeper into everyday business workflows and closer to becoming the interface for your software.
Read →
Webflow’s New AI Agents Don’t Just Build Your Website. They Keep Working on It.
Webflow Source introduces AI agents that continuously monitor, build and optimize websites. Here’s what agent-driven websites mean for small businesses.
Read →
GPT-6 Astra is here, but the bill comparison with Claude Fable 5.1 depends on your stack
OpenAI has launched GPT-6 Astra, aimed at computer-use work: filling forms, updating CRM records, calendar tasks, browser QA and longer agent loops. It is rolling out to ChatGPT Plus, Pro, Business and Enterprise over the coming days, and to the API as gpt-6-astra. For most EU SMB and Spanish agency desks, the useful comparison is cost against Claude Fable 5.1, not the launch rhetoric.
Read →
Yoti is leaving Spanish app stores on 10 September. Here is what that means for age gates
From 10 September, Yoti will remove its Digital ID app from the Apple and Android stores in Spain. People who already have the app can keep using it. New downloads stop. Yoti says it wants to republish after its appeal is decided.
Read →
Claude Fable 5.1: agentic work got cheaper on paper. Art. 50 is the EU bit
Anthropic shipped Claude Fable 5.1 on 1 September. Base prices stayed at $10/$50 per million tokens; cache reads dropped 75% to $0.25. “Up to 45% cheaper” is Anthropic’s agentic scenario. Artificial Analysis still saw higher cost per task because Fable burns more output.
Read →
Meta brand-safety exclusion just became a discount while Ads MCP can edit budgets by default
If you built Sales or Leads brand safety by unchecking placements, that control is leaving the ad set. Meta has not published a hard kill date. The replacement is a value rule that can cut a bid by 90%, not a block.
Read →
Your Google Ads just joined AI Max without asking. Audit the defaults this week
Google’s Ads API team set the clock on 12 August. Through 30 September, Search campaigns still on campaign-level Broad Match or standalone Automatically Created Assets upgrade to AI Max in place. New legacy setups have been blocked since 3 August. Defaults differ by what you had, and extra copy or query matching can turn on without a meeting.
Read →
AI Agents Cost Per Task: the Real Story of Claude Fable 5.1 and Gemini 3.7 Flash
For SMBs, the real AI breakthrough is cheaper repeated agent runs. Claude Fable 5.1 and Gemini 3.7 Flash change cost per completed task.
Read →
AI can send your Klaviyo campaigns, so turn Send off until you mean it
Klaviyo’s remote MCP server that lets Claude, ChatGPT, or Cursor talk to your account now has tools that do more than draft. It now includes send_campaign (“Send a campaign to its configured audience”) and cancel_campaign_send.
Read →
ChatGPT Ads self-serve is live in Europe (Spain included)
OpenAI opened self-serve ChatGPT Ads Manager on 31 August to eligible advertisers across Europe, India, the Middle East and North Africa. Spain is listed as Available. Sign-up is at ads.openai.com. The same Monday, Brussels designated ChatGPT a Very Large Online Search Engine under the DSA — the first standalone AI chatbot treated as search at that scale. The inventory you can buy this morning now sits on a surface the Commission has labelled search.
Read →
WebMCP Might Become the AI Interface for Your Website
WebMCP is an emerging web standard that lets websites expose structured tools that AI agents can use directly.
Read →
Claude can now work inside Salesforce if you already live there
Salesforce and Anthropic announced Claudeforce on 27 August: Claude’s reasoning wired into Salesforce data, workflows, business rules, and permissions. The first product is Salesforce in Claude, a plugin with 37 prebuilt sales skills.
Read →
Google now expands AI Overviews before you click Show more
Google confirmed on 28 August that AI Overviews can now open to full length on some queries, without a tap on “Show more.”
Read →
OpenAI will cut Cursor off its models after the SpaceX deal
OpenAI told SpaceX on 28 August it will wind down the contract that feeds OpenAI models into Cursor. Proposed shutoff is 12 November 2026 the longest notice the change-of-control clause allows.
Read →
Gemini Omni now builds Google Ads video in Asset Studio
Google put Gemini Omni inside Asset Studio this week, so a Demand Gen, Performance Max or YouTube campaign can start from a brief and existing assets instead of a video shoot.
Read →
Don’t label every AI ad, only the ones that can mislead
IAB published Disclosure Framework V2 on August 18: label high-risk AI that can fool a customer, not every caption or colour grade. On a small Meta or Google account that is the difference between keeping your clicks and giving up about 31% of them, or skipping a badge the EU now expects.
Read →
Gemini Live now runs your inbox and Workspace by voice
Google upgraded Gemini Live on August 26 so a voice session can move work, not just answer questions. You can triage Gmail hands-free — search, summarise, star, archive, or delete — and hand multi-step jobs to Spark, which keeps running across Docs, Sheets, Drive, and the web after you hang up. A spoken Daily Brief pulls the day’s priorities from Gmail and Calendar. Personal Intelligence lets Live remember earlier chats and connected apps instead of starting from zero every time.
Read →
ChatGPT Work gains webhook-triggered tasks
OpenAI quietly upgraded ChatGPT Work on August 25 with two practical capabilities that move it closer to a real always-on operator for small teams and solo founders. Scheduled tasks can now be triggered by webhooks from Gmail, Slack, and GitHub instead of only running on a fixed timer. At the same time, the browser agent
Read →
Perplexity’s Portable Computer: local-first AI agent
Perplexity launched Portable Computer on August 25, a fully local version of its agentic “Computer” platform that runs the model, planner, tools, and sandbox on the user’s own hardware. At launch it supports Nvidia DGX Spark and Linux machines with compatible RTX GPUs (24 GB VRAM minimum). Windows support is promised for September. The system
Read →
Claude’s New Memory Is a Bigger Deal for Small Businesses Than It Looks
Anthropic has expanded Claude’s memory so that context can now follow users between regular Claude conversations and Claude Cowork, which until were completely separate. That sounds like a minor convenience at first but for anyone constantly switching between ideation and getting it done, like small businesses, it could be much more important. The biggest problem
Read →
OpenAI drops GPT-5.6 Sol API pricing
The ChatGPT making giant just dropped it’s flagship model’s price by 20% on input and ~33% on output API token usage and on eligible ChatGPT Work/Codex credits. The subscriptioons however remain unchanged. This also meant that almost all API access providers droppped their GPT 5.6 Sol pricing significantly and while OpenAI’s promo lasts until at
Read →