Ontavia's six-stage gated workflows and dedicated specialist agents are designed to keep manuscript authoring tightly connected to source evidence at every step. Vasc AI Inc says the architecture addresses growing concerns about traceability and rigor in AI-assisted academic research.
Toronto, ON, Canada - September 23, 2026 - Vasc AI Inc today detailed the architectural principles behind Ontavia, its multi-agent research and writing platform, emphasizing how a structured gated workflow and purpose-built specialist agents work together to keep manuscript authoring tightly anchored to source evidence throughout the entire research process.
As AI-assisted writing tools become more prevalent in academic and biomedical settings, questions about traceability, source fidelity, and methodological transparency have grown more urgent. Ontavia's architecture is a direct response to those concerns. Rather than offering a single generative interface that produces text from broad prompts, the platform channels researchers through a deliberate sequence of stages, each designed to build upon verified outputs from the step before.
The platform is organized into two connected workspaces. The Research workspace focuses on evidence discovery and synthesis, while the Writer workspace supports the composition of manuscripts grounded in the evidence assembled during the research phase. Each workspace contains a six-stage gated workflow. Progression from one stage to the next requires that the outputs of the current stage meet defined completion criteria. Researchers can return to earlier stages at any time, but they cannot skip ahead, a constraint that enforces the sequential logic underlying rigorous academic inquiry.
Within these workflows, five specialist agents handle discrete tasks. The retrieval agent executes federated searches across PubMed, OpenAlex, ClinicalTrials.gov, and the European Nucleotide Archive, delivering consolidated results into a working corpus. The coding agent applies structured tagging frameworks so that sources can be categorized, compared, and filtered according to the researcher's analytical needs. The statistics agent supports quantitative synthesis tasks, assisting with data aggregation and summary computations when numerical evidence is involved. The review agent evaluates the coherence and completeness of assembled evidence and draft text, flagging gaps or inconsistencies. The writing agent assists with prose composition, helping researchers structure their arguments in formats aligned with academic publishing conventions.
A defining feature of the platform is that every output generated by these agents remains editable. Researchers can modify search queries, adjust coding categories, rewrite drafted passages, or override agent suggestions at any point. This editability is central to Ontavia's design philosophy. The platform treats its agents as specialized assistants operating under researcher direction, not as independent producers of finished work. The result is a collaborative dynamic in which the speed advantages of automation are preserved while the researcher retains full authority over intellectual content.
The gated workflow model also produces a natural audit trail. Because each stage generates documented outputs that feed into subsequent stages, researchers and their collaborators can reconstruct the logic and evidence chain behind any section of a finished manuscript. This traceability has particular value in fields where peer reviewers, editorial boards, and institutional oversight bodies increasingly expect transparency about the role AI tools play in the research process.
Ontavia is currently available to graduate students, physician scientists, and medical professionals through a web-based interface. Vasc AI Inc has stated that it will continue refining agent capabilities and expanding database integrations based on ongoing feedback from the research community. The company's roadmap includes enhanced customization of gated workflow stages to accommodate discipline-specific protocols, as well as expanded export options for direct submission to academic journals.
About Us
Vasc AI Inc is the developer of Ontavia, a multi-agent platform that connects literature discovery, evidence synthesis, and manuscript authoring through specialist agents and gated workflows. The platform serves graduate students, physician scientists, and medical professionals engaged in evidence-intensive academic and biomedical research.
CONTACT: https://ontavia.ai/
This release is published as issued by the organization named above, which stands behind its content.
