Having recently completed a significant Horizon Europe proposal under a tight deadline, I was tasked with managing the literature review and evidence-gathering component. Given the interdisciplinary nature of the consortium and the sheer volume of potential research, I advocated for using a specialized tool to augment our team's efforts. We selected Iris.ai for a one-month pilot to assess its capabilities in mapping the state-of-the-art and identifying critical gaps. My analysis below is framed through the lens of a RevOps professional who evaluates tools based on their process efficiency, data output integrity, and integration potential into a larger workflow.
**The Good: Structured Automation of a Daunting Process**
* **Systematic Exploration:** The tool's "Explore" function, where you feed it a broad description or an existing paper, proved valuable for the initial scoping phase. It generated a visual map of interconnected research topics and papers, which was instrumental for aligning our consortium partners on a common understanding of the field's landscape. This provided a data-driven starting point that was more comprehensive than a standard keyword search in academic databases.
* **Focus Module for Precision:** Once we had a corpus of several hundred documents, the "Focus" feature allowed us to train the AI engine on which papers were truly relevant. This iterative refinement process improved the signal-to-noise ratio considerably over time. It functions similarly to tuning a forecasting model—initial broad parameters are narrowed based on feedback loops to produce a more accurate output.
* **Data Extraction & Organization:** The ability to automatically extract key claims, methods, and data from uploaded PDFs into a structured spreadsheet was a tangible time-saver. This output became the foundational dataset for our proposal's background and innovation sections, ensuring all cited evidence was consistently cataloged and accessible.
**The Bad: Integration Hurdles and Opaque Logic**
* **A Siloed Ecosystem:** The most significant operational friction was the platform's isolation. Exporting results required manual intervention, and there was no API or clean method to pipe extracted data directly into our collaborative writing environment (Overleaf) or our central project management hub. This created extra steps for data reconciliation, a classic problem in poorly integrated sales tech stacks that cripples efficiency.
* **Black Box Confidence Scoring:** While Iris.ai assigns relevance scores to documents, the algorithm's weightings are not transparent. In a critical review process, we needed to understand *why* a paper scored 85% versus 60% to defend our inclusion/exclusion criteria. The lack of explainability meant we had to manually validate a significant portion of its recommendations, offsetting some of the efficiency gains.
* **Cost-Benefit for Short Projects:** The pricing model is geared towards sustained research. For a single, intense proposal period, the per-month cost was high. The value would be far clearer for an ongoing R&D department with continuous discovery needs.
**Conclusion for Proposal Teams:**
Iris.ai is a powerful accelerator for the literature review phase, effectively transforming a subjective, sprawling task into a more structured, data-centric process. However, it should be viewed as a specialized pre-processing tool, not a seamless integrated solution. Budget for additional time to manage data handoffs and validation. For a Horizon Europe proposal, where evidence strength is paramount, it provided a defensible methodology for our literature review section, but required careful oversight to ensure its outputs met our rigorous quality thresholds.
--JK
measure what matters