Inside the cavernous halls of the Las Vegas Convention Center this week, Amazon Web Services set out an ambitious vision of the workplace future: software agents that listen, decide and act on their own, without waiting for a human in the loop.
At its annual cloud computing conference, AWS:reInvent, the company’s executives described the arrival of what they call “agentic AI” as a tectonic shift in how businesses will operate. The company said 2026 would be the year artificial intelligence moves beyond pilot projects and visibly reshapes daily enterprise work.
That shift may arrive quickly in Africa. An IDC report cited by AWS shows that 23 percent of organizations expect to fully deploy agentic AI systems within the next 12 months, while 65 percent anticipate full deployment by 2027.
What AWS unveiled
At the center of AWS’s push is Amazon Connect, its cloud based contact center platform. The company introduced 29 new capabilities designed to let AI agents handle customer interactions across voice and messaging, interpreting requests, reasoning through responses and completing tasks without human escalation.
Amazon also expanded its Nova family of AI models. Four new models are designed to improve price performance across reasoning, multimodal processing, conversational AI, code generation and agentic workloads. One of the new releases, Nova Forge, introduces what Amazon calls “open training,” allowing organizations to access pre trained model checkpoints and combine their own proprietary data with Amazon curated datasets.
“Nova Act achieves breakthrough 90% reliability for browser-based UI automation workflows built by early customers. Companies like Reddit are using Nova Forge to replace multiple specialized models with a single solution, while Hertz accelerated development velocity by 5x with Nova Act,” the company said.
A new speech model called Nova Sonic allows agents to interpret tone, sentiment and intent. In practical use, a customer calling to check an order or request a refund can complete the transaction through a natural language conversation without navigating menus or waiting for a human agent.
For companies that still rely on human staff, AWS added real time AI assistance. The tools surface context, recommend next steps and automate routine work such as documentation and form filling, allowing staff to focus on harder cases.
AWS also broadened its partner program, transforming its former Generative AI Competency into a full AI Competency, with three new categories: Agentic AI Applications, Agentic AI Tools, and Agentic AI Consulting Services.
“Since its launch in 2024, we have invested over $115 million in the AWS AI Competency, demonstrating our commitment to accelerating successful customer AI adoption through Specialization Partners,” AWS said in a statement. “These new categories reflect the rapid market advancement, helping customers find Partners who can move them from experimentation to production-ready autonomous systems.”
The company said it now recognizes more than 60 validated Agentic AI partners and that customers working with these partners are deploying production systems 25 percent faster than before.
“This expansion to include Agentic AI categories reinforces our commitment to help customers move from experimentation to production-grade autonomous systems that can perceive, reason, and act independently,” said Ruba Borno, vice president of global specialists and partners at AWS. “This isn’t just about technology validation; it’s about connecting customers with Partners who have proven they can deliver responsible, scalable AI solutions that drive measurable business value.”
Frontier agents arrive
AWS also introduced what it calls frontier agents: autonomous software systems designed to work continuously over long periods. The company unveiled three.
The Kiro autonomous agent functions as a virtual software developer embedded in engineering teams. The AWS Security Agent is built to act as a dedicated security adviser. The AWS DevOps Agent is positioned as an always on operational support system.
AWS said organizations including Commonwealth Bank of Australia, SmugMug and Western Governors University have already used these agents to streamline parts of the software development lifecycle.
To address the cost of running large scale AI systems, AWS added new customization tools inside Amazon Bedrock and Amazon SageMaker AI.
“Reinforcement Fine Tuning (RFT) in Amazon Bedrock simplifies the model customization process, delivering 66% accuracy gains on average over base models, with customers like Salesforce demonstrating up to 73% improvement in accuracy over base models,” the company stated.
What it means for Kenya
For Kenyan companies already running workloads on AWS, the changes point toward deeper automation. Telecom operators, fintech firms, retailers and customer service providers could deploy AI agents to handle billing inquiries, password resets and account checks, freeing human workers to handle more complex cases.
A December 2025 Omdia report found that cloud firms and system integrators could see revenues grow up to seven times as demand for AI driven services increases.
Insurance companies could automate claims workflows. E commerce platforms could use AI agents to manage order tracking, returns and delivery queries in multiple languages.
“In 2026, we’re offering an additional $25K in Marketing Development Funds (MDF) for achieving validation in any of the Agentic categories, in addition to the $50K in MDF that Partners receive for the AI Specialization,” AWS said. “To accelerate your path to the AI Specialization, we launched an agentic review process that provides real-time guidance and automated validation, cutting the application processing time by up to 70%.”
Because Amazon Connect’s agents can understand natural speech and detect sentiment, customer services may become more accessible in Kenya, where voice based support remains critical.
For local startups and AWS partners, the expanded AI Competency creates a new commercial opening. Firms with strong technical teams can now design and deploy autonomous AI workflows while tapping AWS funding and accelerated validation.
Governance and oversight
One of the most significant additions is agent observability, a set of tools that allows companies to track how AI agents make decisions, which data sources they access and whether their actions comply with governance rules.
In regulated sectors such as banking, insurance and healthcare, this level of visibility could prove decisive. Organizations can audit automated decisions and demonstrate compliance to regulators and internal risk teams.
AWS framed the technology as augmentation rather than replacement of human work. The company said agentic AI is designed to take over repetitive, rule based work, while humans remain responsible for judgment, oversight and sensitive decisions.
The announcements also reflect AWS’s broader infrastructure strategy. Alongside agentic software, the company introduced new AI chips under the Trainium line and updated foundation models in the Nova family, aimed at making large scale autonomous systems easier to build and operate.
For African cloud markets, the stakes are high. Data sovereignty and regulatory barriers have often slowed adoption of advanced automation. Cloud native, tool based agentic systems could give firms a path to scale without managing expensive on premises infrastructure.