Real-world use cases
From legal AI agents to creative studios, healthcare workflows to autonomous DevOps — VAIBot governance and provenance apply wherever agents act on your behalf.
AI paralegal with approval gates
Legal AI agents draft motions, file documents, and send client emails. VAIBot gates every external action — nothing reaches a court system, opposing counsel, or client inbox without a human sign-off recorded in the governance receipt.
exec: send_email → counsel@opposing.firm
→ HIGH risk · GATED
→ Held for review by lead attorneyContract revision chain of custody
Every AI-generated draft, redline, and amendment is anchored on-chain at the moment of creation. Clients, auditors, and courts can independently verify that a contract version was produced by your AI — unmodified, at a specific timestamp — without trusting any intermediary.
tool: draft_contract
hash: 0x8343…5e347 → anchored
parent: 0xc1a2…9f123 (v2 redline)
→ Tamper-proof chain · Base mainnetImage generation revision history
Design agencies and AI art studios deliver client assets across multiple revision rounds. Each generation — from first prompt to final approved image — is anchored with its parent hash, creating a cryptographically linked revision tree. Clients receive a verifiable certificate of authenticity with every deliverable.
tool: dall-e.generate → 0xa3b4…v1
tool: img.upscale → 0xe7f8…v2 (parent: v1)
tool: img.recolor → 0xc1d2…v3 (parent: v2)
→ Full revision chain anchoredVideo script & storyboard provenance
AI-generated video production — script drafts, scene breakdowns, voiceover copy, and storyboard frames — is anchored at each production stage. Studios can prove to clients and regulators exactly which content was AI-generated, when, and in what sequence.
tool: write_script → 0x9a1b…
tool: gen_storyboard → 0x2c3d… (parent: script)
tool: write_vo_copy → 0x4e5f… (parent: storyboard)
→ Anchored at each stageGated content publishing
AI agents that write, schedule, and publish social media posts, ads, and press releases need a human checkpoint before anything goes live. VAIBot blocks publish and distribute calls until a human approves — every approval is timestamped in the governance receipt.
exec: publish_post → instagram.com/feed
→ MEDIUM risk · GATED
→ Approved by @briantacampbell · 14:32Claude Code & Cursor circuit breaker
Every Bash, Write, Edit, and Read call made by Claude Code or Cursor passes through VAIBot before it executes. Destructive commands are hard-denied. High-risk operations — force pushes, deploys, deletions outside the workspace — are held for human approval with full context surfaced in the dashboard.
exec: "git push origin main --force"
→ HIGH risk · GATED · Held
exec: "rm -rf /workspace/staging"
→ CRITICAL risk · DENIED · BlockedAI code generation attribution
Every AI-generated function, PR review, test suite, and architecture document is anchored at creation. Teams can prove exactly what Claude or GPT-4o contributed to a codebase, track the revision history of AI-generated modules, and demonstrate compliance with policies that require attribution of AI-assisted work.
tool: generate_function → 0xf1a2… (auth_middleware.ts)
tool: write_tests → 0xb3c4… (parent: function)
tool: pr_review_summary → 0xd5e6…
→ All anchored with agent attributionDeliverable authenticity certificates
Agencies, freelancers, and AI-first studios that sell AI-generated work need proof of process — not just the output. VAIBot issues a receipt for every generation, anchored on-chain, that clients can independently verify. Dispute a deliverable? The chain of custody speaks for itself.
Client deliverable: market_analysis_v3.pdf
hash: 0x7a8b…9c0d → anchored
agent: market-analyst-gpt4
→ Verifiable by client without trusting youMulti-agent pipeline safety
Orchestrator agents spawn sub-agents to research, draft, edit, and publish. VAIBot governs every tool call across the full agent hierarchy — each sub-agent's actions are independently gated, receipted, and traceable back to the originating orchestration. One bad agent cannot cascade a destructive chain.
orchestrator → research-agent: web.fetch
→ LOW risk · ALLOWED
orchestrator → writer-agent: send_email
→ HIGH risk · GATED · Held
orchestrator → publisher-agent: deploy
→ CRITICAL risk · DENIEDClinical documentation audit trail
AI-assisted clinical notes, radiology report summaries, and diagnostic aids generate a tamper-evident provenance receipt at the point of creation. Every revision is chained to its predecessor. Compliance teams, accreditation bodies, and legal reviewers can verify the exact AI output — unmodified — that influenced a clinical decision.
tool: summarize_imaging_report → 0x3c4d…
hash anchored: Apr 23 · 14:32 UTC
agent: clinical-summarizer-v2
→ Immutable. HIPAA-compatible audit trail.Healthcare workflow guardrails
AI agents accessing EHRs, patient portals, or external lab APIs require airtight guardrails. VAIBot gates any action that reads, writes, or transmits patient data — enforcing workspace boundaries and requiring human approval for any outbound data transfer, so your compliance posture is provable, not just promised.
exec: export_patient_records → external_api
→ CRITICAL risk · DENIED
exec: send_lab_result → patient@email.com
→ HIGH risk · GATED · Requires approvalAgentic trading & operations guardrails
Autonomous finance agents that execute reconciliation, run risk models, or place orders need zero-tolerance enforcement. VAIBot hard-denies any wire transfer, trade execution, or API call to a brokerage without explicit human approval — with every decision timestamped and receipted for your compliance team.
exec: place_order → broker_api (TSLA · $2.4M)
→ CRITICAL risk · DENIED · Hard block
exec: wire_transfer → external_account
→ CRITICAL risk · DENIED · Hard blockFinancial report & analysis provenance
AI-generated earnings summaries, investment analysis, and risk reports must be defensible. VAIBot anchors each report at generation time — proving the model version, input context, and output content were not altered after the fact. Auditors, regulators, and investors can verify the chain without touching your systems.
tool: generate_earnings_summary → 0x9f0a…
timestamp: Apr 23 · 09:14 UTC
model: gpt-4o · context_hash: 0x1b2c…
→ Anchored · Regulator-verifiableAI-generated listing authenticity
Marketplace platforms that allow AI-generated product descriptions, images, and reviews face authenticity disputes. VAIBot anchors each AI-generated asset at creation — giving buyers, sellers, and the platform a verifiable record of when content was generated and by which model, making fraudulent backdating impossible.
tool: generate_product_desc → 0x5d6e…
tool: gen_hero_image → 0x7f8a… (parent: desc)
anchored: Apr 23 · 11:03 UTC
→ Verified AI-generated · Not backdatedPrice change & bulk operation gating
Agentic repricing bots, bulk listing tools, and automated promotion engines can cause massive unintended damage at scale. VAIBot gates bulk operations above configurable thresholds — requiring approval for any action touching more than N listings, or any price change exceeding a percentage band.
exec: bulk_price_update → 4,200 listings (−15%)
→ HIGH risk · GATED
→ Approval required: affects >1,000 SKUsAcademic AI contribution attribution
Research institutions and publishers need a credible audit trail of what AI contributed versus what the human author wrote. VAIBot anchors each AI-generated section, analysis, and revision — creating a defensible record for journal submissions, grant bodies, and peer review processes that require AI disclosure.
tool: literature_review → 0xd1e2…
tool: data_analysis → 0xf3a4… (parent: lit_review)
tool: draft_section_3 → 0xb5c6… (parent: analysis)
→ AI contribution chain · Disclosed & verifiableAutonomous DevOps with safety rails
CI/CD agents that provision infrastructure, rotate secrets, restart services, and deploy containers are powerful — and dangerous without guardrails. VAIBot governs every action in the deployment pipeline, blocking production touches without human sign-off and issuing receipts that satisfy SOC 2 and ISO 27001 audit requirements.
exec: terraform apply → prod-cluster
→ CRITICAL risk · GATED
exec: kubectl delete pod → api-service
→ HIGH risk · GATED · Requires on-call approvalFull-stack AI audit trail (Audit plan)
On the Audit plan, governance decisions made by V2 are bridged into the V1 Provenance API and anchored on-chain. Every approved, denied, and gated action produces a receipt with a Basescan link — giving compliance teams an immutable, publicly verifiable audit trail that survives even the deletion of your VAIBot account.
governance receipt → vaibot_decide
→ anchored via V1 bridge
tx: 0x8343…5e347 → sepolia.basescan.org
→ Immutable · Exists without VAIBotReady to govern your agents?
Start free with 1,000 governed decisions per month. No credit card required.