Triple Threat: The AI That Works Only From Your Files · Klarna’s Costly Lesson in Speed vs Trust · A GPT That Turns Competitor Intel Into Instant Battlecards

💥 Power: A tool that unlocks a new superpower for your business

Google NotebookLM: The AI That Works Only From Your Files

What it is: Google NotebookLM is Google’s AI research and briefing assistant that works only with the materials you provide—PDFs, Google Docs, YouTube videos, audio files, and more. Every answer links back to the original source so you can verify facts and keep messaging on point.

Instead of pulling from the open internet, it’s built on your sources—so outputs are accurate, relevant, and aligned with your strategy. You can upload up to 50 files in the free tier (300 with Plus), from PDFs to transcripts to videos, and work collaboratively with your team in a shared notebook.

Why it’s worth trying: NotebookLM streamlines the shift from raw information to actionable output. In one workspace, you can produce summaries, FAQs, timelines, mind maps, narrated videos, and podcasts. It’s as useful for sales as it is for market updates, and it lets you visualize how topics connect across documents, helping you spot patterns you might miss in a manual review. Not to mention it saves you a ton of time. I can honestly say it gives me back at least 10 hours in my week, if not more.

How I’m Using It:

• Competitive Briefs for Sales – I upload competitor sites, pricing pages, and product docs, then generate a differentiator chart with objection handlers.

• Market Pulse for Leadership – At the end of the week, I drop in articles, industry newsletters, and podcast transcripts for a concise, cited update.

• Campaign Messaging Alignment – I combine creative briefs, research reports, and past campaign results into a unified, fact-checked messaging playbook.

Amazing Use Cases for Leaders:

1. Customer Insight Digest – Turn sales calls, support logs, and NPS surveys into a report highlighting recurring objections and triggers—linked to verbatim quotes.

2. Market Expansion Brief – Consolidate regional studies, competitor profiles, and partner intel into a single, decision-ready plan.

3. Event Briefing Pack – Upload agendas, attendee lists, and press coverage to create a quick-reference prep doc for your leadership team.

4. Training Refreshers – Blend old and new playbooks into a “what’s changed” narrated update for the sales org.

5. Regulatory/Compliance Snapshot – Summarize updated industry regulations into an executive-friendly brief with key takeaways.

How it stacks up:

• vs. Generic AI tools → Only uses your sources, so it stays on-message and fact-checked

• vs. Manual prep → Turns one set of documents into multiple, ready-to-use formats in minutes

• vs. Multiple point tools → Consolidates research, summarization, visualization, and repurposing into one workspace

How to use it (high level):

1. Upload your approved source materials (docs, reports, transcripts, videos)

2. Ask NotebookLM for the format you need: briefing, comparison, playbook, or update

3. Share or repurpose the results across sales, marketing, and leadership teams

First step: Try Google NotebookLM and run one current project through it: see how much faster you get to a usable deliverable.

Want to see it in action?
Here’s a full tutorial showing all of NotebookLM’s features in under 15 minutes, covering setup, file uploads, multi-format outputs, and real-world use cases.

🎯Precision: A real-world application and what it means for growth, trust, and traction

Fast Answers, Broken Trust: Klarna’s AI Customer Service Pivot

Klarna went all-in on AI support in late 2024. The plan: have a generative AI assistant handle most customer inquiries, fast responses, lower costs, fewer manual tickets.

At first, it looked like a win. Routine questions got answered instantly. Handle times dropped. The CFO was happy with all of the cost savings.

But then came the edge cases. Fraud disputes. Chargebacks. Identity verification. The AI started making bad calls on high-stakes issues—closing legitimate fraud cases, mishandling customer data, and sending inaccurate decisions to partner banks.

What happened next:

• Customers escalated complaints after “resolved” cases

• Regulators started asking questions about automated decisions

• Partner banks flagged clusters of errors

• Klarna had to reintroduce human agents and tighten controls

Why it failed:

• No clear “do not automate” list for high-risk cases

• Confidence thresholds not enforced—AI acted even when unsure

• Lack of human oversight for Tier 2/3 support

• Rushed rollout without shadow testing complex workflows

What it teaches: Automating the easy stuff is fine—but once AI touches regulated or high-value transactions, the brand carries the risk. Precision isn’t just accuracy on average—it’s about not failing where it matters most.

How to apply it:

• Train AI on real cases, especially the messy ones

• Define red-line categories for human-only review (fraud, legal, compliance)

• Use confidence gating so the AI only acts above a set threshold

• Run AI in shadow mode before granting autonomy in high-stakes workflows

• Make public your escalation and human-override process

Bottom line: When AI decisions can cost you customers, money, or regulatory trust—it’s not about speed, it’s about governance. Automate with precision, not just ambition.

🚀 Momentum — A custom GPT I create to help you take action today

Competitor Battlecard Agent: Turn competitor moves into sales-ready battlecards

Most teams track competitors in scattered docs and DMs, and they miss the updates that matter. This GPT runs on demand and turns your latest checks across news, product pages, social posts, hiring, and reviews into a single, living battlecard your reps can use on calls.

Who it’s for: CEOs, CROs, Heads of Sales and Marketing, Product Marketing, RevOps, and Operations leaders who want current, usable competitive intel without manual research.

What it does: You add your company, domain, and optionally a competitor list. Each time you run it, the GPT reviews press, product pages, LinkedIn and X, hiring signals, and review sites, then generates markdown battlecards with price, features, ICP, value props, recent moves, your edge, strengths and weaknesses, and objection handlers. It also calls out what’s new versus your last saved card so you can keep a clean history.

Why it’s worth using:

• Turns noisy updates into clear talk tracks your team can use today

• Flags pricing and packaging shifts, partnerships, and customer wins quickly

• Keeps one source of truth so decks and messaging stay fresh

• Reduces prep time for reps and PMM while improving call readiness

👉 Try Competitor Battlecard Agent GPT

_{*Note: You will need a ChatGPT account (ideally a Plus or Team plan) to run this GPT.}