JP
Joshua Peskay
Personal AI Cohort: Building a Thinking Partner That Challenges You
The problem this solves
Personal AI assistants that reinforce a user's biases instead of challenging them
Joshua Peskay runs a cohort program teaching people to build their own Personal AI assistant — a configured Claude instance intended as a working and thinking partner rather than a tool. The program centers on building metacognitive safeguards into the AI itself so it challenges rather than flatters the user.
Technique
Configure the AI to attack your ideas before you act
“A Red Team prompt that attacks your ideas before you act on them. (It can be as simple as: "Argue the strongest case against the position I just took. Do not reassure me.")”
— Solve Tuesday #9 - Your Personal AI Has a Security ...
Technique
Build a monthly audit: when did your AI last say you were wrong?
“A monthly disagreement audit: when was the last time your AI told you that you were wrong? ... If you can't remember, your AI is mirroring you, not helping you.”
— Solve Tuesday #9 - Your Personal AI Has a Security ...
AI adoptionPersonal AICritical thinkingAI safetyCybersecurity
Similar solutions
AI Ethics and Responsible AI ConsultingWebfriesThe problemBias and risk in AI adoptionResponsible AIAI ethicsAI governanceResponsible AI Governance and Bias ConsultingSparxIT SolutionsThe problemAlgorithmic bias and AI compliance riskTechniqueReal-time monitoring corrects algorithmic bias as it occursResponsible AIAI governanceBias mitigationAI Awareness Talk for All-Hands and CPD DaysSafeAIThe problemStaff using AI at work without any guidanceEU AI ActAI awarenessAI literacyBusiness Connectors: scoped, revocable source access for AI clientsPulse Business AIThe problemAI assistants reaching into data they shouldn't touchJudgementSource expansion is a named-reviewer decision, not an automatic oneAI governanceData accessSecurityAI Training for Workplace Tool Evaluation and Responsible UseSkill Builder SeminarsThe problemEmployees using AI tools without knowing which are safe or appropriateAI trainingAI governanceAI literacyRight-sized AI governance for UK SMEsLyticoreThe problemStaff using consumer AI tools informally without a usage policyNovel practiceMeet shadow AI with a policy, not a banAI governanceAI complianceData protection
Have a problem like this?
- Only matched specialists see it
- Free to send
- Proposals scoped to your brief
Appears in