Every few years someone in research distribution proposes the obvious fix. Put all the research in one place, give readers one way in, and stop making everyone integrate with everyone else.
The idea is right. The trouble is usually who owns the place.
When the single platform belongs to a bank, other banks are wary of putting their research on it. When it belongs to a vendor whose main business is selling its own data, providers worry about what happens to theirs. When it belongs to a buy-side firm, it serves that firm. Each version solves part of the problem for one side and creates a new one for the other.
This piece sets out why we think the answer is a neutral hub, and why neutrality is a design requirement rather than a slogan.
The rail, not the train
Railways are a useful picture. The operators run the trains, set the timetables and sell the tickets. The rail underneath is shared, built to a common standard and open to any operator that meets it. Nobody expects the track to favour one company's trains.
Research distribution needs the same split. Brokers and independent providers are the trains. They write the research, set its price and decide who gets it. Readers choose which trains to ride. What has been missing is the rail: a shared layer that carries research from any provider to any entitled reader, under the provider's rules, without itself competing for the business.
Research Protocol is built to be that layer. A sell-side firm connects its library once and it is available, under its own entitlements, to every connected buy-side reader and AI agent. A buy-side firm connects once and queries across every library it is entitled to read.
What only a neutral hub can do
The strongest argument for a neutral hub is not convenience. It is that some questions can only be answered from the middle.
You can ask a bank's own API about that bank's research. You cannot ask it what other brokers think about the same stock. A neutral hub can.
On Research Protocol that shows up as cross-library intelligence.
- Per-broker forecasts in each broker's own line-item structure, so a reader sees the numbers as the analyst built them rather than squeezed into someone else's template.
- Cross-broker consensus, with the median, the interquartile range, flags for outlying forecasts and the distribution of ratings.
- Consensus history, so a reader can see how the view on a company has moved over time, not just where it sits today.
For a fund manager, or an AI agent working for one, that is the difference between reading one opinion and seeing where it sits among the others. For a provider, it means its forecasts are compared on its own terms and in context, within the entitlements it has set.
None of this works if one participant owns the hub. A bank would not want a rival's platform computing consensus from its numbers. A reader would not trust consensus produced by a firm with a view to sell. Neutrality is what makes the data usable.
What neutral has to mean in practice
Saying "neutral" is easy. We think it has to mean a few specific things.
- Providers keep control of their content. Research can be hosted encrypted in UK South Azure storage, or stay on the provider's own servers and be served by signed redirect. Either way the provider decides.
- Providers set the entitlements. Who sees what is decided by the provider, per organisation and per user, by sector, company, analyst or entitlement code. Every read is checked. AI agents only see what their organisation is entitled to.
- The hub should carry research, not compete with it. Its job is to move other people's research to entitled readers, not to publish its own against the firms that use it.
- The hub should be owned by nobody's sales desk. No provider's notes should be ranked above another's for commercial reasons, and readers should not be steered towards anyone's products.
- Readership goes back to the provider. Providers see who read their work, across channels, with reads by people and by AI agents shown separately.
Why now
Two things make this more pressing than it was.
The first is fragmentation. Since MiFID II unbundling, research distribution has spread across many platforms, each with its own integration. Providers maintain a growing list of connections, and readers juggle a growing list of logins.
The second is AI agents. They are becoming a new class of research reader, and they need machine-readable, entitlement-aware access rather than PDFs in inboxes. An agent wired to one bank's API sees one bank. An agent wired to a neutral hub sees everything its firm is entitled to, through one connection. That is why the hub offers an MCP server with 40 tools alongside a REST API with more than 200 endpoints and the rpt command-line tool. Details are on the MCP page and the API page.
The longer the industry goes without a shared rail, the more each new reader type multiplies the integration work. Agents make that cost much more visible.
The rail stays the rail
We do not think every part of research distribution should move to one place. Providers will keep their own clients, their own relationships and, often, their own branded portals, which can run on the same hub on their own subdomain. Readers will keep their preferred tools. Platforms will keep their place, and a provider can publish to them from the hub in each destination's format.
What changes is the layer underneath: one connection for each side, one set of entitlements per provider, and a place where cross-broker questions finally have an answer.
If you are a provider or a buy-side firm and want to talk about connecting, email [email protected]. More on how the hub works is on the home page.