42% of Companies Are Abandoning Their AI Projects, Why?
AI just got real for business owners: it’s now deciding health insurance claims at scale, screening 40% of job applications out before human review, and fundamentally changing how the next generation of workers learns. Today’s edition breaks down the new federal Genesis Mission accelerating AI breakthroughs, shows why most companies aren’t seeing ROI from AI yet (and what the successful ones do differently), and covers the tools helping businesses adapt.
Word of the Day: AI VALUE GAP
Definition: The widening disconnect between what companies spend on AI and what they actually get back in measurable business results.
Why it matters: According to AWS and S&P Global, 42% of companies abandoned most of their AI initiatives in the first half of 2025, up from just 17% last year. Meanwhile, 92% of companies are increasing AI spending. The gap isn’t about technology limitations; it’s about implementation approach.
The numbers:
Most companies: Running scattered AI experiments with no clear ROI tracking
Successful companies: 45% more cost savings, 60% more revenue growth (Boston Consulting Group)
The difference: Synchronized transformation across people, process, and technology, not just deploying tools
What successful companies do differently:
Leadership: Business leaders (not just IT) drive the AI agenda from day one
Incentives: Reward actual AI adoption with measurable outcomes, not theoretical interest
Governance: Balance innovation speed with compliance (centralized guardrails, federated execution)
Metrics: Track 1-2 clear business outcomes, not just “AI costs”
What this means for you: If you’re experimenting with AI tools but haven’t seen meaningful P&L impact, you’re not alone. The fix isn’t better AI, it’s better implementation strategy.
Action step: Pick your current AI tool or experiment. Write down: (1) The specific business metric you’re trying to move, (2) The baseline number before AI, (3) How you’ll measure the after. If you can’t answer all three, you’re likely contributing to the value gap.
Skill of the Day: How to Use AI for Resume Screening Without Getting Sued
The challenge: AI is now screening 40% of job applications before humans see them, but multiple lawsuits allege AI hiring systems discriminate based on race, age, or disabilities. Meanwhile, 41% of job seekers admit to using “prompt injections” (hidden text) to bypass AI filters, creating an arms race where trust is at an all-time low.
What’s actually happening:
Application volume up 45% on LinkedIn due to AI-assisted applications
Trust is down across all parties: only 8% of job seekers believe AI makes hiring fairer
Recruiters uncertain: 25% admit they’re not confident in their AI systems, 8% say they “have no idea” what their algorithms prioritize
Candidates fighting back: 54% of candidates who don’t use prompt injections are considering it
The risk: Workday is in a lawsuit over alleged discrimination. The EU AI Act classifies recruitment AI as “high-risk.” California Consumer Privacy Act gives candidates the right to know how their data is used.
How to do it right:
Step 1: Understand what “AI-assisted” legally means
Not illegal: Using AI to rank candidates by skills match, schedule interviews, or draft job descriptions
High-risk: Using AI to make final hiring decisions without human review of individual circumstances
Requires transparency: California Consumer Privacy Act gives candidates right to know how AI was used in their application process
Step 2: Build in human oversight
Minimum: Every AI-flagged rejection requires human review of the candidate’s individual circumstances
Best practice: AI suggests and ranks, humans make final decisions—with clear audit trails
Test regularly: Run bias audits (gender, age, race) on AI recommendations to catch systematic discrimination
Step 3: Set clear guidelines and communicate them
Transparency: Tell candidates in job postings when and how AI is used in your screening process
Appeals process: Provide human contact for candidates to question or dispute AI decisions
Documentation: Keep records showing a human reviewed individual circumstances before any rejection
Step 4: Focus AI on the right parts of hiring
Good use cases: Initial skills matching, interview scheduling, administrative tasks like email responses
Bad use cases: Final hiring decisions, personality assessments without validation, anything creating legal liability without human judgment of individual context
The key distinction: Use AI to help humans make better decisions faster, not to replace human decision-making entirely
Real-world example: BCG reports 92% of companies using AI in hiring see benefits, with 10%+ reporting 30% productivity gains. The successful ones use AI for “marketing and administrative tasks that are a core part of the hiring process” while keeping humans central to relationship-building and final decisions. They’re freeing recruiters to spend more time with qualified candidates, not replacing recruiter judgment.
Action step: If you use AI in hiring or are considering it, answer these three questions: (1) Can you explain exactly what the AI prioritizes when screening candidates? (2) Does a human review individual circumstances before rejecting any candidate? (3) Do your job postings disclose that AI is used in screening? If you answered “no” to any of these, you’re in the regulatory risk zone and should address these gaps before they become legal problems.
Tools and Tips
1. Automat - Turn Screen Recordings Into AI Automations
What it does: Record yourself doing a repetitive work task (or upload your process docs), and Automat’s AI builds and maintains the automation for you. Deploy through API, Chrome extension, or desktop app.
Why it matters: Most business owners know they should automate repetitive tasks but don’t have time to learn automation tools. Automat bridges that gap. Just show it what to do, and it handles the technical implementation.
Business use cases:
Data entry across multiple systems
Invoice processing and document extraction
Customer onboarding workflows
Report generation from multiple sources
Link: https://www.runautomat.com/
2. Claude Opus 4.5 - The World’s Best Coding AI Model
What it does: Anthropic’s new flagship AI model scored 80.9% on solving real-world software issues from GitHub (first to break 80%). Built specifically for coding, complex reasoning, and orchestrating teams of smaller AI agents.
Why it matters: If you have developers on staff or work with development agencies, this is the AI they should be using. It’s 3x cheaper than previous Opus models while being significantly more capable, meaning your development costs can actually go down while quality improves.
Business applications:
Accelerate custom software development
Reduce development costs (66% price reduction from Opus 4.1)
Debug and fix code issues faster
Build AI agent workflows for your business
Link: https://www.anthropic.com/news/claude-opus-4-5
3. ChatGPT Shopping Research - AI-Powered Product Discovery & Comparison
What it does: Describe what you’re shopping for, and ChatGPT builds a personalized buyer’s guide by asking clarifying questions, researching across the web, and comparing products with up-to-date pricing and specs. Takes a few minutes, replaces hours of research.
Why it matters: Whether you’re sourcing office equipment, comparing software vendors, or finding the right tools for your team, this handles the tedious research part. Works especially well for electronics, office supplies, and business equipment.
Business use cases:
Compare business software and tools
Source office equipment and supplies
Research vendor options with detailed comparisons
Make informed purchasing decisions faster
Available to: All ChatGPT users (Free, Plus, and Pro plans) with nearly unlimited usage through the holidays.
Link: https://openai.com/index/chatgpt-shopping-research
4. RightBlogger - Automated Blog Posts & Content Creation
What it does: Full suite of 80+ AI tools for bloggers and content creators. Write SEO-optimized articles in minutes, turn videos into blog posts, auto-optimize for search engines, and publish directly to WordPress with one click.
Why it matters: If you’re trying to maintain a blog, newsletter, or content marketing strategy while running a business, RightBlogger dramatically reduces the time investment. Their “MyTone” feature learns your writing style so content actually sounds like you.
Business use cases:
Maintain company blog without hiring content team
Convert webinars/videos into written content
SEO optimization without technical expertise
Generate leads with custom AI tools on your site
Pricing: Free trial available, paid plans from $29.99/month for unlimited usage.
Link: https://rightblogger.com/
5. Excelmatic - AI-Powered Excel Analysis & Visualization
What it does: Upload your Excel file, ask questions in plain English, and get instant insights, charts, and analysis. Automatically identifies data types, suggests visualizations, generates formulas, and spots anomalies.
Why it matters: Most business owners have critical data in Excel spreadsheets but lack time or expertise to analyze it properly. Excelmatic turns “I know this data matters but I don’t know what it’s telling me” into actionable insights in minutes.
Business use cases:
Analyze sales data and spot trends
Create client-ready reports and charts
Understand financial metrics without accounting degree
Clean messy data and catch errors
Pricing: Free tier (10 queries/month), paid plans start at $9/month.
Link: https://excelmatic.ai/
6. TubeDummies - Turn YouTube Tutorials Into Step-by-Step Interactive Learning
What it does: Converts any YouTube tutorial into an interactive, step-by-step guide you can follow at your own pace with the ability to ask questions and get clarification.
Why it matters: I watch YouTube videos almost daily to learn how to do something new. It’s cumbersome to have to continually pause the video or rewind it while going through the steps. TubeDummies makes it easy to review the steps at your own pace.
Business use cases:
Employee training: Turn vendor webinars or training videos into interactive courses
Customer onboarding: Convert your product demo videos into self-service learning paths
Skills development: Make YouTube business tutorials actually stick with interactive practice
Link: https://tubedummies.com
7. SafeNew - AI Text Humanizer with Personal Writing Model
What it does: Transforms AI-generated content into natural, human-like text by building a personalized rewriting model trained on your specific writing patterns using vocabulary, rhythm, and tone. Goes beyond word-swapping to reorganize ideas and adapt style to your identity.
Why it matters: If you’re using AI to draft content but concerned it sounds robotic or generic, SafeNew makes it sound like you actually wrote it. Particularly valuable for customer-facing content, marketing copy, or anything where your voice matters.
Business use cases:
Polish AI-drafted emails and proposals
Make marketing copy sound authentic
Maintain consistent brand voice across AI-assisted content
Avoid AI detection while maintaining quality
Pricing: Free trial available with credits to test.
Link: https://safenew.ai/
IN THE NEWS
1. PBS: College Students Using AI for Nearly Everything
What happened: 86% of college students now use AI tools like ChatGPT for schoolwork, and this year’s graduating class is the first to have spent nearly their entire degree in the age of generative AI.
Why business owners should care: These are your future employees. They’re being trained to use AI as a fundamental tool, not an optional add-on. Ohio State now requires all undergrads to learn AI tools across all disciplines.
The controversy: Some professors see 50%+ of writing assignments showing AI detection flags, leading to “teaching feeling like policing.” Others are embracing it. One music professor uses AI to analyze thousands of tuba recordings to help students find the perfect note.
The shift: Universities are moving from “ban AI” to “teach AI fluency.” Professor Ravi Bellamkonda: “If you don’t use AI to be more effective, you’re not going to be competitive in the job market.”
Action for you: Ask your newest hires how they use AI. You’ll likely find they’re already 2-3 years ahead of your current workflows and can teach you.
2. Trump Launches ‘Genesis Mission’ - Federal AI Supercomputing Initiative
What happened: President Trump signed an executive order launching the Genesis Mission, mobilizing 17 national laboratories, federal supercomputers, and decades of scientific data into a unified AI platform. Called “comparable in urgency to the Manhattan Project.”
What it means: The Department of Energy will create an “American Science and Security Platform” providing:
High-performance computing resources across national labs
AI agents to automate research workflows and test hypotheses
Access to federal scientific datasets (world’s largest collection)
Goals: Compress scientific discovery from years to days or hours
Business implications:
Nvidia and Anthropic already announced partnerships
Private sector can collaborate on challenges around energy, materials science, medicine
Accelerated breakthroughs in energy (grid optimization), biotech, quantum computing
Signals federal government going all-in on AI leadership
Timeline: Initial operating capability within 270 days, initiative runs through 2028 with yearly updates.
Why it matters: This level of federal coordination and data access could dramatically accelerate AI capabilities across industries. If you work in energy, materials, healthcare, or advanced manufacturing, partnerships may be coming.
Link: https://www.whitehouse.gov/presidential-actions/2025/11/launching-the-genesis-mission/
3. MIT: How AI Can Help Achieve Clean Energy Future
What happened: MIT researchers detailed how AI is already helping the energy transition from optimizing wind/solar installations to predicting grid equipment failures before they cause blackouts.
Real-world applications already working:
Grid operations: AI forecasts which power plants should run hour-by-hour while maintaining grid stability
Predictive maintenance: Spots equipment issues before failure, preventing blackouts and reducing costs
Materials discovery: AI-guided experiments shortening development from decades to a few years
Demand management: Smart thermostats and EV charging that responds to real-time grid pricing
Business owner takeaway: If you’re in construction, facilities management, or have high energy costs, AI-powered energy management is no longer theoretical. Buildings, transportation, industrial processes already seeing consumption reductions.
The infrastructure challenge: Yes, AI data centers use a lot of power (1.5% of global electricity now, doubling by 2030). But properly deployed AI tools can actually optimize overall energy use, evening out the net impact.
Link: https://news.mit.edu/2025/how-ai-can-help-achieve-clean-energy-future-1124
4. Health Insurance AI Creating “Arms Race” Between Patients and Insurers
What happened: Insurers increasingly use AI to process claims, with denial rates hitting nearly 1 in 5 (up from 17% in 2021). In response, patients are now using AI to fight back—creating appeal letters, catching billing errors, and forcing insurers to defend denials.
The numbers:
73 million Americans had in-network claims denied in 2023
Less than 1% appealed (too time-consuming)
Over 60% of physicians fear AI is increasing prior authorization denials
California’s response: SB 1120 (effective Jan 1, 2025) requires human physician review for AI-based medical necessity decisions
Tools emerging:
Counterforce Health (NC nonprofit): Free AI assistant to draft customized appeal letters
Sheer Health: Identifies coding errors, handles insurer communications
Generic LLMs: Patients using ChatGPT/Claude to draft appeals
Business impact: If you provide employee health benefits, know that:
Your employees may face more denials
AI tools are making appeals easier (success rates are actually high when people appeal)
Expect regulatory pressure on AI-based healthcare decisions to increase
Quote: “Insurers profit from claims denials. We should be very suspicious when they adopt technologies that make it easier to deny claims.” - Jennifer Oliva, Indiana University law professor
6. Hiring “AI Doom Loop”: Trust at All-Time Low for Job Seekers and Recruiters
What happened: New Greenhouse survey of 4,100+ people shows hiring is in an “AI arms race” where neither side is happy. Job seekers use AI to apply to more jobs, employers use AI to filter them back out. The collective result is worse for everyone.
The numbers:
41% of job seekers admit using “prompt injections” (hidden text to bypass AI filters)
52% of those who don’t use this tactic are considering it
49% submitting more applications than last year (many AI-assisted)
Only 8% of candidates believe AI makes hiring fairer
62% of Gen Z entry-level workers have lost trust in hiring
Recruiter side:
70% of hiring managers say AI helps make faster, better decisions with fewer resources
Only 21% of recruiters are “very confident” their AI systems aren’t rejecting qualified candidates
25% admit they’re “not confident in their AI systems at all”
8% say they “have no idea” what their algorithms prioritize
Business implications: If you’re hiring:
Expect more applications (quantity), but question quality
Your AI screening tools may be creating legal risk (see Skill of the Day)
Consider how you’re being perceived. Over half of candidates have had AI-led interviews, which many find “downright insulting and inhumane”
Quote: “Trust is at an all-time low for both job seekers and recruiters.” - Daniel Chait, CEO, Greenhouse
Link: Greenhouse
7. Tennessee Releases First State AI Action Plan
What happened: Tennessee AI Advisory Council released the state’s first comprehensive AI action plan, organized around four pillars: pilot programs, infrastructure, workforce development, and governance.
What’s in it:
Pilots: Demonstrate real improvements in government services
Infrastructure: Secure systems for AI experimentation while protecting sensitive data
Workforce: Boost AI literacy among public employees, expand training through education partnerships
Governance: Risk management, oversight, accountability frameworks
Why it matters beyond Tennessee: States are creating their own AI regulations while federal action remains fragmented. If you operate across multiple states, tracking state-level AI laws is now essential.
Quote: “AI is no longer theoretical—it is already transforming how government provides services, how businesses operate, and how Tennesseans work and learn.” - Tennessee AI Advisory Council
Timeline: Council continues work through 2028 with yearly updates.
8. The State of AI Report 2025: Barriers No Longer Technological
What happened: New report declares 2025 the “beginning of the industrial age of AI,” noting that barriers to AI’s economic potential have shifted from technological challenges to social and material constraints.
What this means: The technology works. The bottlenecks are now:
People: Adoption, training, cultural resistance
Infrastructure: Energy, data centers, chips
Organization: Processes, incentives, governance
Business takeaway: Stop waiting for AI to “get better.” The tools work now. The question is whether your organization is ready to use them effectively.
FINAL THOUGHT
The theme across all of today’s stories? AI is moving from experimental to operational and that shift is revealing who’s ready and who’s not.
College students don’t ask “should I use AI?”—they ask “how do I use it effectively?” The federal government isn’t debating AI investment—they’re mobilizing the largest scientific resources since Apollo. Insurance companies and employers aren’t wondering if AI can screen at scale—they’re already doing it.
The question for business owners isn’t “will AI change my industry?” It’s “am I building the culture, processes, and metrics to use it well?”
Because the companies seeing 45% more cost savings and 60% more revenue growth aren’t using different AI tools. They’re using the same tools differently, with business leaders driving strategy, incentives aligned to outcomes, and clear metrics on what matters.
The technology gap is closing fast. The implementation gap is widening.
Which side are you on?
-Scott