Zapier flags AI builders, Google splits data: July 7 dispatch

AI is not just writing copy today. It is building small business software, changing account privacy, and filtering the places your customers research you. In 35 minutes, you can audit who is building tools, lock down Google data settings, and tighten where Reddit research enters your marketing stack.

Zapier says 34% of AI builders lack formal coding training

Zapier published survey data on July 6, 2026 from about 800 US employees who have built or deployed tools with AI coding apps. The headline number is useful for owners: 34% of people shipping software with AI tools have no formal programming background.

That changes the risk inside a normal business. AI coding apps are tools that turn plain-language requests into software. They let a non-engineer build a dashboard, form, or customer-facing calculator, but they also make it easier for shadow tools to touch private data before anyone has approved them. CRM means the software that stores your customers and sends the follow-ups. If an employee connects an AI-built form to your CRM with a personal login, you inherit the risk.

If you are already comparing automation platforms, pair this audit with Zapier alternatives for automation and Zapier pricing at volume. The point is not to stop small tools. It is to stop mystery tools from becoming customer infrastructure.

Google separates Search data from Web & App Activity

TechCrunch reported on July 6, 2026 that Google has split some Search data controls away from the older Web & App Activity setting. Google's own help page now says it is updating settings for Google Search services, including Search, Maps, Shopping, Flights, Hotels, Translate, and News.

This matters if your team uses Google accounts for customer calls, uploaded files, local search work, or ad research. Web & App Activity is the account setting that stores activity across Google sites and apps. TechCrunch says the new Search data setting is on by default, and the article notes that Google may store media such as images, files, audio, and video recordings to improve AI models.

The wiring move is simple. Open myactivity.google.com, choose Controls, then check both Web & App Activity and the separate Search data control. For shared business accounts, turn off anything that stores customer files, calls, or internal research unless you have a written reason to keep it on. Add this to the same privacy checklist you use for email deliverability, meaning whether your emails actually reach inboxes.

Reddit uses LLMs to cut spam exposure

Reddit said on July 6, 2026 that its updated automated systems are blocking 23 million spam views per day, catching about 25,000 new spam posts and comments a day, and reducing user exposure to spam by about 20% from January to March 2026 compared with the prior three months. LLM means large language model, the AI system that reads and writes text.

The business consequence is not moderation theory. It is market research quality. If you use Reddit threads to understand complaints, product language, competitor anger, or buying questions, the input is being filtered by AI before you ever see it. That is mostly good. It also means you should stop treating one viral thread as proof of demand.

Wire Reddit research into your stack with a source rule. Save the thread URL, subreddit, date, top three repeated phrases, and whether the post looks like a real customer question or artificial hype. Use it as a signal, not the whole brief. For broader search visibility work, connect this to AI recommendations discovery and AI Overviews traffic planning.

Zapier updates the database app-builder shortlist

Zapier updated its database-powered app builder guide on July 7, 2026 with seven tools: Airtable, Progress Podio, Google AppSheet, Zenkit, Smartsheet, Memento Database, and Baserow. A database is a structured home for business records. Think customer requests, jobs, inventory, quotes, and service history.

The useful change is the pricing and AI fit. The guide lists Airtable paid plans from $24 per user per month, Google AppSheet deployment from $5 per user per month, and Baserow paid plans from $12 per user per month. It also notes Airtable's AI upgrades and AppSheet's Gemini and Gmail connections.

The wiring move: stop running repeated work from a spreadsheet once three people edit it or customers depend on it. Pick one process, such as quote requests or job status, and map the fields before choosing a tool. Fields means the columns you need: name, phone, job type, status, owner, next action, and last update. Then connect the chosen app to the automation layer you priced in Make vs Zapier or the lead-to-CRM Make build.

On the bench

Ternlight, a 7 MB browser embedding model, is worth testing for private search inside small help docs once it has clearer packaging.

OfficeCLI, an open-source tool for AI agents to read and edit Microsoft Office files, belongs on the watch list for proposal and report workflows.

Zapier's ambient AI explainer is worth reading, but do not buy around the phrase yet. Price the actual automation job in /topics/automation first.