Microsoft has introduced new artificial intelligence features that allow its Copilot assistant to use multiple models simultaneously, as the company seeks to improve reliability and accelerate adoption of its enterprise AI tools.
The upgrade centres on a new “Critique” feature within Copilot’s Researcher tool, which combines OpenAI’s GPT models with Anthropic’s Claude system in a single workflow. GPT generates responses while Claude reviews them for accuracy and quality before output is delivered to users.
Nicole Herskowitz, corporate vice president of Microsoft 365 and Copilot, told Reuters that “having various different models from different vendors in Copilot is highly attractive,” adding that the company is “taking this to the next level” by enabling the systems to work together. She said the approach is designed to improve productivity and reduce so-called AI hallucinations, where systems generate incorrect information.
Microsoft has also launched a “Council” feature, which presents side-by-side outputs from different AI models along with a summary highlighting where they agree or diverge. The company said the architecture mirrors academic research processes by separating content generation from evaluation, aiming to improve analytical depth and presentation quality.
According to Microsoft’s own blog, the multi-model system has shown measurable gains in internal testing, outperforming single-model approaches on benchmarks assessing accuracy, completeness and objectivity. The company said the Critique feature improved aggregate scores by around 13.88 per cent compared with rival systems in the same category.
The upgrades are being rolled out through Microsoft’s Frontier programme, which provides early access to experimental AI tools, including the Copilot Cowork agent introduced earlier this month. The product reflects growing demand for autonomous AI agents capable of handling complex, multi-step tasks.
The developments come as Microsoft faces intensifying competition in generative AI from rivals such as Google’s Gemini and Anthropic’s Claude offerings. The company’s shares rose about 1 per cent following the announcement, although Reuters noted the stock remains on track for its worst quarter since the 2008 financial crisis, with a decline of nearly 25 per cent as investor enthusiasm for AI moderates.
Separately, Reuters reported that Microsoft plans to invest $1 billion in Thailand over the next two years to expand cloud and AI infrastructure, alongside initiatives to develop digital skills in the local workforce. The move aligns with broader efforts across Southeast Asia to scale data centre capacity and support AI-driven growth.
.jpeg "UK chip start-up seeks 0m to expand manufacturing capabilities")
