ChatGPT's Shopping Revolution?

Nathan Weill here.

I'm the CEO @ Flow Digital, where we help companies unleash their full potential by strategically automating every inch of their workflows. Each week, I share hot AI and automation tips to help you move your business into the future successfully.

Speaking of future planning, we're offering limited 2026 Strategy Sessions this quarter.

ChatGPT Shopping Research

ChatGPT now offers a guided product researcher that creates custom buyer's guides instead of generic product links. Type a shopping query with multiple constraints—like "laptop under $1,000 for video editing, light, good battery"—and ChatGPT asks clarifying questions, then synthesizes Reddit threads, reviews, and retailer data into curated recommendations with trade-offs explained in plain language.

What It Does:

New AI discovery layer: Converts tens of millions of daily shopping questions into structured guides, creating an "AI shelf" where visibility depends on relevance and consumer discussion—not just search ads.

Higher-intent traffic: Multi-constraint research (typically several minutes per query) delivers clicks that behave more like trusted buying-guide traffic than casual browsing.

Early-mover advantage: Merchant integrations with OpenAI's checkout infrastructure could gain outsized visibility as this channel scales over 12–24 months.

What's Working vs. Current Limitations:

Wins:

• Excels at complex, multi-constraint purchases where users need synthesized comparisons fast…

• Surfaces non-obvious picks from niche communities that traditional search misses

• Higher product-fit accuracy and conversion rates compared to standard search flows

Watch-outs:

• Response times can hit several minutes for complex queries—slower than typical AI answers

• Coverage varies by category and region; some testers still prefer Google Shopping for certain searches

• Occasional off-target recommendations as the system continues learning merchant catalogs

• It doesn’t capture products currently listed on Amazon

ChatGPT Shopping = product research that sounds like asking a knowledgeable friend, now with a direct path to checkout.

Make

Scenario Run Replay: Test safely with real data

Ever had an automation fail and lose critical data? Or need to test a workflow update without risking your live processes?

Make's Scenario Run Replay feature lets you rewind and re-run any past automation using the exact same data from the original trigger. If a payment processor hiccup causes 100 orders to fail overnight, you can fix the workflow once and replay all 100 transactions without asking customers to resubmit.

Business Impact:

• Recover lost data instantly – Replay failed runs after fixing errors instead of manual cleanup

• Test safely with real data – Validate workflow changes using actual past transactions before going live

• Fill data gaps retroactively – Missed adding leads to your CRM last month? Replay those runs to backfill

Testing a new step in your lead routing? Replay last week's real leads to see how they'd flow through your updated process—no need to wait for new leads or risk breaking what's working.

Scenario Run Replay offers an automation safety net and time machine in one.

Work With AI-Powered Automation

Every automation platform you use has likely added AI features in the past year—which means you now have two types of automation to choose from:

• Deterministic (If This Happens, Do That): Rule-based, like a vending machine (press B4, get chips, every single time).

• Agentic (AI-powered): Is your personal assistant (figure out what I need and handle it).

Two Different Mindsets

Deterministic Automation follows exact rules. If a form is submitted, send this email. Same input, same output, every time.

AI-Powered Automation uses judgment. Summarize this ticket. Decide who should handle this request. The output can vary because the AI is interpreting, not just executing.

When to Choose Each
Whether you look at it as a spectrum or a definitive choice, here are some guidelines to get started:

Stick with If/Then when:

• The rules are clear and unlikely to change

• Consistency matters more than flexibility

• You need an audit trail with predictable outcomes

• The data is structured and uniform

Adopt AI-Powered when:

• Inputs are messy, inconsistent, or written in natural language

• The task requires interpretation, prioritization, or summarization

• You're spending hours on repetitive decisions that follow loose patterns

• Speed and scale matter more than pixel-perfect consistency

AI-powered automation can handle work that If/Then never could—but it asks more from you upfront. The teams that win aren't the ones who adopt fastest. They're the ones who adopt with the right expectations.

If you're evaluating where to adopt AI features next year, our 2026 strategy Sessions are a good place to start.

Self-Hosted AI Tools

Enhanced Security & Customization Control

Enterprises are shifting toward self-hosted AI models to ensure privacy and customization while reducing third-party API costs.

Think of it like the difference between renting software (Salesforce) versus hosting your own CRM. You're not building AI from scratch—you're taking models that already work and bringing them in-house.

• Data stays yours: Customer conversations, proprietary documents, and strategic plans never leave your control—critical when working with sensitive client data or meeting compliance requirements like GDPR.

• Predictable costs: Replace variable per-query charges (which can reach $10K+/month for heavy users) with fixed server costs, making budget planning straightforward.

• Customization control: Fine-tune models on your specific industry terminology, client base, or operational processes without waiting on a vendor's roadmap.

The shift mirrors what happened with cloud computing a decade ago—early adopters who invested in the right infrastructure gained lasting competitive advantages.

For enterprise businesses handling confidential information, self-hosted AI is becoming the default choice for 2025.

SMART WORDS OF THE WEEK:

— Leonardo da Vinci

The most powerful workflows aren't complex—they're elegantly simple. Strip away the unnecessary, and what remains will work beautifully.

Stay in touch with our team from anywhere.