Subscribe | Free AI | CRO Audit | Cool Tools

New AI Law

Last week, a new law was passed called the Colorado Automated Decision-Making Technology (ADM) Act (formerly the Colorado Artificial Intelligence Act) Bill SB26-198. This law is set to take effect on January 1, 2027. Pay attention if you are in these industries: employment, housing, education, healthcare, insurance, credit, AI development, and legal services. Here is a summary:

Who and What is Covered

• Consequential Decisions: Covers AI tools used to materially influence or decide access to vital necessities, such as employment, housing, education, healthcare, insurance, credit, and legal services.

• Developers: Creators of covered ADMT must provide deployers with comprehensive technical documentation (intended uses, training data categories, and known limitations) and notify them of material updates.

• Deployers: The organizations using the AI to make decisions. They are expected to conduct human reviews of outcomes and establish methods to contest adverse decisions.

Consumer Rights & Protections

• Point-of-Interaction Notice: Deployers must provide clear and conspicuous notice to consumers that an automated decision-making tool is being used before or at the time of interaction.

• Right to an Explanation: If an AI materially influences a "consequential decision" that negatively impacts a consumer (such as a denial of a job or a higher price), the deployer must provide a plain-language explanation of why the adverse action occurred.

• Right to Appeal: Consumers have the legal right to appeal decisions made by covered AI systems and request that a human review the choice. 

Enforcement & Compliance

• Exclusively Enforceable: The Attorney General enforces the legislation under the Colorado Consumer Protection Act.

• Right to Cure: Before taking action, the state provides a 60-day opportunity to correct violations, unless the violation is proven to be knowingly or repeatedly committed.

• Liability: Both developers and deployers can be held liable, but only to the extent of their relative fault (e.g., if a deployer uses a tool exactly as intended and documented, liability falls on the developer). 

This week’s sponsor is Ometrics AI

News This Week

AI

Google Gemini 3.5 Flash Launches — Opus 4.7-Level Intelligence at 4x Speed and Half the Cost. It is now powering Google Search as the new default for AI Mode.
Google's Gemini 3.5 Flash is the model powering most of Google I/O's agent announcements, and its benchmark position is striking: Every's on-the-ground review from Google I/O places it in a unique quadrant of speed and intelligence — fast enough to make AI Mode feel like traditional search and smart enough to handle complex agentic tasks. TLDR AI confirmed it's specifically designed for agentic workflows, coding, and long-horizon task execution with a 256K context window, now available via the Gemini API. (Link) (Link)(Link)

America Is Turning Against AI — Every Major Poll Now Shows Declining Public Trust
The Algorithmic Bridge compiled the most comprehensive polling analysis on AI public opinion yet, covering every major US survey from Pew, Gallup, AP-NORC, Ipsos, and others. The finding is consistent across all methodologies: American trust in AI has declined significantly since 2023, with concerns centering on job displacement, accuracy, and lack of accountability. For marketers, it's a crucial signal — the gap between executive AI enthusiasm and consumer AI skepticism that Stacked Marketer documented last month appears to be structural, not cyclical. (Link)

US agentic commerce revenue forecast to reach $1 trillion by 2030
But even as the technology grows in popularity, the physical shopping experience will remain important, per a new report. Retail in the U.S. is experiencing dramatic changes in the era of AI. Brick-and-mortar stores have the opportunity to survive and thrive if they know how to differentiate the in-store experience and substantially invest in an AI strategy (Link)

Digital Marketing

Google I/O SEO Verdict: AI Mode Is Going Default — Keyword Rankings Are No Longer the Primary Metric
Product-Led SEO's post-I/O analysis is the most important SEO read of the year: Google did not announce AI Mode as the default, but its rollout trajectory makes the outcome inevitable — and the implications are severe for any SEO strategy still built around keyword rank tracking. A business can hold strong positions for high-volume keywords and see zero traffic because the AI routes users based on contextual fit, not keyword match. The new metrics that matter: revenue from search, qualified pipeline from search, and LTV from search — not clicks, not impressions, not rank positions. (Link)

37% of Users Now Start Search Journeys Directly in AI Tools — Bypassing Google Entirely
37% of users now begin their search journey directly in AI tools like ChatGPT, Gemini, or Copilot, compared to 47% who still start in Google. For nearly 4 in 10 users, the search journey starts and ends inside a chatbot interface without ever touching a brand website. The actionable response: use schema markup, structured FAQ content, and clear direct answers to common questions so AI tools have accurate content to cite as the authoritative first response. (Link)

Amazon Brings Alexa AI Directly to Its Search Bar — Replacing Rufus for Millions of Shoppers
Amazon deployed Alexa AI capabilities directly into its main search bar, supplanting the Rufus shopping assistant as the primary AI touchpoint for product discovery. Retail Dive reports this is Amazon's most significant search UX change in years — shoppers can now ask natural-language purchase questions directly in the search box and get AI-synthesized answers rather than a results list. For brands selling on Amazon, it accelerates the shift from keyword optimization to product data accuracy as the primary driver of visibility. (Link) (Link)

How to Build an AI Content Engine With Claude Code That Doesn't Hallucinate Your Editorial Reputation
Ahrefs' Ryan Law published a detailed system for content engineering with Claude Code — a 23-skill-file architecture that chains editorial tasks from keyword research to near-finished draft, with every stage saving its own output file for troubleshooting. Key principles: connect Claude to real data via the Ahrefs MCP to eliminate hallucinated SEO metrics, front-load human editorial direction before writing begins, and build recursive self-testing loops that keep skill files lean and output quality high over time. (Link)

Conversion Optimization

The "Checkout Gap" Is Costing Retail and Hospitality Brands Loyalty and Revenue — New Research
A new study finds that friction at checkout — slow payment processing, forced account creation, limited payment methods, poor mobile UX — is not just a conversion problem but a loyalty problem: customers who experience checkout friction are measurably less likely to return for a second purchase, even if they completed the first. The "checkout gap" between what brands believe their checkout experience delivers and what customers actually experience averages 23 percentage points — and brands that close it see compounding loyalty effects, not just one-time conversion lifts. (Link)

Cool Tools

OpenClaw is morphing into many other Claws depending on your use case. The alternatives were created due to security, setup friction, less standared features and its size.

• OpenClaw - The orignal, For security and not configuration headaches you can install it on a server Hostinger 

• NanoClaw - Runs in a container for security, with small code base

• IronClaw - security, open-source

• TrustClaw - many tools and very secure

• ZeroClaw - Runs on 5 MB of RAM

• Hermes - Good for reptitive tasks that need thier own muscle memory