MIT Tech Review Insights — Building the AI-enabled enterprise of the future
TL;DR
Sponsored research from MIT Technology Review Insights in partnership with Cisco. Argues 2024–25 is an inflection point for enterprise AI: 98% of companies feel increased urgency, 85% give themselves <18 months to deploy strategy, but only 13% are globally ready to leverage AI to its full potential. Five “foundations” framework (Strategy / Infrastructure / Data / Governance / Culture and Talent) plus heavy emphasis on the AI agent transition and AI security.
Key claims
Urgency and readiness gap
- 98% of companies feel increased urgency over the past year.
- 85% say they have less than 18 months to deploy an AI strategy or face negative business effects.
- For 50% of organizations, CEOs and leadership are driving AI urgency more than any other factor.
- AI is a priority IT spend: 50% of companies have already dedicated 10–30% of their IT budget to AI.
- Only 13% globally are ready to leverage AI to its full potential.
- Two-thirds (68%) say their infrastructure is at best moderately ready to adopt and scale AI.
Industry impact (selected)
- Healthcare/pharma: AI reducing drug discovery timelines by as much as 50%.
- R&D cycles: AI may reduce time-to-market by 50%.
- Automotive and aerospace: AI lowering costs by up to 30%.
- Manufacturing: 72% of manufacturers report AI has reduced costs and improved operational efficiency (per National Association of Manufacturers).
Manufacturing AI use by function (NAM data, 2025)
| Function | % of manufacturers using AI |
|---|---|
| Manufacturing & production | 39% |
| Inventory management | 33% |
| Quality operations | 24% |
| R&D | 24% |
| IT/OT | 21% |
| Equipment maintenance/installation | 17% |
| Supply chain | 11% |
| Product design | 11% |
| Distribution/logistics | 9% |
| Sales & marketing | 7% |
| Finance | 7% |
| HR | 6% |
| Customer service | 4% |
| Legal | 3% |
| Procurement | 3% |
| Sustainability | 3% |
Ford case study (concrete numbers)
- AI-augmented vision systems inspect part kits before assembly. On one assembly line for “squish tubes” (rubber seals in electric oil pumps cooling hybrid vehicle motors), defects went from 63 per month to zero.
- Computational fluid dynamic test simulating airflow around vehicles: previously 15 hours per run, now 10 seconds with AI prediction.
- Ford CISO Patrick Milligan quoted on AI as core to transformation strategy.
The agent transition (3-stage model)
The report frames a progression: AI chatbot → AI agent → multi-agent system:
- Chatbot: simulates and processes human conversation
- Agent: pursues complex goals autonomously, with independent decision-making, planning, adaptable execution in dynamic environments
- Multi-agent system: multiple AI agents collaborate to pursue complex goals in dynamic environments
Stats from external research:
- AI agents could double the capacity of knowledge professionals and field-support roles (PwC).
- >80% of organizations plan to integrate AI agents within 1–3 years (Capgemini research).
- 64% believe AI agents will significantly improve customer service.
- Customers are 3.8× more likely to purchase again following a successful experience.
Cisco’s Capgemini-cited stats on agent expectations:
- 71% AI agents will help drive higher levels of automation
- 64% AI agents will significantly improve customer service
- 64% AI agents will help focus on more value-added activities
- 57% the potential of AI agents to improve productivity outweighs its risks
The “Foundations” framework
To capture full AI value, Cisco argues for getting five foundations right:
- Strategy — well-thought-out, clearly defined AI strategy aligned with business goals
- Infrastructure — evaluate network capacity for AI workloads (compute, energy, data, security; reaccelerating private data center for data sovereignty)
- Data — good quality, company-wide data; break down silos; AI tools to query unstructured data
- Governance — strengthen data governance per local/global regulations; responsible AI; regular review of policies
- Culture and talent — develop tomorrow’s skills today; allay job-displacement fears; reward successful initiatives
AI security (sidebar)
Cisco’s fundamentals for AI safety:
- Understand AI security and safety taxonomy
- Identify vulnerabilities in AI models, software, and hardware code
- Secure vector databases
- Use established security best practices for AI training environments
- Establish AI security as ongoing practice
- Use reference architectures for LLM training environments
- Select secure embedding models for content generation
Quoted: Jeetu Patel (Cisco President & CPO): “Safety and security are fundamental, because they’re one of the big fears impeding adoption for AI technologies today. So if you don’t trust something, you’re not going to use it.”
Notable quotes
“I don’t worry about AI taking my job, but I definitely worry about another person or company that uses AI better than me taking my job or making my company irrelevant.” — Jeetu Patel, President & Chief Product Officer, Cisco
“It’s not about replacing roles. It’s about where we can give agency, with some human oversight and governance, to improve tasks within a workflow.” — Liz Centoni, EVP & Chief CX Officer, Cisco
“This is one of those inflection points where I don’t think anybody really has a full view of the significance of the change it is going to have on not just companies but society as a whole.” — Patrick Milligan, CISO, Ford
My take
This is sponsored research, not journalism — every framing benefits Cisco’s positioning as an “AI infrastructure / network / security” vendor. Discount accordingly when reading the “Infrastructure” and “Security” sections.
That said, the agent-transition framing (chatbot → agent → multi-agent) is genuinely useful and reinforces what AI Index 2025 (RE-Bench, Salesforce Agentforce) and MIT Sloan (Italgas DANA, Stage 3+ “exploring autonomous agents”) said separately. With four sources now converging on agents as a near-term enterprise reality, ai-agents earns its own concept page.
The 13% / 98% / 85% triad (ready / urgent / <18-month deadline) is a useful complement to the AI Index adoption stats and the MIT CISR maturity stages. It’s a third measurement instrument for the ai-maturity-measurement-comparison thread — measuring urgency and infrastructure readiness rather than adoption breadth or maturity stage.
The Ford case study (63 → 0 defects on squish tubes; 15 hours → 10 seconds for fluid dynamics) is concrete and quantifiable; together with the Guardian RFP example (1 week → 24 hours) and Italgas’s WorkOnSite (+40% construction speed), we now have a small portfolio of concrete enterprise AI ROI numbers to cite.
Linked entities and concepts
Entities (this wiki): Cisco, MIT Technology Review Insights. Dangling: Jeetu Patel, Liz Centoni, Patrick Milligan, Ford, Capgemini, PwC, National Association of Manufacturers, Virginia Wilson, Nicola Crepaldi.
Concepts: ai-agents (created by this ingest as concept threshold reached), enterprise-ai-adoption, responsible-ai (Cisco AI security fundamentals), generative-ai, industry-4-0 (manufacturing context).
Threads: ai-maturity-measurement-comparison (enriched), organizational-frameworks-for-ai-adoption (new — Cisco’s “5 foundations” enters the framework comparison).
Source
- Raw PDF (8 pp): report file
- Publisher: MIT Technology Review Insights (custom publishing arm of MIT Technology Review)
- Sponsor: Cisco (Cisco enabled the research; MIT TRI claims independent reporting)
- © 2025 MIT Technology Review Insights