AI Trends to Watch in 2026
By John Martines · May 20, 2025 · 8 min read
The most important AI developments coming in 2026 and how SMBs should prepare.
2025 was the year AI became real. 2026 is the year it becomes embedded. If you're waiting for AI to "settle down" before adopting it, this year will be the wake-up call.
Here's what's coming, and what you need to be ready for.
Agentic AI isn't new, but it's been mostly experimental. This year, it's moving into production across industries.
What This Means
Instead of asking an AI to do a single task ("write me an email"), you give it a goal and it breaks it down into steps, executes them, handles errors, and reports back. It doesn't need a human to supervise every step.
Example: "Generate a customer invoice from the CRM data, send it via email, update the AR aging report, and flag any overdue accounts for the accounting team." One instruction. The agent does all of it.
For NEPA Businesses
- Look for agentic features in your existing software (ERP, CRM, accounting platforms). Most major vendors are adding them.
- Expect ROI conversations to shift from "time saved" to "processes completed end-to-end with less human oversight"
- Security and governance become critical — you need to know what your agent is doing and who can authorize what
- Expect the first generation of industry-specific agents (construction agents, manufacturing agents, healthcare agents)
Action: Audit your top 3 workflows. Which ones could an agent handle end-to-end? That's your pilot candidate.
AI that only reads text is 2023. In 2026, expect AI that fluently handles images, audio, video, and structured data in the same conversation.
Business Impact
- Construction/trades: Upload a site photo, AI identifies materials, estimates costs, generates a materials list
- Manufacturing: Upload a design sketch or 3D model, AI checks it against standards, flags issues, suggests improvements
- Healthcare: AI listens to a patient encounter, reviews attached images (X-rays, lab results), generates a complete clinical note
- Professional services: AI reviews contracts, scans supporting documents, extracts terms, flags risks in one pass
The shift is from "AI handles one data type" to "AI handles all your data, whatever form it's in."
Action: Inventory your data. What's text? What's images? What's video? What could you automate if AI could see/hear it?
Generic AI is giving way to AI built specifically for your industry and your software.
What's Happening
- ERP systems: SAP, NetSuite, Acumatica are all embedding industry-specific AI agents into their platforms
- Accounting software: QuickBooks, Xero, and niche accounting platforms now have native AI that understands accounting workflows
- Construction software: Procore, Bluebeam, and PlanSwift are adding AI that understands construction terminology and processes
- Healthcare platforms: Epic, athenahealth, and Medidata are embedding clinical AI that knows HIPAA and healthcare workflows
- Manufacturing: Manufacturers are seeing industry-specific AI for demand forecasting, quality control, and supply chain optimization
For Your Business
The AI that will actually work for you isn't generic ChatGPT. It's the AI that lives in your software and understands your industry.
If you're still using general-purpose AI for specific business tasks, you're doing it wrong (or at least inefficiently).
Action: Check with your software vendors. What AI do they have? What are they building? Make this part of your vendor evaluation.
Text-based AI is convenient, but voice is natural. In 2026, expect voice agents to handle customer service, scheduling, payment collection, and more.
What's Changing
- Latency is solved: You can have a natural conversation with an AI with minimal lag — no more "the robot is thinking" pauses
- Phone quality is real: AI agents can make and receive phone calls that sound natural, handle accents and background noise, and navigate complex conversations
- Security is baked in: Voice agents can verify identity via voiceprint and handle sensitive calls securely
Use Cases
- Appointment scheduling: Instead of texting confirmation links, patients/customers call a number and talk to an agent
- Payment collection: AI agents can call overdue accounts, verify information, and collect payments (with proper compliance)
- Customer support first line: AI agent answers the phone, triages the issue, and routes to a human if needed
- After-hours support: Your business never closes — an AI agent handles calls when your team is sleeping
Action: Identify phone-based workflows that could be automated. Outbound calling (confirmations, collections, surveys) is the first target.
2025 saw some initial regulation (EU AI Act, California's SB 1047). 2026 is when it gets real and spreads.
What to Expect
- Industry-specific compliance: Healthcare, finance, and legal AI will face stricter rules. Behavioral health is already seeing it.
- Data privacy rules tightening: What you can do with your data when running it through AI models will be more restricted
- Audit and transparency requirements: You may need to be able to explain why your AI made a decision (especially in hiring, lending, healthcare)
- Liability shifting: If your AI causes harm, who's responsible? Expect more clarity (and more finger-pointing)
A free account opens the rest of this article and the whole library.