Ethora Research maps conversational AI’s shift to agents and open source

Ethora Research has released a free, vendor-neutral Q3 2026 report on conversational AI, built from primary data and released under CC BY 4.0. The findings show search interest shifting toward agents, self-hosting gaining traction with developers, and buying-intent queries increasingly answered by AI before users reach vendors.

Why it matters: - The report gives builders, analysts and journalists a reusable map of a fast-changing conversational-AI market. - The release is free to read and reuse under CC BY 4.0, which lowers the barrier to independent scrutiny and reuse. - The report is based on measured sources rather than vendor surveys or a single-use enterprise ranking.

What happened: - Ethora Research published The State of Conversational AI, Q3 2026 on August 28, 2026. - The report covers the conversational-AI stack and is designed as a vendor-neutral, primary-data map. - Ethora Research says the report, charts and underlying datasets are free to read and reuse under a Creative Commons licence. - The release includes a report link, a data repository and a DOI: Read the report, Data and scripts, and DOI 10.5281/zenodo.22131590.

The details: - The report is built from GitHub, Hugging Face, package registries, search demand, sampled AI-assistant answers, open-job counts and talent data. - US search for "ai agents" rose from roughly 60 searches a month in early 2023 to about 50,000 a month in 2026. - That growth is close to a 700-fold increase and now leads the category’s vocabulary. - The US leads on absolute AI-role demand and supply. - India is the largest net exporter of practitioners. - Switzerland and Singapore top the per-capita talent tables. - The report says that result matches the Stanford AI Index. - Open-source projects that let teams run models and chat on their own infrastructure are the most-starred projects in the space. - The report frames that as developer and early-adopter momentum, while noting most enterprises still use cloud-hosted AI. - The Model Context Protocol, or "mcp server," had effectively no search volume before late 2024. - Search for the protocol reached tens of thousands of monthly queries in about 18 months. - The protocol’s developer libraries are already among the most-downloaded packages in the field. - Of 28 buying-intent search terms tracked, 24 trigger a Google AI Overview. - The assistants pull recommendations largely from a small set of third-party "best of" lists rather than directly from vendors. - The report also introduces a "Periodic Table of Conversational AI" covering 18 notable systems verified against current documentation. - The report adds a five-layer model showing how a conversational-AI product is assembled, from interface to infrastructure.

Between the lines: - The report suggests discovery is shifting away from vendor-controlled channels and toward AI-mediated summaries and third-party listicles. - The rapid rise of agent-related search and MCP interest points to a market reorganising around orchestration, tool use and infrastructure choice. - Open-source momentum may be leading commercial adoption, but the report treats enterprise self-hosting as an open question rather than a settled trend. - The talent and demand data point to a more geographically distributed field, with concentrated hubs outperforming on a per-capita basis.

What's next: - Ethora Research says every figure traces to a dated source and every table links to full underlying data. - The collection scripts are published so the numbers can be reproduced and rerun in future editions. - Journalists and analysts can quote the report and reuse charts and data with attribution, and can request underlying datasets or additional cuts from Ethora Research. - The social links in the release point readers to Ethora’s LinkedIn, Facebook, YouTube and X accounts for updates.