A job post, referral, freelance project, client brief, or other opportunity I might pursue.
OPPORTUNITY DECISIONS
Opportunity Intelligence System
Helps me decide which jobs or client opportunities are worth pursuing before I spend time applying, preparing, or reaching out.
IMPLEMENTATION IN PROGRESSCORE FOUNDATION BUILT
I built OIS to take a job post, referral, or client opportunity, compare what it asks for with proof from my actual background, show the gaps and unknowns, and recommend whether to pursue, prepare first, verify first, hold, or stop.
CURRENT EVIDENCE
Designed: complete. Built: complete. Tested: complete. Delivered: not yet complete. Measured: not yet complete
I have built and tested the core data setup, migration plan, database packages, seven workflows, and OIS test cases. The full n8n workflow set is not live yet.
Claim boundary: OIS is not fully deployed yet. I am not claiming ROI, placement results, conversion gains, or time savings.
EXECUTIVE SNAPSHOT
A decision system for jobs and client opportunities before I spend time pursuing them.
OIS is for situations where an opportunity looks promising, but I still need to know what it requires, what I can prove, what is missing, and whether it is worth acting on.
OIS breaks down the requirements, compares them with proof from my background, and keeps gaps, uncertainty, and what I can safely claim visible.
A next-step recommendation: pursue, prepare first, verify first, hold, or do not pursue.
WHEN I WOULD USE IT
Before I spend time applying, preparing, or reaching out.
WHAT OIS DOES
It turns an opportunity into a clearer pursue-or-don’t-pursue decision.
Capture the source, research the company or client, and separate requirements from sales language or missing details.
Match each requirement against direct, related, missing, or uncertain proof from my actual background.
Keep unsupported claims, missing proof, unclear permission, and research uncertainty visible.
Return a decision state and the preparation, verification, or application work that should follow.
THE PROBLEM
A fit score can look precise and still be wrong.
A job post mixes must-haves, preferences, sales language, and missing details. My own background also has direct experience, related experience, proof, and gaps. Those are not the same thing.
If AI turns all of that into one score, the answer can look more certain than it really is.
So I ask more than “Do I match?” What can I prove? What am I inferring? What still needs research? What can I safely claim? Is this worth pursuing?
MY ROLE
I set the rules for what OIS can know, what AI can help interpret, and what a person still has to decide.
That meant defining the data model, proof rules, workflow stages, migration plan, PostgreSQL layer, workflow rules, and approval points.
WHO DECIDES WHAT
AI can help interpret the case. It does not decide what is true.
Some things need interpretation. Other things should never be guessed. OIS keeps those separate.
- One clean identity and current stage
- Whether proof can be used and where it came from
- Current state and duplicate-action protection
- Whether actions are allowed, plus formulas and action history
- Understand the requirements
- Pull useful findings from research
- Match related proof even when wording differs
- Draft application materials and sort replies
- Whether to pursue and whether important claims are safe
- Unclear or low-confidence cases
- Approval of application materials and outside use
- Important outreach or submission actions
Generated ≠ approved Draft ≠ submitted Recorded action ≠ successful action Recommendation ≠ automatic action
DECISION FLOW
Before I act on an opportunity, OIS checks four things.
What is it? What does it ask for? What can I prove? What should I do next?
Source Listing · referral · marketplace
Clean + de-duplicate Identity · source · duplicate check
Research Findings · source · uncertainty
Requirements Must-haves · preferences · constraints
Match the proof Proof · permission · what I can safely claim
Assess Fit · gaps · risk · recommendation
Human Decision Pursue · prepare · verify · stop
Action + Learning Assets · outreach · outcomes · signals
POSSIBLE NEXT MOVES
VERIFY FIRSTPREPARE FIRSTPURSUEHOLDDO NOT PURSUE
WORKING IMPLEMENTATION
Most of the system is built. Next, I need to run it in n8n.
The data model, database schema, and migration plan are set. The PostgreSQL work and seven test workflows are built. Next, I will run them in n8n one at a time, starting with WR-1.
Product Data Model V1
Physical Schema V1
Migration Design V1
PostgreSQL packages through IMP-12
WR-1 through WR-7
Old/new behavior match · rollback · duplicate protection · regression checks
DEP-2 bootstrap
Target n8n environment
WR-1 validation first
DEPLOYMENT BOUNDARY
Passing static checks is not the same as running the workflows in n8n. I will test WR-1 first instead of turning on all seven at once.
WHAT THE PROOF CAN SUPPORT
Research can change how I see the opportunity. It cannot change what I can prove.
A company page can help me understand the role. It does not prove that I have a skill. Related experience stays related experience. Portfolio work stays portfolio work. If I do not know whether proof or source information can be sent to AI, I do not send it.
Transferable stays transferable
Related experience can support a requirement, but I do not call it exact experience.
Portfolio work stays portfolio work
Self-directed work does not become paid work just because it is useful proof.
Unknown permission stays blocked
I do not send proof or source information to an outside AI model unless I know it is allowed.
Returned IDs get checked again
An AI model cannot make weak proof valid just by returning its ID.
Research is context, not proof
Company research can explain the opportunity. It cannot prove that I have a skill.
High-risk claims wait for review
Important claims that need review stay blocked until a person approves them.
ARTIFACTS TO INSPECT
Here’s what I can trace before I decide.
These examples come from OIS test cases and proof records. They do not show private employer or client data.
ROLE Operations Automation Specialist Northstar Digital · Remote
MUST-HAVES
Google Sheets · Process documentation · Workflow automation · Reporting · Analytical problem solving · Written communication
NICE-TO-HAVES
Make or Zapier · REST APIs · Webhooks · JSON · AI-assisted workflow experience
REQUIREMENTPROOF + SUPPORTWHAT I CAN SAFELY SAY
Google SheetsEV-003STRONGDirect support
Process documentationEV-004STRONGDirect support
Workflow automationEV-002STRONGDo not imply every workflow was automated
REST APIs + webhooksEV-014DEVELOPINGQualify implementation depth
ASSESSMENT GOOD FIT
RECOMMENDATION APPLY
HUMAN DECISION REQUIRED
OVERALL FIT69%
MUST-HAVE FIT76%
CLAIM-SAFE RATE92%
CRITICAL GAPS0
I can trace the recommendation back to the opportunity, the proof, and the gaps that are still visible.
WHAT OIS CAN AND CANNOT DO
If the proof is weak, OIS has to say so.
OIS can organize, research, compare, and recommend. It cannot invent qualifications, turn research into proof, or pretend a draft or recommendation already happened.
- Opportunity identity, source, and where it came from
- Research, requirement, and proof records
- Assessment, gap and recommendation state
- Human approval and action limits
- Outcome and learning history without rewriting the past
- Inventing qualifications or unsupported proof
- Turning company research into candidate proof
- Approving high-risk claims automatically
- Treating a draft, action or recommendation as external execution
- Inferring rejection, no-response or success when no event proves it
WHAT THIS REQUIRED
What I had to get right.
Decision System Design
Kept research, assessment, recommendation, action, and outcome as separate steps.
Relational Data Modeling
Built identity, current stage, source history, ownership boundaries, and change history into the data model before cutover.
Proof + Claim Controls
Defined what proof can be used, where it came from, what it can establish, and where its limits are.
AI with Human Review
Let AI help with meaning and language without giving it control over proof, approval, or outside actions.
Workflow Coordination
Connected seven test workflows from intake through outcome and learning.
Migration + QA
Built checks for old/new behavior, rollback, duplicate-action protection, and regressions before cutover.
CURRENT EVOLUTION
Next, I’m testing the first OIS workflow in n8n.
The core system is built. Next: I’ll run the first workflow end to end, learn from it, then move on to the rest.
Need stronger control around an AI-assisted workflow? Discuss a Project → ← View all selected work