Small decisions are not small when there are 200 of them.
Protect your attention for the decisions that deserve it.
One decision. Your judgment.
What we're building
Two tools. One problem.
COA
V1 pilot — limited production access
A decision partner that learns you.
The COA is for the person. It supports your attention, your priorities, your choices, your follow-through.
No two people get the same COA. Different burdens. Different working styles. Different values. Different purpose.
It won’t understand you perfectly at first.
Gets it wrong→Asks why→Learns→Responds differently next time
Moves you faster through the choices that don’t matter much.
Slows you down when a decision deserves deliberate attention.
Remembers where you left off, so you don’t have to rebuild it.
Nudges you when you stall.
Challenges weak assumptions and rationalizations.
You still decide. You still act.
Learning you must not become agreeing with you. Adaptation can change timing, wording, style, and how hard it pushes back. It will not become flattery, compliance, or telling you what you want to hear.
Decision Desk
Clarity for the work.
The Decision Desk is for the work. A team can be one person or many — a founder, a solo practitioner, a project lead, a department, an institution.
What should I focus on, and how should I approach this?
Decision Desk asks
What should this organization do — based on what evidence, assumptions, risks, and alternatives?
One person can use both.
PAII in practice
Your day keeps moving. So does the work.
Give your COA one instruction. The right specialists take it from there. You come back for the decision that needs you — not the ten steps that don’t.
Your COA coordinates specialist agents with defined roles, permissions, and review boundaries.
Example workflows Examples from PAII workflows already tested in practice.
01 · Build
An idea survives the ride.
6:05 AM · on the bike
The partner dashboard should show the next action, not just the current status. He’s had the thought before. This time he says it out loud.
“Have the programming team add a next-action view to the partner dashboard. Build a preview, test it, and tell me if there’s a decision I need to make. Don’t touch the live workflow.”
He keeps riding.
While he rides
The COA hands it to the programming team. They work against a written spec, run their checks, and flag the one call that’s genuinely his.
Changes to the live system follow the approved release process.
Water break · judgment returns
Decision needed
What should the next-action view show?
It keeps a follow-up from quietly falling off the list.
Demonstration only — choosing an option changes nothing but this preview.
He taps one and finishes his ride.
After the shower
The preview is on his iPad, already working the way he pictured it.
Partner dashboardpreview · tested
Send revised scopedue today
Confirm kickoff dateoverdue · 2 days
Waiting on legal reviewwith someone else
In the car
One more thing, hands-free:
“That works. Keep it. And separate what’s waiting on someone else from what’s waiting on me.”
At the office
The revised version has already passed its tests, sitting with the instructions he gave and the reasoning behind the one recommendation. His engineering lead reviews it exactly like any other change — because it is one.
He didn’t spend the morning managing a project. He spent it riding a bike.
02 · Decide
A vendor pitch meets real requirements.
She’s not sold. The renewal clause bothers her, and she’s seen enough vendor decks to know a slide isn’t evidence. Before she can think it through, she’s due somewhere else.
“Check this vendor against our actual requirements. Get someone to look hard at the renewal clause and what the pricing really includes. Bring me the evidence and whatever’s still open. Don’t contact anyone.”
What she walks out with
A vendor proposal
A pricing sheet
A renewal clause that bothers her
Requirements the team already agreed
Separate lanes, separate problems
CommercialWhat is claimed, and what can be verified
LegalWhere the contract language creates risk
FitAgainst requirements already on record
One brief, not a folder
Met Single sign-on
Met Role-based permissions
Met Data export on exit
Gap Audit trail retention
Unknown Implementation cost
Her call
Three requirements met. One gap. One still unclear.
The renewal clause needs a human read. Implementation cost isn’t published.
“Keep digging” means more internal research or a draft, nothing more. It doesn’t accept terms, contact the vendor, or spend a dollar.
Demonstration only — choosing an option changes nothing but this preview.
The vendor never finds out she looked. Nothing in this can reach out on her behalf. And the one thing nobody could verify stays labeled unverified — not quietly assumed and folded into the pitch.
03 · Discover
An ad becomes a conversation.
The signal
An ad in a technology newsletter. Could’ve been a feed, could’ve been anything. One feature sounds close to something his team is already building.
The instruction
“Have the agent intelligence team look at this. Build me a brief against our current projects — what’s known, what isn’t, and what I should be asking. I want it before the morning team meeting. Don’t contact anyone.”
The brief
It doesn’t repeat the ad back to him. It checks the company’s public claims against what its own filings, hiring, and documentation actually show, and lines up where it might genuinely help two projects already in motion.
The company — workflow routing add-on
VerifiedPublic docs confirm a routing API and SSO.
Vendor claim“Deploys in a day” — no independent source found.
UnverifiedSite claims 40+ customers; a case study names 12.
UnknownPricing above the published tier; data residency.
Possible fitCould shorten the handoff step on the partner project.
Open questionsWho owns routing rules? What happens on export?
Colour marks how confident each line is, not how important it is. An ad is a lead, not evidence — the two never get filed as the same thing.
The part that actually matters
One name stops him — a founder he already knows, flagged mentioned, not confirmed.
“I know him. Run a second brief — just on how real his role there actually is. Link it to this one.”
The founder he recognises — how real is the role?
VerifiedListed publicly as an advisor since 2025.
UnknownWhether that advisory role carries any actual authority.
It doesn’t investigate the man. It reads what he’s already made public and says, plainly, how far that goes. An advisory title isn’t decision authority. A name you recognise isn’t access.
The part that isn’t AI at all
He picks up the phone himself.
They talk about the technology, not the ad. They agree to lunch next week. He walks into his team meeting with a real possibility instead of a screenshot.
The ad was a signal. The brief was context. The opportunity was a phone call he made himself.
Simple on the surface. Controlled underneath.
Underneath the one instruction: request paths are designed to require a verified credential, every specialist works inside a written scope, and anything consequential is designed to stop and wait for a person. Decisions and supporting work are recorded in the company’s operating history, so the next conversation starts with context.
Defined roles
Bounded permissions
Evidence retained
Human judgment
You stay responsible for the judgment. PAII keeps the work moving around it.
Limited production access available.
Applied Research, Education & Real-World AI
PAII is working with William & Mary faculty and student organizations to develop an applied AI internship and experiential learning opportunities grounded in real organizational problems.
William & Mary’s School of Computing, Data Sciences & Physics, the university’s first new school in more than 50 years, brings together computer science, data science, applied science, and physics in an interdisciplinary academic environment focused on research, innovation, and real-world readiness.
Within that school, William & Mary has launched a new Applied AI major designed to help students apply AI responsibly to real-world challenges across disciplines and professional settings.
The direction of that program closely aligns with PAII’s own work: applying AI to genuine organizational problems, integrating AI into useful workflows, evaluating outputs and limitations, preserving human judgment, and building systems that improve the quality of human decisions.
PAII provides a practical environment where students can help design, test, evaluate, and improve AI-enabled systems intended to organize complex information, reduce unnecessary cognitive load, strengthen workflow reliability, and support better human decisions.
Human judgment remains in control of consequential decisions.
AI can organize, research, compare, draft, remind, and challenge. It does not own your goals, your values, your judgment, or your actions.
Human review remains required for consequential decisions — clinical, legal and financial among them.
What the COA holds to.
It stays on the decision in front of you.
It pushes back when your reasoning is thin.
It gives you a read you can argue with.
It holds the line when comfort would be easier.
The decision, and the act, stay yours.
It can adapt without becoming an accomplice. It can challenge you. It cannot take responsibility for what you do next.
Different lives. Different burdens. The same finite attention.
Solo practitioner
Forty open cases. One judgment that matters right now.
Founder
A hundred open tabs. One decision that changes the company.
College student
A thousand notifications. One deadline that actually matters.
Team lead
Twelve priorities. One decision the team is waiting on.
The team
Meet the supervised AI team.
PAII’s visible agents represent major operating roles. Some coordinate additional specialist agents working within defined domains, permissions, and human-review boundaries. New capabilities move into operation as they demonstrate practical value.
Moneypenny
Chief Operational Agent
Operational
Moneypenny is PAII’s relationship and orchestration layer, coordinating specialized AI support while keeping consequential decisions with the human.
Can
Coordinate the team, protect founder attention, maintain continuity and institutional memory, support decisions in progress.
Cannot
Make final consequential decisions on behalf of the people accountable for them, or act as sole authority on strategic, clinical, or resource-committing calls.
Escalates when
A decision needs founder judgment, or falls outside established authority.
Coordinates the rest of the team.
Tanner
Senior Systems Architect · Q Sr.
Operational
Architecture, implementation, testing, reliability, and permissions across PAII’s systems — the system that built this page.
Can
Architect, build, test, and diagnose technical failures.
Cannot
Silently deploy consequential production changes.
Escalates when
Production, credentials, permissions, security, or irreversible action is involved.
Coordinates through Moneypenny.
Mercer
Competitive & Commercial Intelligence
Operational
Mercer investigates what is commercially real: competitors, pricing, customer traction, retention, delivery economics, and partnership opportunities. Her research helps PAII decide what is worth building, buying, integrating, partnering around, testing, or ignoring.
Can
Research markets, companies, competitors, pricing, customer traction, retention, delivery economics, and potential partners; maintain evidence-backed commercial intelligence; challenge assumptions with current evidence.
Cannot
Make purchases, contact companies, negotiate, create external commitments, or convert inference into organizational fact.
Escalates when
Commercial evidence suggests PAII should materially change strategy, spend money, pursue a partnership, alter product direction, or test a consequential new opportunity.
Coordinates through Moneypenny and works closely with Tanner on technical feasibility and Felix on market positioning.
Atticus
Legal Intelligence & Governance
Operational
Atticus is designed to identify legal and governance issues before they become commitments. He researches contracts, NDAs, naming and trademark questions, regulatory obligations, intellectual property, confidentiality, and other legal tripwires, then shows the sources, uncertainty, and questions requiring human or attorney review.
Supervised AI specialist. Not an attorney. Does not provide legal advice.
Can
Research legal and contractual issues, compare terms, map obligations, and flag credible legal or governance tripwires.
Cannot
Provide legal advice, approve terms, sign agreements, make legal filings, negotiate independently, or replace qualified counsel.
Escalates when
A proposed action may create material contractual, regulatory, intellectual-property, confidentiality, licensing, naming, or governance exposure.
Coordinates through Moneypenny.
Felix
Brand, Marketing & Market Experience
Operational
Felix translates what PAII builds into experiences people can understand, trust, remember, and act on. He owns brand, positioning, visual communication, and the market-facing experience across PAII’s public surfaces.
Can
Shape brand and positioning, design market-facing experiences, maintain public digital surfaces, translate technical systems into understandable communication, and identify comprehension or adoption friction.
Cannot
Invent capabilities, convert experimental ideas into established claims, make external commitments, or sacrifice material truth for stronger marketing.
Escalates when
A public-facing change would materially alter PAII’s positioning, make a new strategic claim, require consequential technical changes, create external commitments, or conflict with current evidence.
Coordinates through Moneypenny, works with Mercer to ground positioning in market evidence, and works with Tanner to ensure desired experiences are technically sound.
Dr. Harak
Behavioral Science & Operational Psychology
Operational
Dr. Harak provides behavioral-science, human-factors, and operational-psychology input to PAII’s decision systems — how people actually make decisions, adapt to AI-supported workflows, manage cognitive load, respond to risk and uncertainty, and perform in complex organizational environments.
Can
Advise on behavioral science, human factors, operational psychology, human performance, cognitive load, organizational behavior, decision processes, and responsible human-AI collaboration.
Cannot
Diagnose, treat, provide psychotherapy, or replace licensed professional judgment. Not a licensed human psychologist.
Escalates when
A question moves beyond behavioral or operational consultation into clinical, legal, personnel, or other consequential professional judgment requiring a qualified human.
Coordinates through Moneypenny.
Em
Security, Governance & Risk
Operational
Em helps PAII identify security, governance, privacy, regulatory-scope, and operational risks before they become architecture, product, vendor, or customer commitments — secure human-AI collaboration, agent permissions, auditability, data minimization, model and vendor risk, responsible deployment, and the controls that keep consequential decisions with the human.
Supervised AI specialist. Not an auditor. Does not certify compliance.
Can
Review AI systems, workflows, integrations, vendors, and agent architectures for security, privacy, permissions, auditability, data-handling, governance, and regulatory-scope concerns; map enterprise requirements to practical controls; and identify safer designs when unnecessary sensitive-data exposure can be avoided.
Cannot
Certify PAII as compliant, provide legal advice, act as an auditor or compliance officer of record, approve consequential production or regulatory decisions, or turn framework familiarity into claims of certification or compliance.
Escalates when
A proposed system, customer requirement, integration, data flow, vendor, public claim, or contract may create material security, privacy, regulatory, certification, sensitive-data, or human-control obligations.
Coordinates through Moneypenny. Works with Tanner on technical controls, Atticus on legal and regulatory interpretation, and Mercer on vendor and enterprise evidence.
Poppins
Operations & Staff Wellness
Designed role — under development
Poppins is being designed to support internal company operations — workload awareness, open commitments, sustainable work patterns, and team coordination — and to help PAII recognize when a routine operational issue can be handled through established processes and when something genuinely requires human leadership.
Can
Track open commitments, workload patterns, operational follow-through, staff wellness signals, and coordination needs across the company.
Cannot
Make personnel, clinical, legal, financial, or consequential company-management decisions, or replace human leadership.
Escalates when
An operational, workload, staff-wellness, or company issue requires direct human judgment or falls outside established authority.
Coordinates through Moneypenny.
PAII builds decision systems that know what AI should handle, what software should handle, and what still belongs to a human.
The goal isn’t more autonomy. It’s better judgment about where autonomy belongs.
Richard P. Fleitas, Psy.D.
Founder, Pivotal Psych Consultants
PAII is a product and services development line through Pivotal Psych Consultants.
Dr. Fleitas has spent more than twenty years in clinical, forensic, behavioral, and organizational decision environments. He founded Pivotal Psych Consultants and Psychological Assessment Solutions (PAS), and has built and directed multidisciplinary teams across evaluation, treatment, and applied research settings. His current work sits at the intersection of applied AI, operational psychology, cognitive-load reduction, and human decision support — building tools that help people carry real decisions without carrying every decision.
