AI in January 2026: agents move onto the desktop
January set the tone for the year. AI agents moved out of developer tools and into the everyday software that operations teams already use, while one of the world's biggest analyst firms told businesses to expect a more sober year of proving value. Here are the developments worth knowing about, and what they mean for your organisation.
Agents arrive for everyday office work
On 13 January, Anthropic launched Cowork, a research preview that brings the agent technology behind its Claude Code developer tool to non-technical users. Rather than answering one question at a time, Cowork can read, organise and create files on a user's computer and work through multi-step tasks. Anthropic's examples included turning photos of receipts into an expense spreadsheet and drafting documents from scattered notes. At launch it was limited to subscribers on Anthropic's Max plans.
The same day, Salesforce made its rebuilt Slackbot generally available to Business+ and Enterprise+ customers. Salesforce describes it as a personal agent for work: it draws on Slack conversations, files and Salesforce data, stays within each user's existing permissions, and can draft notes, schedule meetings and kick off workflows.
What it means for your business: Agents are no longer something only engineering teams experiment with. The tools your staff already use are gaining the ability to carry out tasks, not just suggest text. That makes now a good time to identify the repetitive, document-heavy processes (expenses, reporting, admin) where an agent could take on the legwork, and to agree the guardrails before staff adopt these tools on their own.
Apple chooses Google's Gemini to power the new Siri
In mid-January, Apple and Google confirmed a multi-year agreement under which Google's Gemini models will underpin Apple's own foundation models, including a revamped Siri due later in 2026. In a statement, Google said Apple had determined, after careful evaluation, that Google's technology "provides the most capable foundation". Financial terms were not disclosed.
What it means for your business: Many of your customers and employees use Apple devices. A more capable voice assistant will raise expectations of how quickly and conversationally people can get answers, book services or find information. It is also a reminder that even the largest technology companies combine their own work with partners' models. For most organisations, choosing and integrating the right models matters more than building them.
Gartner forecasts $2.5 trillion of AI spending, with a warning
On 15 January, Gartner forecast that worldwide AI spending will reach about $2.52 trillion in 2026, up 44% on 2025, with infrastructure accounting for more than half. More interesting for business leaders was the commentary. Gartner says AI sits in the "Trough of Disillusionment" throughout 2026, and expects it will most often be sold to enterprises by their incumbent software provider rather than bought as part of a new moonshot project. Its analyst John-David Lovelock added that improved predictability of return on investment must come before AI can truly be scaled.
What it means for your business: Spending is rising, but so is scrutiny. Projects that start with a clear business case, a measurable baseline and a realistic scope are the ones that will survive budget reviews. It is also worth checking what AI features are already included in the software you pay for before buying something new.
The UK offers free AI training to every adult
On 28 January, the UK government expanded its AI skills programme with the aim of giving 10 million workers key AI skills by 2030, including at least 2 million employees of small and medium-sized businesses. The free online courses, many under 20 minutes long, cover practical tasks such as drafting text and handling admin, and lead to a foundations badge benchmarked by Skills England. Partners include Microsoft, Google, Salesforce, Sage, the Federation of Small Businesses and the British Chambers of Commerce. Government research cited alongside the launch found only 21% of UK workers feel confident using AI at work.
What it means for your business: Low confidence, not access to tools, is often what holds adoption back. Free, short courses are an easy first step for building a shared baseline across your team, especially in smaller firms without a training budget. Pair them with clear internal guidance on which tools are approved and what data can be used.
Practical takeaways
List three to five repetitive, file-based processes that a desktop or workplace agent could support, and pilot one with clear success measures.
Review the AI features already included in your existing software licences before committing to new platforms.
Set out simple rules on approved AI tools and data handling now, before agents spread informally across teams.
Encourage staff to complete free foundation courses, such as the UK government's AI skills programme, to build confidence.
Build every AI business case around a measurable baseline so you can show return on investment.
If you are wondering where agents could make a real difference in your organisation, or how to turn this year's tighter focus on ROI into a clear plan, we would be glad to help. Book a call with the Manta Global AI team and we can talk through what is possible for your business.
Comments