AI is moving faster than anyone can forecast with confidence, so treat what follows as our opinion, not a prediction backed by data. The first wave of generative AI was about chatbots and content. In our view, the next wave is about AI that connects to your systems, acts on its own and does one job well. The shift is from AI as a tool you use to AI as an agent that works for you.

Agents and autonomous workflows

The biggest change we see coming is agentic AI. Today's chatbots need a prompt for every action. An agent is given a goal and works out the steps itself [1].

From copilot to autopilot

Most AI today is a copilot: it helps a person finish a task. We think more of it will become an autopilot that plans a multi step project, carries it out across different software, and fixes its own mistakes along the way.

For example, today a person asks AI to write an email and then books the meeting by hand. An agent could find the lead, research their background, draft the outreach, book the meeting through your calendar and update the CRM, all without a person in the middle.

Personalization for every customer

We expect personalization to move past generic templates. With live data, a system could adjust prices or change what a customer sees on screen based on what they're doing at that moment.

How different industries might change

Each industry will feel this differently. What they share is less manual thinking work.

Healthcare and biotech

AI is already used to screen huge numbers of candidate molecules, and we expect it to take on more of early drug discovery. People would still run the clinical trials and make the final calls.

Manufacturing and logistics

Robots paired with AI make "dark warehouses" possible: facilities that run with the lights low and few people on site, because machines handle sorting and packing.

Finance and legal

Document review is a natural fit. We expect AI to go beyond summarizing contracts and start flagging regulatory risk as rules change, which would speed up compliance checks that now take a long time.

What it could mean for jobs

Some repetitive roles will shrink. We also expect more demand for people who can design, run and oversee AI systems, the "orchestrators" who connect business needs to what the tools can do.

Obstacles

None of this is guaranteed. Three problems stand out.

Privacy and control of data

Agents need access to internal data to be useful, which raises the stakes on security. One response we're watching is small language models that run on a company's own hardware instead of in the cloud.

The skills gap

There aren't enough people who understand both the business side and the AI side. Companies will need people who can translate between the two.

Regulation

We expect governments to ask for more transparency about how AI makes decisions. Risk management is becoming the price of admission for AI [2]. A business running autonomous agents should be able to show the steps an agent took to reach a decision.

How to prepare

You can start on all of this now.

Review your manual work

Find tasks that are repetitive and follow clear rules. Those are the first ones an agent could take over.

Clean up your data

AI is only as good as the data it reads. Getting your data into structured, accessible formats now makes it much easier to plug in new tools later.

Get your team comfortable

Encourage people to try AI tools today. The goal is for staff to see themselves as the ones directing the tools, not the ones being replaced by them.

Our take

We see AI automation as a way to give people more reach, not to remove them. If agents and personalized systems keep improving, they'll change how work gets done in most industries.

If you want to work out where AI fits in your business, contact us and we'll help you build a plan.