Microsoft has lowered sales growth targets for its AI agent products after many salespeople missed their quotas in the fiscal year ending in June, according to a report Wednesday from The Information. The adjustment is reportedly unusual for Microsoft, and it comes after the company missed a number of ambitious sales goals for its AI offerings.
AI agents are specialized implementations of AI language models designed to perform multistep tasks autonomously rather than simply responding to single prompts. So-called “agentic” features have been central to Microsoft’s 2025 sales pitch: At its Build conference in May, the company declared that it has entered “the era of AI agents.”
The company has promised customers that agents could automate complex tasks, such as generating dashboards from sales data or writing customer reports. At its Ignite conference in November, Microsoft announced new features like Word, Excel, and PowerPoint agents in Microsoft 365 Copilot, along with tools for building and deploying agents through Azure AI Foundry and Copilot Studio. But as the year draws to a close, that promise has proven harder to deliver than the company expected.
On Tuesday, Microsoft and Nvidia announced plans to invest in Anthropic under a new partnership that includes a $30 billion commitment by the Claude maker to use Microsoft’s cloud services. Nvidia will commit up to $10 billion to Anthropic and Microsoft up to $5 billion, with both companies investing in Anthropic’s next funding round.
The deal brings together two companies that have backed OpenAI and connects them more closely to one of the ChatGPT maker’s main competitors. Microsoft CEO Satya Nadella said in a video that OpenAI “remains a critical partner,” while adding that the companies will increasingly be customers of each other.
“We will use Anthropic models, they will use our infrastructure, and we’ll go to market together,” Nadella said.
Cisco plans to acquire NeuralFabric, a Seattle-area startup founded by a group of Microsoft veterans that makes back-end software for companies to build and run their own generative AI models. Financial terms were not disclosed.
The Silicon Valley enterprise tech mainstay said the deal will bolster its AI Canvas initiative, a generative UI and collaboration environment announced earlier this year.
In its announcement Thursday morning, Cisco highlighted NeuralFabric’s expertise in distributed systems, model training, and flexible deployment as a complement to its existing AI assistant, cybersecurity models, and data fabric strategy.
DJ Sampath, senior vice president for AI software and platforms, said in the announcement that the startup has “cracked a crucial part of this puzzle” by building technology that lets companies develop their own domain-specific small language models using proprietary data across cloud or on-premises environments.
NeuralFabric, based in Redmond, was founded in 2023 by former Microsoft Azure engineering veteran Weijie Lin (CEO), longtime Microsoft executive John deVadoss, AI entrepreneur Jesus Rodriguez (president), and cloud and security veteran Mark Baciak (CTO), with former Microsoft director Drew Gude (chief revenue officer) also listed as an early exec.
The startup employs about nine people, according to LinkedIn. Cisco said the acquisition is expected to close in the second quarter of its 2026 fiscal year (by the end of January), after which NeuralFabric’s team will join the company’s AI Software and Platform organization.
NeuralFabric had raised at least $5 million in funding as of February 2024 announcement. PitchBook lists investors including Collab+Currency, CMT Digital, and New Form Capital.
With its no-code app builder now widely available, Google is betting that tools like Opal can turn everyday users into AI builders and expand the reach of its cloud ecosystem.
AI voice startup WellSaid Labs is doubling down on its niche of enterprise customers and regulated industries — hoping that a more judicious, behind-the-scenes approach will pay off for its business in the long run even as flashier rivals draw widespread attention and controversy.
The company, based in Bellevue, Wash., launched a new version of its text-to-speech AI voice platform Monday with redesigned Studio software and its next-generation Caruso voice model, promising better workflows, improved audio quality, and fine-tuned controls, among other features.
Unlike open voice-generation models that scrape public data, WellSaid’s system is trained exclusively on licensed voice actor recordings, a closed-model approach that it says respects intellectual property and appeals to sectors such as healthcare, legal, and finance.
WellSaid’s latest release is a pivotal moment for the company — the result of years of internal research now coming to market in a form that refines its focus on business and institutional users, said Chris Johnson, WellSaid’s chief product and technology officer, in an interview.
Chris Johnson, WellSaid’s chief product and technology officer. (WellSaid Photo)
“We put our stake in the ground in being the best solution for enterprises in the market,” Johnson said. “A lot of these innovations accrue to making that a reality for us.”
WellSaid, which spun out of Seattle’s AI2 Incubator in 2019, works with large enterprise customers including LinkedIn, T-Mobile, ServiceNow, and Accenture.
The company made an impression on the public in 2023 when NPR’s Planet Moneyused WellSaid’s technology to create a synthetic version of former host Robert Smith’s voice — a near-perfect replica that surprised listeners and showed both the promise and potential challenges of realistic AI audio.
But WellSaid has struggled at times to break into the larger industry conversation. The challenge was underscored by its absence from a CB Insights market map of leading voice-AI startups — topped by buzzy ElevenLabs, which has been at the center of controversy over the use of its technology to make fake AI voices of public figures and others.
WellSaid executives say they’re hoping to correct that specific oversight, but the challenge reflects a broader pattern among enterprise AI companies, particularly those in the Seattle region — which often emphasize trust, governance, and regulatory scenarios in a tech culture still captivated by headline-generating Silicon Valley experiments and consumer apps.
A spokesperson said enterprise customers are WellSaid’s fastest-growing segment, expanding six-fold in three years with net retention of more than 150%. Its business model is lean, having raised about $20 million, letting it run efficiently while paying voice actors royalties and offering an equity program. The company employs about 70 people, down slightly from a year ago, according to LinkedIn data.
WellSaid says it’s seeing new momentum in the application of its AI technology to advertising, due to an increase in voice quality and a decrease in related content production costs.
The company has seen some turnover in its executive suite, including three CEOs in less than two years — starting with founder Matt Hocking (who is still chairman), then Brian Cook, and now Benjamin Dorr, who succeeded Cook earlier this year after serving as chief financial officer.
“Every voice is connected to a real person, and that person receives royalties from the revenue that’s generated on WellSaid,” Dorr said on a recent episode of the Master Move podcast. “I think the things we do right by our voice actors allow us to do right by the enterprises that choose us, and I don’t think everyone else can say that.”
Login
