A digital workforceyour C-suitecan govern.
Codi is a managed platform where specialised AI agents work as employees across marketing, sales, finance, and HR — governed, audited, and metered, and supervised by your own people at every step.
- 1Confirm new ETA with the supplier (Sysco contact for #1182: Lewis K.).
- 2If new ETA pushes past prep window, swap to backup inventory schedule.
- 3File a delay ticket with reason code DL-04 in ServiceNow within 30 minutes.
AI pilots don't fail on intelligence. They fail on control.
Four problems stall every corporate rollout — leaked data, runaway spend, unprovable audits, and siloed systems. Codi was built to close all four.
Sensitive data leaks.
Unauthorised tool usage and exfiltration blocked by inline PII redaction and prompt-injection defences.
Unpredictable token expenses.
Surprise API charges removed via hard spend caps, threshold alerting, and usage metered at list price plus 5%.
Compliance auditing hurdles.
Tamper-evident, hash-chained activity logs with POPIA and GDPR erasure tooling.
Disconnected enterprise systems.
Isolated sources such as Confluence connected with hierarchical, quality chunking algorithms.
Hire, supervise, meter.
Hire the agent
Configure a specialised digital employee in the agent studio, then test it in an isolated sandbox before publishing.
- Instructions, personality, knowledge, tools, voice
- Guardrail policy per agent
- Draft → Publish → Archive lifecycle
Supervise the work
Your people stay accountable. Agents answer in the streaming workspace, on WhatsApp, or by voice — every turn inspected and logged.
- Group-scoped visibility and access
- Cross-agent conversation hub with global search
- Mid-stream guardrail notifications
Meter the spend
Model consumption behaves like a utility bill with a breaker on it: metered at list price plus 5%, with hard caps.
- Admin spend caps enforced pre-call
- Alerts at 50%, 80%, 100%
- Turn-level traces for cost attribution
Agent Studio
Eleven configuration tabs to hire and train a digital employee, with optimistic locking so two admins never overwrite each other.
- MBTI-driven personality engine: 4 axes, 3 tone sliders, 9 presets
- “Improve with AI” prompt optimizer
- Admin-authored memory — up to 50 always-on facts
- Immutable versions, provenance and one-click rollback
Built like your org chart, not your tech stack.
Human leadership and the C-suite keep supervision at every level. Every agent reports to a real department head — access, spend, and output are attributed down to the person who owns them.
Brand & growth operations
- SEO agentsBrand-tuned personality
- Content agentsConfluence-grounded citations
- Campaign agentsBrand-tuned tone controls
Pipeline & enterprise revenue
- Lead-gen agentsSocial messaging + voice
- Proposal agentsCustom personality engine
P&L, audit & spend governance
- Budget agentsSpend caps per department
- Audit agentsTamper-evident activity log
Talent & organizational staffing
- Recruiting agentsPII redaction on intake
- Onboarding agentsNumber-scoped memory
Chief-in-command monitoring
Real-time agent health, guardrail intercepts, and turn execution traces across every department in a unified pane.
Quantifiable billing & allocation
Tokens and credits attributed down to department, user, and turn, on a tamper-evident audit trail.
Where the work lands
Web workspace, mobile iOS and Android, rich social messaging, and spoken voice — all through the same governed platform.
Role-based access
Supervisor, admin and super admin roles federated through OIDC SSO, with data resident in AWS eu-west-1.
Every seat at the table gets its own answer.
The same platform, argued five different ways — because the CFO, CISO, and CTO are not buying the same thing.
Operating leverage without headcount risk
- Scale your operations, not your payroll.
- Deploy an agentic workforce governed by human intent.
- Turn enterprise AI from an ungoverned risk into your strategic advantage.
Orchestration that survives an architecture review
- Enterprise-grade agent orchestration without vendor lock-in.
- Connect any model, enforce strict guardrails, and audit every token.
- Deploy localised AI agents with built-in MCP tools and zero model training on your data.
A number finance can forecast
- Predictable AI expenditure with hard spend caps and zero seat-licensing waste.
- Granular credit transparency down to the individual department, user, and turn.
- Eliminate shadow AI SaaS subscriptions with a single, controlled base platform.
Proof that holds up under examination
- Tamper-evident, hash-chained AI audit logs.
- Multi-layer PII redaction and real-time semantic guardrails that stop data leaks before they happen.
- Built for strict global compliance — complete tenant isolation, prompt-injection defence, and localised data residency.
Departmental staff that sound like the brand
- Equip your department leaders with bespoke, MBTI-tuned virtual staff.
- Instant access to enterprise knowledge bases without sacrificing brand integrity.
- Seamless voice, chat, and multi-modal execution tuned specifically for your team's operational flows.
One platform for the three jobs your team is doing already.
Wire in your sources. Trust the citations.
Sync a Confluence space in seconds. Codi chunks, embeds, and refreshes it on a schedule — and every answer cites the exact paragraph it came from.
- Confluence sync with hierarchical, quality-aware chunking
- Confluence permissions honored — users only see what they're allowed to
- Stale-source alerts when documents drift from the live policy
Every chat. Every tool call. Forever.
The audit log isn't an afterthought — it's the same primitive the agent runs on. Hash-chained, tamper-evident, and streamable to your SIEM the minute an event occurs with the Webhook Events add-on.
- Tamper-evident audit log · hash-chained to the byte
- Exportable to your SIEM or any webhook endpoint (Webhook Events add-on)
- PII redaction in transit with a configurable allowlist
| Time | Event | Actor | Result |
|---|---|---|---|
| 10:14:02 | agent.invoke | ms@northwind.com | ok |
| 10:14:03 | kb.retrieve · 3 src | agent:franch | ok |
| 10:14:05 | tool.call · refund | agent:franch | ok |
| 10:14:05 | policy.check | system | ok |
| 10:14:06 | pii.redact · email | system | redacted |
| 10:14:07 | audit.commit | system | ok |
Route by intent. Spend less by default.
Codi inspects each turn and routes to the cheapest model that can answer it. Set hard budgets per agent; cap your worst-case month before it happens.
- Real-time cost meter with projected month-end spend
- Per-agent budgets & per-team chargeback
- Routing recommendations from your own usage
Evidence, not assurances.
Every turn passes five inspection surfaces before a token reaches a model, and every event is hash-chained so a missing or altered log is immediately detectable in-app.
Tamper-evident audit log
Hash-chained log events with an in-app verifier that flags tampering or gaps across the whole workspace.
Custom regex & semantic AI judges
Pattern rules combined with parallel intent classifiers that intercept policy violations without perceptible latency.
Enterprise SSO
Self-serve OIDC federation with Okta, Entra ID and Google Workspace, domain enforcement and Just-In-Time (JIT) provisioning.
Residency, retention & erasure
European cloud hosting (AWS eu-west-1), isolated tenants, configurable retention, and self-service export or permanent erasure routines.
The controls your security team will ask about — and your board will expect.
SSO & JIT provisioning
Single sign-on with OIDC. New hires get access the moment a role is granted — no manual provisioning step.
Where your data lives
Hosted on AWS eu-west-1. We don't train models on your prompts or your data.
Tamper-evident logs
Tamper-evident, hash-chained event log for every chat and tool call. Streamable to your SIEM with the Webhook Events add-on.
Scoped permissions
Per-agent access controlled by department directors. Spend caps enforced per department, before the call is made.
Every workspace runs in AWS eu-west-1. Guardrails sit in front of every model call — PII redaction and prompt-injection defences aren't bolted on after the fact.
Seven virtual floors of platform.
Everything from agent lifecycle and personality tuning to rich social messaging and omnichannel routing ships in the same audited workspace.
Agent building & lifecycle
01- Eleven-tab configuration studio with optimistic locking
- Personality engine and prompt optimizer
- Immutable versioning, rollback, 30-day recovery
- Isolated pre-publication test sandbox
Knowledge base & RAG
02- Confluence connector with sync progress
- Hierarchical structural chunking
- Citations with relevance scoring
- Multi-tenant vector isolation
Voice
03- Multilingual text-to-speech on demand
- Bring your own ElevenLabs voice clones
- Real-time spoken mode
- Audio-in / audio-out API endpoints
Rich social messaging
04- Social & conversational gateway integration
- Rich text, media & interactive formatting
- Quick-reply buttons & action cards
- Channel-scoped memory + verified erasure
Guardrails
05- Multi-stage input / output / tool inspection
- Per-source PII redaction, 5 surfaces
- Regex plus semantic AI judges
- Retroactive history redaction
Security & infrastructure
06- Tamper-evident, hash-chained audit trail
- OIDC SSO with JIT provisioning
- POPIA and GDPR erasure automation
- AWS eu-west-1 residency, custom retention
Developer & observability
07- REST API with live interactive docs
- MCP servers per agent
- 21 webhook event scopes, HMAC signed
- Turn-level execution span trees
Virtual staffing tiers, not seat licenses.
- 1 internal + 1 customer-facing agent
- 2 supervisor seats
- 7-day data retention
- AI usage at retail token rates · voice $0.01/min
- 5 internal + 2 customer-facing agents
- 10 supervisor seats
- 30-day data retention
- AI usage at retail token rates · voice $0.01/min
- 20 internal + 10 customer-facing agents
- 30 supervisor seats
- 90-day data retention
- AI usage at retail token rates · voice $0.01/min