Sep 1 – Oct 1, 2026 · 5 items
The AI week, for a clinical research associate
A real edition, written for: Clinical Research Associate
The one thing
Which of these agent capabilities could cut your protocol deviation rate the most this quarter?
OpenAI Dots launch for persistent agents
OpenAI announced over twenty updates at DevDay 2026. It introduced Dots as always-on agents for ongoing tasks. GPT-6.1 Sol launched with near-Astra performance at one-fifth the token cost. Ultrafast mode offers up to eight times faster token generation in Codex. Private Intelligence now supports zero data retention with private safety processing. Codex can now run in the cloud or via a refreshed CLI with voice control. Code Review in ChatGPT desktop app enables automated and manual reviews. Codex Security Cloud scans repositories and prepares fixes in the cloud. Decisions API entered limited preview for real-time classification. Agents API now supports computer use for software interaction. Bedrock Managed Agents brings OpenAI agents to AWS. Plugin extensions allow sidebar integration in ChatGPT. Improved plugin creation and discovery tools were released.
Why it matters for you
Dots run 24/7 on tasks like monitoring trial data streams. That matches your need for continuous oversight.
- Always-on agents can watch for protocol deviations in real time
- You already use ChatGPT, and Dots extend what it can do
- Enterprise rollout means admin approval is the only gate
Try this
Ask your admin to enable Dots beta and set one to watch for data anomalies
Paste this into your AI tool
Context: I monitor clinical trial sites for protocol deviations and data anomalies Task: watch this data stream for values outside expected ranges and flag them immediately Constraints: - only use data from [trial ID] and [protocol version] - report each deviation with timestamp and value I'll paste: [sample data stream format] Output: a list of flagged deviations with context
Sources
IBM Pass^k metric for agent consistency
IBM Research introduced a new metric called Pass^k to measure agent consistency. It found a 24.4-point gap between average success and consistent success for a ReAct agent. The gap arises from flat decision distributions that flip under small perturbations. IBM built a Consistency Analyzer to detect flip-prone steps in agent trajectories. It generates consistency guidelines that reduce the gap from 24.4pp to 12.0pp. The guidelines improve Pass^5 by 16.0pp on same tasks and 13.0pp on similar tasks. Mean accuracy remains unchanged. The method works offline with one recorded trajectory per task. It requires no ground truth or model internals. Full details are in an arXiv technical report.
Why it matters for you
Your trial monitoring needs consistent results, not average ones. This metric targets that gap.
- Catches agents that flip decisions under small changes
- Works offline with recorded trajectories
- No ground truth or model internals needed
Try this
Run your last monitoring session through Pass^k to check consistency
Paste this into your AI tool
Context: I need to evaluate consistency of my trial monitoring agent Task: analyze this recorded agent trajectory for flip-prone steps Constraints: - use IBM's Pass^k metric - focus on protocol deviation checks I'll paste: [recorded agent trajectory] Output: list of flip-prone steps with suggested fixes
Perplexity Automations for long-running agents
Perplexity Computer launched Automations for long-running agents. Agents remember prior work and pick up where they left off. They run on a schedule or when events occur in Slack, Gmail, Outlook, Linear, or GitHub. Users set triggers, define tasks, and specify outputs like Slack or Gmail. Automations can flag missed updates or deadlines. Users create them via the omnibar or the Automations interface. Admins can limit connector permissions. Automations do not use credits while waiting for triggers. Existing Scheduled Tasks can be migrated to Automations. Automations are available to all Computer users.
Why it matters for you
Automations remember prior work and trigger on events. That fits your need for ongoing trial oversight.
- Agents pick up where they left off after interruptions
- Can trigger on new data or deadline alerts
- No credit cost while waiting for triggers
Try this
Set an automation to check trial data daily and alert on deviations
Paste this into your AI tool
Context: I need daily checks of clinical trial data against protocol Task: create an automation that runs daily and flags deviations Constraints: - only use data from [trial ID] - compare against [protocol version] I'll paste: [data schema] Output: automation definition ready to deploy
Cisco Talos CAIRN for AI malware detection
Cisco Talos built an open-source framework called CAIRN. It flags AI-integration traits in malware to classify and track samples. CAIRN already found a Windows hacking tool named CLOSEDQUORUM. CLOSEDQUORUM polls DeepSeek, Qwen, Mistral and Google Gemini for directives. The tool has no human input mechanism. It steals login credentials and cryptocurrency. Researchers linked it to cybercriminal forums from 2025. They could not confirm its real-world use. CAIRN also uncovered about 20 other AI-integrated malware examples. Researchers say AI is becoming operationalized for attackers.
Why it matters for you
Your trial data security matters. CAIRN helps detect AI-integrated threats that might target it.
- Flags AI-specific traits in malware
- Already found real-world examples
- Open-source and available now
Try this
Ask IT to evaluate CAIRN for monitoring trial data systems
Sources
Nous Research Hermes refactor with agents
Nous Research used Hermes Agent to refactor its own codebase. It deployed 1,393 subagents over nineteen active hours. The refactor cut non-test Python by 34.4%. Model spend totaled roughly $19,300. A manual effort was estimated at $150k to $1.8M. The largest file shrank from 34,847 to 5,512 lines. Workers split giant files and consolidated duplicate helpers. Reviewers caught regressions missed by existing tests. Lookup costs for agents fell sharply after the refactor. The team updated Hermes skills with lessons from the run.
Why it matters for you
Large codebases like trial systems benefit from agent-driven refactoring. This shows it works at scale.
- Cut non-test code by 34.4% in a real project
- 1,393 subagents coordinated the work
- Cost was ~$19,300 vs estimated $150k-$1.8M manually
Try this
Review your monitoring scripts for agent-assisted refactoring opportunities
Build this
Every week, one small thing to build with AI in something you actually care about. No work in it. Five minutes to set up, and worth keeping if it earns a second run.
A training plan for the nine people you already have time for this month
Five minutes to set up
Context: I need to create a training plan for my team Task: build a one-month training schedule using only existing team members Constraints: - use only [list of team members] - focus on [specific skill gap] - no new hires or external resources Output: a four-week training plan with assigned mentors and topics
- List the nine team members and their current skills
- Use ChatGPT to identify skill gaps from your last trial audit
- Assign each person to mentor or learn one specific protocol area
- Schedule weekly check-ins to review progress
Yours arrives Thursday.
This one was written for a clinical research associate. Tell us what you do and the next one is written for you — same news, your job, once a week.