Evidence-backed
Credentials
Human-reviewed artifacts, public verification, actively maintained status, and revocation trails make capability inspectable.
Agency Script OS: 521 skills and five execution modes, one click from your own model
Agency Script
OS
Most AI tools answer. Agency Script runs the work: robust skills you chain into workflows, hand to supervised loops, and wire to fire on the right event, governed and auditable the whole way. Builders, CRM, automation, and Studio, with 521 skills, five execution modes, the Vault, and training built in. Point your own model at it with the OS Plugin, or run it all in our platform.
- 521 skills, five execution modes
- First-party builders + CRM
- Run in our platform or your model
SOC 2 Control Map
Audit-readiness posture
Actively Maintained
Live-checked at every verification
Human-Reviewed
No auto-pass algorithms
Revocable
Standards enforced
Product positioning proof
Agency Script OS, not a course platform
The Agency Script public platform connects credentials, capability graph, governed automation, buyer proof, and commercial rails into one operating layer. Public claims are labeled so users can see what is evidence-backed, modeled, or target-state.
Evidence-backed
Capability graph
Paths connect roles, checkpoints, rubrics, reviewer signals, and buyer-visible proof of what someone can actually do.
Target-state
Governed automation
Workflow adapters, webhooks, policy gates, and audit trails are positioned as governed automation rails as each integration clears launch evidence.
Modeled
Buyer proof
ROI, readiness, benchmark, and proof surfaces stay modeled or aggregate-only until sourced customer evidence is approved.
Target-state
Commercial rails
Pricing, packs, team plans, and procurement paths form commercial rails with capacity and evidence gates instead of open-ended promises.
Claim safety: Evidence-backed means live product surfaces or source artifacts exist. Modeled means a clearly labeled forecast, benchmark, or aggregate model. Target-state means the roadmap is declared before evidence is available.
Built for teams across operating contexts
AI Agencies
Tech Companies
Regulated Enterprises
Startups
Government
Healthcare
Enterprise Teams
Consultancies
AI Agencies
Tech Companies
Regulated Enterprises
Startups
Government
Healthcare
Enterprise Teams
Consultancies
THE DIFFERENCE
Most AI tools answer.
Agency Script runs the work.
Robustly, as a workflow, in a loop, on a trigger, governed and auditable the whole way.
Robust
Deep capability, not a thin prompt
Every skill is a complete, governed system held to a binary 10/10 bar: reliable, on-brand work, not generic AI filler.
Workflow-able
It runs the whole job, not one step
Chain skills into repeatable, named workflows: one input becomes a finished deliverable.
Loop-able
It works while you sleep, supervised
Hand work to autonomous loops with a verifier and a human gate: autonomy that compounds, never runs away.
Hook-able
The right work fires on the right event
Wire skills to triggers (a signup, a schedule, a threshold) so the system acts the moment something happens.
Built this way from day one, not bolted on later.
See the OS in action →EXECUTION MODES
Five modes. One word each.
A mode changes how a job runs, never what it is allowed to do. Every gate that binds the underlying skill still binds inside a mode: a hundred agents have exactly the authority of one.
ultraThe flagship
The full protocol. Question the requirement and write the 10/10 bar before any work exists, split the job into independent specialist lanes, disclose the cost, fan out, verify adversarially, measure against the rubric, and stop on dry, budget, or green, saying which.
swarmBreadth
Discovery that misses nothing. Many finders, each with a different search lens, deduped against everything already seen. It stops when two consecutive rounds surface nothing new, never at a round number, because a counter always leaves the tail unexplored.
arenaDepth
One artifact, championship quality. Competing attempts from genuinely different strategies (different bets, not drafts of one idea), scored by an independent judge panel, then synthesized: winner as the spine, the best of the runners-up grafted in, every graft noted.
proofVerification
Attack something that already exists. Every claim refuted or confirmed with evidence, every number re-measured, every "it works" exercised. Proof never fixes. It reports, so the verdict stays untainted by the wish to have been right.
shipThe last mile
Turn done into delivered. Run every gate the work declares, self-score against the rubric, package it with paths and handoff notes, and state plainly what is live, what is owner-gated, and what was deliberately cut.
Say the word from any surface running the OS Plugin, like "ultra: rebuild the offer page", and the mode takes it from there. Modes compose, and the budget stays disclosed.
RUN THE AGENCY
Training plus the operating system
Agency Script runs the work you do for clients: builders, CRM, automation, and Studio in one platform, driven from your own model through the OS Plugin and agents.
Website & funnel builder
Design sites and funnels in a visual builder, capture leads, and publish on your own domain.
Native CRM
Contacts, companies, deals, and pipeline, with bidirectional Salesforce and HubSpot sync.
Automation & workflows
Triggers, webhooks, and multi-step workflows to run the operational busywork for you.
Studio
Record, auto-clip, and produce branded video content from a transcript-first editor.
THE PLATFORM
Everything Inside Agency Script
Read the OS. Run the OS. Prove the OS. Scale the OS. Every module below is live in the platform.
Interactive Labs
Hands-on RAG and agent builders with real datasets. Force the tradeoffs you face in real client work, without multiple-choice shortcuts.
AI Tutor
Adaptive 1:1 guidance across your path. It explains, reviews, and pushes you, grounded in the canon.
The 5 Scripts
Discovery, Architecture, Delivery, Optimization, Scale: one-click tools that run a real client engagement end to end.
Capability Graph
Map and verify what you can actually do: a living model of proven, evidenced capability rather than a list of badges.
The Vault
Governed templates, scoping frameworks, and every artifact you produce: your deliverables, organized and reusable.
Certification & Credentials
Human-reviewed against calibrated rubrics. Expiring, revocable, and independently verifiable via a public endpoint.
Cohorts & Community
Cohort-based programs, events, mentorship, challenges, and a growing alumni network with shared standards.
Enterprise Suite
AI policy, incident playbooks, a regulatory sandbox, maturity assessment, ROI modeling, training credits, and contracts.
Analytics & Automation
Performance dashboards, automation workflows, leaderboards, and replays so you can see results and put delivery on rails.
AGENCY SCRIPT OS
Run your agency from your LLM
Connect your model to the Agency Script MCP and operate the whole business (builders, CRM, automation, sites, and workflows) from any surface. Portable across your tools. The engines stay on our server; you keep the operation.
The OS Plugin ships 521 governed skills served live, plus the five execution modes (ultra, swarm, arena, proof, ship) that raise how hard any of them runs.
- Claude
- Claude Code
- ChatGPT
- Codex CLI
- Cursor
- Gemini CLI
- Perplexity
- VS Code
- GitHub Copilot
- Windsurf
- Terminal
…and any MCP-capable client. One server, every surface.
Also runs as SaaS, mobile, desktop, SDK, and API. CLI coming soon.
35+ courses included
Learn at Your Own Pace
From AI fundamentals to advanced engineering. Short, focused courses built to produce real competency you can demonstrate.
AI FundamentalsBeginner
Introduction to AI
Core AI concepts, how large language models work, and why AI changes everything for agencies.
5 lessons45 min
AI StrategyIntermediate
Context Engineering for AI
Design the information environment around AI systems to get reliable, high-quality outputs.
5 lessons55 min
Prompt EngineeringBeginner
Prompt Engineering Fundamentals
Systematic approaches to writing prompts that deliver consistent, professional results.
5 lessons40 min
AI EngineeringAdvanced
AI-Assisted Development
Pair-program with AI coding assistants and build production-ready AI-powered applications.
6 lessons65 min
AI StrategyIntermediate
Agentic Thinking
Design AI workflows that observe, decide, act, and self-correct with minimal supervision.
5 lessons50 min
LeadershipBeginner
Agency Leadership in the AI Era
Lead confidently when the technology changes faster than any plan you can write.
4 lessons35 min
THE COMMUNITY
You do not build this alone
The platform runs the work. The community keeps the bar high: cohorts, live challenges, mentorship, and an alumni network that shares one standard of proof.
Cohorts & mentorship
Cohort-based programs and 1:1 mentorship, so you build alongside operators working the same problems on the same standards.
Challenges & events
Live build challenges and events where the deliverable is real work, reviewed against the same rubrics the certification uses.
Shared standards
An alumni network that holds one public bar: proven capability, evidenced deliverables, and credentials that stay revocable.
COMPARISON
Built for proof,
not completion
How Agency Script differs from typical vendor certifications and course completions.
Feature
Agency Script
Typical Certifications
Assessment Type
Artifact-based + scenarios
Multiple choice
Review Process
Human evaluators
Auto-graded
Credential Validity
Actively maintained (live-checked)
Lifetime (static)
Revocation Policy
Active enforcement
None
Public Verification
Unique URL per credential
PDF certificate
What It Tests
Judgment under constraint
Knowledge recall
Pass Rate Target
60-75% target band
90%+ (maximized)
Audit Trail
Immutable logs
None
GOVERNANCE
Standards Over Scale
We monitor what matters. The launch target is a calibrated 60-75% pass-rate band. If rates exceed 85%, standards may be drifting. If below 50%, assessment design needs review.
Monthly reviewer calibration exercises
Inter-rater reliability monitoring (Cohen's kappa > 0.7)
Documented dispute resolution process
Public change logs for all policy updates
60-75%
Target Pass-Rate Band
>0.7
Inter-rater Reliability
24mo
Credential Validity
5d
Review Target
100%
Human Review
Calibration targets by design, not current performance.
THE PROBLEM
Everyone Claims AI Expertise
Few Can Prove It
The AI agency space has a trust problem. Most credentials measure familiarity, not judgment under real constraints.
Proof
buyers need verifiable AI-operations evidence
Credentials Without Capability
Vendor certifications validate tool familiarity, not operational judgment. Enterprises can't distinguish qualified operators.
Practice
real work needs more than completion signals
Consumption Over Execution
Course platforms measure completion rates, not demonstrated ability. Videos don't build judgment for real constraints.
Standards
AI agency work needs public verification rails
No Institutional Standards
The absence of professional standards allows unqualified operators to damage client trust and undermine credibility.
CERTIFICATION
Get Certified on Building With It
Certification proves you can build with AI and the Agency Script OS: judgment under constraint, shown in real deliverables, beyond knowledge recall.
Artifact-Based Assessment
Real work products (offer specs, SOPs, system blueprints) evaluated against structured rubrics. No multiple-choice shortcuts.
Calibrated Human Review
Every submission reviewed by trained evaluators. Monthly calibration exercises. Inter-rater reliability monitored.
Actively Maintained
Credential status is live-checked at every verification. Staying active takes an active account and a day of verified training a year, so the credential always reflects current practice.
Public Verification
Every credential has a verification URL. Employers and clients verify status without accessing private data.
Revocation Authority
Certifications can be revoked for fraud, ethics violations, or misrepresentation. Standards are enforced, not aspirational.
Immutable Audit Trails
Every action logged. Submission history, review decisions, credential changes. Enterprise-oriented governance.
HOW IT WORKS
Your Path to
Certification
01
01
Choose Your Path
Start with any component (Foundation, Operator, Architect, or Principal) in any order. Hold all four to earn the Master credential (AS-M).
02
02
Complete Path Requirements
Work through structured content, submit checkpoint artifacts, and receive feedback from reviewers.
03
03
Portfolio Assessment
Submit your artifact portfolio. Real deliverables evaluated against structured rubrics by calibrated reviewers.
04
04
Scenario Examination
Written scenarios test judgment under constraint. Higher tiers include proctored live evaluations.
05
05
Earn Your Credential
Upon passing, receive your verifiable credential with a public verification URL. Status stays active while your account and annual training are current.
AGENCY SCRIPT CERTIFIED
Four Components.
One Master.
Four independent components. Earn one, several, or all four in any order. Hold all four active and pass the capstone to earn the Master credential (AS-M), required for Partner status.
Founders: $2,500 for all four → Master (AS-M)(Retail: $10,000)
LEVEL IAS-F
Foundation
Validates fundamentals: ICP selection, offer design, basic delivery frameworks, governance awareness.
- 3-5 artifact portfolio
- Written scenarios
- Human review
30-day path
$2,500
Most Popular
LEVEL IIAS-O
Operator
End-to-end operations: client lifecycle, delivery governance, team coordination, quality assurance.
- 5-8 artifact portfolio
- Complex scenarios
- Proctored walkthrough
60-day path
$2,500
LEVEL IIIAS-A
Architect
System design at scale: multi-client architecture, risk frameworks, strategic planning.
- Architecture portfolio
- Panel defense
- System design review
90-day path
$2,500
LEVEL IVAS-P
Principal
Institutional leadership: building, governing, and scaling an AI agency as an organization.
- Leadership portfolio
- Panel defense
- Organizational review
120-day path
$2,500
LAUNCH PROOF
Evidence-Gated
By Design
Agency Script publishes proof only when it is backed by product data, signed approval, or a documented launch-readiness artifact.
4
Certification tiers mapped
100%
Human-reviewed assessments
24mo
Active credential window
60-75%
Target pass-rate band
No fake proof
Public traction claims require source evidence before publication.
No blind trust
Compliance claims stay control-mapped until formal reports exist.
No provider drift
External provider claims publish only when configured and evidenced.
Every public claim passes a gate
Customer proof requiredProvider readiness verifiedCompliance evidence mappedOutcome metrics sourcedCase studies approved
WHY NOW
The market is moving fast
Industry and analyst estimates of the AI-skills market. Directional context, not Agency Script metrics.
$0B+
AI Education Market by 2030
0%
Enterprises with AI Skills Gap
0%
AI Training YoY Growth
0%
AI Skills Wage Premium
FAQ
Frequently Asked
Questions
Agency Script OS is the operating system for agency work: a website and funnel builder, native CRM, automation and workflows, the Vault, training, and community in one platform. You can run it in our app, or point your own model at it with the OS Plugin, which puts 521 skills directly in your AI client. Certification sits on top as proof you can build with it.
A mode changes how a job runs, never what it is allowed to do. ultra runs the full protocol with specialist lanes and adversarial verification, swarm is discovery that stops only when nothing new surfaces, arena builds one artifact from competing attempts scored by a judge panel, proof attacks existing work and reports with evidence, and ship turns done into delivered. Every gate that binds a skill still binds inside a mode.
Agency Script certifies judgment under constraint through artifact-based assessment and human review. Unlike vendor certifications that test tool familiarity with multiple choice questions, we evaluate real work products against structured rubrics. Credential status is actively maintained: it stays valid with an active account and annual ongoing training, is live-checked at every public verification, and can be revoked for integrity violations.
Platform ($99/month) gives you access to learning content and community. Certification is four independent components (AS-F/O/A/P) at $2,500 each, earned in any order with structured testing, live Q&A, and human review. Hold all four and pass the capstone to earn the Master credential (AS-M), required for Partner status. Platform and certification are separate purchases.
Certification timelines vary by component. Foundation requires a minimum 30-day path, Operator 60 days, Architect 90 days, and Principal 120 days, and you can take them in any order. These minimums ensure adequate time for learning, artifact creation, and review cycles. Candidates should plan for review and revision cycles beyond the minimum duration.
Failed assessments include detailed feedback from reviewers. You can resubmit after addressing the feedback, typically within 30 days. Each level has a maximum of 3 attempts within a 12-month period. If all attempts are exhausted, you must wait before trying again.
Agency Script certifications are designed to be easy for employers and buyers to verify. Each credential includes a public verification URL that can confirm status, validity dates, and certification scope without accessing private data. Employer-recognition claims publish only after dated source artifacts and rights approval exist.
Launch ($5,000) is a 90-day cohort program for people ready to start their AI agency. It includes live sessions, peer accountability, and templates. Launch Premium ($12,500) adds done-for-you deliverables and 1:1 coaching. Both are separate from certification.
Ready to
Run the Work?
Point your own model at the OS and run 521 skills in five execution modes, or run it all in our platform. Learn as you build, and get certified on building with it.
Human-reviewed certificationPublic credential verificationEvidence-backed claims policySecure checkout via Stripe
Still have questions?