Hebbia launches Headless Hebbia, an API and MCP server for firm knowledge

What Headless Hebbia is

According to Hebbia, firms hold knowledge across documents, emails, databases and internal systems, and even the best models need the right context from that material to do useful work. Getting that context into an application takes effort: teams must find relevant sources, understand how they relate, coordinate analysis across several steps and check the results.

Headless Hebbia is the company's answer to that problem. It consists of an API and an MCP server, which let Hebbia's retrieval, orchestration and analysis run inside bespoke applications that Hebbia builds with customers, or inside the AI assistants those customers already use.

Step-by-step retrieval and analysis

Hebbia says it works through a question step by step, retrieving and analyzing the information it needs along the way. Developers can test the workflow, and users can check the findings against the source material.

The company puts weight on the relationships between sources. A question about a company may depend on its subsidiaries, previous transactions or notes filed under a different name. Hebbia says it works with customers to map those relationships and to account for the firm's own terminology and classifications. According to the company, this helps retrieval find relevant material that a search for the company's name alone would miss.

Hebbia also says it supplies relevant evidence to each model call as the analysis progresses. This lets an application work across a large collection while using smaller context windows where appropriate.

Custom applications for banking and advisory teams

Hebbia gives two examples of what it can build with customers.

• Banking origination tool: an application that triggers on a newly announced deal and draws on research, market maps and internal notes to surface the right client to call, the reason to call now and the pitch to make. Every claim links to evidence the banker can review.
• Acquisition advisory: for a professional services team assessing the cost of integrating a target's technology, Hebbia can connect system records, contracts and workshop notes to identify consolidation opportunities and integration risks. Supporting evidence is available in the firm's own client portal.

The company says it works with each team to define the analysis, connect the relevant information and build the application around its process, including how the firm interprets its records and where people need to review results. Customers can use Hebbia's intelligence entirely through an interface built for them or by them.

Connecting Claude, ChatGPT and other assistants

Through MCP, assistants such as Claude and ChatGPT can call Hebbia to retrieve and analyze a firm's connected information, and they return answers with the supporting evidence. According to Hebbia, the choice of model and interface stays with the firm.

For firms, the pitch is that they can keep the AI tools their staff already use while giving those tools access to internal knowledge. Hebbia invites companies building an internal application, or wanting existing AI tools to work better with their knowledge, to contact it about what could be built together. The source does not give pricing or availability details.