DistillerSR launches Smart Chat: an AI chat that only answers from your own validated evidence

DistillerSR has put a chat window inside the place where its customers already do their evidence work. Smart Chat, announced on October 27, 2026, brings the company's Agentic AI capabilities directly into the DistillerSR platform, so researchers can question their own literature reviews without exporting data to an outside chatbot.

The pitch: talk to your evidence, not the internet

The defining constraint is also the product's argument. Smart Chat works only with the data in a customer's own DistillerSR projects. It cannot reach the open web or any external source, which the company says dramatically reduces the risk of hallucination and keeps every answer tied to the organization's own validated evidence.

That matters because of who DistillerSR serves. The company describes itself as the pioneer and market leader in AI-enabled literature review automation, and says 80% of the top pharmaceutical and medical device companies trust it to produce transparent, audit-ready and regulatory-compliant literature reviews. For those customers, sending sensitive research data to an external large language model is not a minor policy question — it is the reason they have not already pasted their evidence into a chatbot.

DistillerSR chief executive Peter O'Blenis put it plainly in the announcement: many life sciences organizations cannot risk connecting external LLMs to their sensitive research data, and Smart Chat removes that barrier because it is built into the platform teams already use and works exclusively with their own projects.

What researchers can actually do with it

Smart Chat covers three kinds of work. Researchers can interrogate their data — asking questions, summarizing findings and comparing extracted data across projects. They can draft reports and other outputs, including summaries of findings, regulatory submission sections, tables and formatted content, which their experts then review and finalize. And they can configure protocols and workflows: building and editing forms, adding questions across many projects, and managing reference metadata and labels.

The through-line is that experts stay in the loop at every step. Smart Chat sits alongside Smart Screening and Smart Evidence Extraction, the company's existing tools for screening and extracting evidence; the three together are meant to cover the full arc from finding evidence to conversing with it, without handing the final judgment to the model.

No integration, no separate licence — and a deliberate alternative to MCP

Smart Chat is positioned as an alternative to DistillerSR's comprehensive Model Context Protocol (MCP) interface, aimed at organizations that lack the IT support or permissions to connect one. It needs no integration, no separate LLM licence and no setup, and it is included with every version of DistillerSR.

That trade-off is worth naming clearly. An MCP connection lets an outside agent reach into a system with its own tools and context; Smart Chat gives up that reach in exchange for a closed loop. The company is betting that for regulated evidence work, the closed loop is the feature: customer data is never used to train AI models, Smart Chat respects copyright obligations, and every action is governed by DistillerSR's existing roles and permissions system and is fully auditable.

What to watch

Smart Chat is available now, according to the announcement, so the test is no longer the pitch but the practice: whether answers grounded only in a customer's own projects stay useful when the evidence is thin, contradictory or spread across projects, and whether audit trails hold up under the regulatory scrutiny these reviews are built for. DistillerSR has not published pricing beyond inclusion in every version, nor independent accuracy figures — this launch is a company announcement, and the claims above are the company's, not independently verified results. Still, the design choice is a clear one, and it cuts against the grain of the chatbot era: in a market racing to connect models to everything, DistillerSR's new chat connects to less, on purpose.