TL;DR
AWS announced new agentic AI tools at re:Invent 2025, including upgraded Amazon Connect capabilities, new Nova models and three frontier agents for development, security and operations.
The company expanded its AI Competency to cover agentic applications and tools, offering partners new funding and faster validation.
Kenyan businesses can automate customer service, claims handling, software tasks and operational workflows using these systems as demand for automation grows.
Agent observability, new hardware and updated foundation models signal AWS’s plan to make agentic AI a core part of enterprise operations in 2026.
AWS used its re:Invent 2025 conference to push agentic AI into the mainstream, positioning it as a core part of how enterprises will run their operations in 2026. The announcements comes at a time when Kenyan firms are already preparing for heavier automation across customer service, software development, security and compliance. Agentic AI featured heavily in the event, and you can see why many local organisations are now reassessing their cloud plans.
AWS introduced 29 updates to Amazon Connect and expanded its Nova model lineup to support reasoning, multimodal processing, conversational tasks, code automation and long-running workflows. The company said early adopters of its Nova Act model recorded “breakthrough 90 percent reliability for browser-based UI automation workflows”, naming Reddit and Hertz as examples of companies consolidating several tasks into a single agent-based setup. AWS executives described these changes as part of a wider push to move AI out of pilot tests and into daily operations.
AWS builds out its agentic system
The company highlighted a set of upgrades designed to help AI agents work across customer-facing and backend environments. The updates to Amazon Connect give agents the ability to “understand, reason, and act across voice and messaging channels” and handle both basic and complex customer interactions. The new Nova Sonic speech model adds tone and sentiment detection so callers can complete routine tasks through natural conversation instead of navigating menus or waiting for staff.
Also Read: Your ChatGPT conversations could end up in court
AWS also expanded its partner program. Its former Generative AI Competency is now a full AI Competency with new categories that include Agentic AI Applications, Agentic AI Tools and Agentic AI Consulting Services. According to AWS, “customers working with AI Competency Partners are rolling out production-grade AI solutions about 25 percent faster than before”, and the company noted that it has invested more than 115 million dollars into the competency since 2024.

Three frontier agents were also introduced. Kiro functions as a virtual software engineer integrated into development teams. The AWS Security Agent acts as an automated security adviser, while the AWS DevOps Agent supports day-to-day operational tasks. Organisations such as SmugMug, the Commonwealth Bank of Australia and Western Governors University have already incorporated these agents into software development and maintenance workflows.
The company also detailed improvements to Amazon Bedrock and SageMaker aimed at deeper model customisation. AWS stated that “Reinforcement Fine Tuning in Amazon Bedrock simplifies the model customization process, delivering 66 percent accuracy gains on average over base models, with customers like Salesforce demonstrating up to 73 percent improvement in accuracy over base models.”
What this shift means for Kenyan organisations
Kenyan businesses already running workloads on AWS now face a more capable set of automation tools. A December 2025 Omdia report cited in the release found that local cloud companies could increase revenues up to seven times through AI-driven services as client demand for automation grows. Telcos, fintech firms and retailers can assign routine work to AI agents, from billing checks and password resets to order tracking and delivery updates. Insurance companies can automate basic claims intake and data collection, while e-commerce platforms can use voice-based agents to assist customers in multiple languages.
AWS said that “in 2026, we’re offering an additional 25K dollars in Marketing Development Funds for achieving validation in any of the Agentic categories, in addition to the 50K dollars that Partners receive for the AI Specialization.” The company added that it has introduced an agentic review process that cuts validation times by as much as 70 percent. Kenyan partners therefore have an opportunity to expand into agentic deployments backed by funding and faster approvals.
One of the more relevant updates for regulated industries is the addition of agent observability tools. These tools allow organisations to track how AI agents reason, what data they use and how their decisions align with governance requirements. This will matter for banks, insurers and healthcare organisations that need clear audit trails.
AWS also reinforced its infrastructure-first strategy with new Trainium hardware and updated Nova foundation models. This suggests the company intends to bake agentic AI into core cloud services rather than position it as an optional layer. Kenyan organisations preparing for 2026 will likely need to factor this into their planning, especially as customer expectations shift toward faster and more responsive digital services.





Leave a Reply