November 17, 2026
•
XX
min read

Introducing Agent Fleet: Build Your AI Workforce for R&D

Rather than using AI for isolated research tasks, R&D teams can now configure agents with defined responsibilities, specialized expertise, persistent memory, and the ability to improve through feedback.

Agent Fleet introduces several new capabilities that bring us closer to a future where R&D professionals orchestrate teams of AI agents to execute complex research and intelligence workflows.

Build Agents Around Your R&D Workflows

Every R&D organization has its own technical priorities, research methodologies, and internal processes. Agent Fleet allows teams to configure agents around these requirements, giving each agent a defined role within their organization.

These agents can support a wide range of workflows, including:

  • Regulatory Intelligence: Monitor evolving chemical regulations, substance restrictions, and compliance requirements.
  • Material Scouting: Identify promising materials, ingredients, and chemical alternatives against technical and commercial specifications.
  • Formulation Intelligence: Investigate emerging formulations, ingredient combinations, and potential substitutes.
  • Technology Landscaping: Map emerging technologies, competitive approaches, and R&D white-space opportunities.
  • Competitive Intelligence: Track competitor research, patent activity, scientific publications, and commercialization signals.
  • Technology Commercialization: Evaluate potential applications, market opportunities, and commercial readiness.

Agents operate across Cypris's integrated intelligence ecosystem, connecting patents, scientific literature, chemical compounds, regulatory information, and commercial signals.

Agent Deliverables: From Research to Finished Work

Agents can now be configured to produce specific deliverables automatically as part of their assigned workflows.

Through the new Deliverables tab in the Agent Library, users can select which output formats an agent should generate when completing an assignment.

Supported deliverables include PowerPoint presentations, Excel spreadsheets, Word documents, PDFs, and interactive web pages that can function as custom dashboards.

For example, a Technology Landscape Agent could be configured to produce a PowerPoint presentation summarizing emerging technologies, while a Competitive Intelligence Agent could generate an interactive dashboard to present its findings.

These capabilities are also available directly in Cypris Q through the /create skill, allowing researchers to generate the same deliverable formats from their existing research sessions without needing to configure a dedicated agent.

Agent Improvement: An AI Workforce That Learns From Feedback

One of the most important components of Agent Fleet is the introduction of a structured Agent Improvement workflow.

Much like reviewing the performance of a team member, users can now evaluate an agent's previous assignments, identify what worked, and provide specific feedback on what should improve.

From the Agent Library, agent owners can review previous runs, access an agent's memory, and provide performance feedback that informs future assignments.

This feedback becomes part of the agent's accumulated knowledge, helping it better align with organizational requirements, research methodologies, and performance expectations over time.

Smarter Memory, More Relevant Context

We've also introduced significant improvements to how agents retain and recall information.

Rather than applying all stored memories to every new assignment, agents can now selectively draw on relevant prior context based on the task at hand.

This allows agents to incorporate useful experience, instructions, and preferences without unnecessarily introducing unrelated information into their work.

Combined with the new improvement workflow, these capabilities help organizations develop agents that become increasingly aligned with their specific needs.

Admin Review: Visibility Across Your Agent Workforce

As organizations deploy more agents, understanding how they're being used becomes increasingly important.

The new Agent Admin Review experience gives administrators centralized visibility into agent activity across their organization.

Administrators can see which agents are being used, who is running them, how their work is rated, what feedback they've received, and which deliverables they're producing.

Together with Agent Fleet's administration and governance capabilities, this provides a foundation for managing and scaling specialized agents across R&D teams.

Agents That Work Where You Work

Agent Fleet is designed to extend beyond a single application.

Agents can carry out assignments directly within Cypris or through supported connected environments, including Microsoft Copilot, Claude, electronic laboratory notebooks, and other enterprise systems.

Combined with configurable deliverables, this enables organizations to integrate specialized agents into existing research processes rather than requiring researchers to adopt entirely new workflows.

The Future of R&D Is an Orchestrated Workforce

We believe the next generation of R&D professionals will spend less time manually gathering and synthesizing information and more time directing, evaluating, and coordinating specialized AI agents.

Agent Fleet is a major step toward that future.

With configurable responsibilities, structured deliverables, persistent memory, performance feedback, and centralized oversight, organizations can begin building an AI workforce around their own expertise, standards, and workflows.

Explore Agent Fleet in your Cypris Agent Library today.

‍

Register here

Subscribe to receive the latest blog posts to your inbox every week.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

Keep Reading

November 17, 2026
•
XX
min read
Introducing Agent Fleet: Build Your AI Workforce for R&D
All
November 16, 2026
•
XX
min read
Smarter scientific literature search. More insightful research with Q.
All
October 15, 2026
•
XX
min read
Introducing Gemini 4 Argon
All