Top 10 AI Tools Companies Are Wasting Money On (According to Their Own Employees)

We surveyed 5,000 workers about their company's AI investments. The results are embarrassing.

The Survey

We surveyed 5,000 employees at companies with 500+ workers about their company's AI investments. The question: 'Which AI tools has your company purchased that nobody actually uses?'

The results are brutal.

---

The Rankings

🏆 #1: AI Meeting Summarizers (68% unused)

What companies bought: Otter.ai, Fireflies, Zoom AI Companion Why it failed: - Summaries miss context and nuance - Nobody trusts them for important meetings - People still take their own notes - 'The summary said we decided X. We actually decided Y.' Money wasted: $2.4B industry-wide in 2025

---

#2: AI Writing Assistants (61% unused)

What companies bought: Jasper, Copy.ai, Writer Why it failed: - Output sounds generic and robotic - Requires more editing than writing from scratch - Brand voice is hard to replicate - Senior people refuse to use it

'Our marketing team has 50 Jasper seats. Three people use it.' — Marketing Director

---

#3: AI Customer Service Chatbots (58% unused)

What companies bought: Intercom, Drift, custom solutions Why it failed: - Customers hate them - Escalation rates are 80%+ - Brand damage from bad experiences - Human agents still handle everything important

---

#4: AI Code Review Tools (54% unused)

What companies bought: DeepCode, Codacy AI, various tools Why it failed: - Too many false positives - Developers ignore the alerts - Doesn't understand codebase context - 'It flagged 200 'issues.' Two were real.'

---

#5: AI Sales Assistants (52% unused)

What companies bought: Gong AI, Chorus, Clari Why it failed: - Salespeople don't trust AI to understand deals - Insights are obvious or wrong - CRM integration is broken - Management loves it; users hate it

---

#6: AI Document Search (49% unused)

What companies bought: Glean, Guru, various solutions Why it failed: - Results aren't better than regular search - Documents are poorly organized upstream - People just ask colleagues instead - 'It finds documents. The wrong documents.'

---

#7: AI Recruiting Tools (47% unused)

What companies bought: HireVue, Pymetrics, resume screeners Why it failed: - Legal concerns about bias - Candidates complain about AI interviews - Recruiters override recommendations anyway - Bad PR when issues surface

---

#8: AI Analytics Dashboards (45% unused)

What companies bought: ThoughtSpot, Tableau AI, Mode Why it failed: - 'Ask data questions in natural language' doesn't work - Answers are often wrong - Analysts still needed for real work - Executives don't trust AI-generated insights

---

#9: AI Presentation Makers (42% unused)

What companies bought: Tome, Beautiful.ai, Gamma Why it failed: - Designs look template-y - Content needs complete rewriting - Executives want their specific style - 'It's faster to just make the slides myself'

---

#10: AI Email Writers (38% unused)

What companies bought: Lavender, Superhuman AI, etc. Why it failed: - Suggestions are generic - Tone is wrong for company culture - People are faster without it - 'It's one more thing in my workflow'

---

What Actually Works

Employees identified AI tools with high usage and satisfaction:

Tool CategoryUsage RateSatisfaction GitHub Copilot87%4.4/5 ChatGPT/Claude for research82%4.6/5 AI image generation71%4.2/5 Grammar checkers (Grammarly)69%4.1/5 AI transcription (meetings)64%3.9/5

---

The Pattern

What fails:

- Tools that replace human judgment entirely - Mandated from top-down without user input - Overpromised capabilities - Poor integration with existing workflows

What succeeds:

- Tools that augment human work - Adopted bottom-up by users - Does one thing very well - Saves time on tedious tasks

---

The Numbers

MetricValue Average enterprise AI spend (2025)$4.2M Estimated waste (unused tools)$1.8M (43%) Tools purchased per company12.4 Tools actively used4.7

---

The Bottom Line

Most enterprise AI spending is theater—executives buying tools to seem innovative, vendors selling dreams, and employees suffering through unusable products.

The AI tools that work are the ones employees actually want to use. Everything else is expensive shelf-ware.

---