Know the true economics of your agentic AI — before you build it.
Agenomy is Cognizant's platform for scoping, pricing, and optimizing enterprise agent programs end to end. Model the cost to build and the cost to run with two proprietary engines — and walk into any CFO conversation with defensible numbers.
70–85%
Typical run-cost reduction
2 engines
Build Sutra + Run Sutra
CFO-ready
Defensible cost models
The Framework
Three engines, one economics story
Agenomy unifies build-side complexity scoring, run-side token economics, and platform industrialization into a single defensible model.
🏗️
Build Sutra · ACP
Scope & price the build
Agentic Complexity Points score each use case across orchestration, data, integration, and governance — then map directly to a commercial model, from T&M to Phased SOW.
💵
Run Sutra · TCP
Forecast & optimize the run
Token Complexity Profile establishes a production baseline from seven cost drivers, then models achievable savings across routing, caching, compression, and more.
⚙️
Agent Foundry Ops
Industrialize & govern
The observability and control plane — FinOps, guardrails, human-in-the-loop, agent registry, and eval suites — turns a pilot into a governed enterprise program.
User Guide
How to run an assessment
Follow the workflow left-to-right through the sidebar. Each step feeds the next; the dashboard rolls it all up into an executive summary.
1
Set Project Parameters
Define client, currency, blended rates, margin, and model rate references. These drive every downstream calculation.
2
Build the Use Case Portfolio
Add each agent or use case in scope with its volumetrics, architecture, and governance profile.
3
Profile TCP & Context
Set input/output token volumes and configure RAG chunk economics per use case to establish the run baseline.
4
Assemble the Full TCO Stack
Layer in infrastructure, ops, platform, and observability to build the complete run-cost breakdown.
5
Simulate & Compare Scenarios
Toggle optimization levers and compare OSS/Sovereign, Mid-Tier, and Frontier cost shapes side by side.
6
Present the Dashboard
Roll everything into a CFO-ready summary — TCO, savings, and payback — then export to CSV or save your state.
💡 Tip: A full in-app reference lives under Help → User Guide and Literature in the sidebar once you launch the platform.
Capabilities
What you can do inside
📋
Use Case Portfolio
Catalogue every agent with volumetrics, priority, and deployment phase.
🪙
TCP Token Profiling
Model input, output, and context token economics per use case.
⚡
Optimization Simulator
Toggle six Run Sutra levers and watch savings recompute live.
🔀
Scenario Comparison
Compare frugal, balanced, and lavish cost shapes side by side.
📈
Executive Dashboard
TCO, savings, and payback — packaged for the boardroom.
📚
Case Studies & Literature
Credentialing cases and framework references at your fingertips.
Ready to model your program?
Launch the platform and start with Project Parameters. Everything saves to your browser — nothing leaves your device.
Agenomy
Build Sutra · Run Sutra · Agent Foundry Ops
Cloud SQL: Agenomy
No project loaded
Project Parameters
Define the engagement scope. All inputs feed Build Sutra (ACP scoring) and Run Sutra (TCP model).
Client & Engagement
Commercial Rates
Model Rate Reference (per million tokens)
Cloud Selection
Define the cloud platform and the enterprise's approved model ecosystem. Enterprises typically approve several models rather than a single one — agents downstream select from this approved set.
☁️ The selected Cloud Provider drives the pricing catalog used in the Cloud Cost Model tab (Azure · AWS · GCP), whose monthly total flows into the Full TCO Stack. For Multi Cloud / On-Premise, pick a catalog to price against on that tab.
Cloud Environment
These are the models cleared for use across the program. Each agent must be assigned one of these models on the Token Complexity Profile screen.
Agent Details
Define the agent inventory. A single use case may contain multiple agents, and different agents can run on different approved models. Assign each agent to a use case and a model here.
Add Agent
Registered Agents
Use Case Portfolio
Define all agentic use cases in scope. Each becomes a line item in build and run cost calculations.
Accepts .xlsx / .xls / .csv. Recognized columns: Use Case Name, Description, Daily Active Users, Agent Calls per User / Day, Priority, Deployment Phase. Rows whose name matches an existing use case are skipped as duplicates.
📋
No use cases yet. Add your first agentic use case above.
Run Sutra — Token Complexity Profile (TCP)
Cognizant proprietary model. Forecast token volumes and monthly run cost per use case. Model tier, cache hit rate, and retry rate drive the TCP output.
Use Cases
Agents
Run Sutra — Context & RAG Economics
RAG, embedding, and retrieval cost modeling. Optimizing chunk count from 10–15 to 3–5 cuts retrieval cost by ~70% and reduces hallucination-driven retries.
Run Sutra — RAG cost driver: Quality plateaus at 3–5 retrieved chunks. Fetching 10–15 is wasteful — embedding, storage, and re-indexing cost compound with corpus size. Enable Precision RAG Optimization below to model the reduction.
Cloud Infrastructure Cost Model
Bottom-up cloud run-rate for the selected provider (set in Setup → Cloud Selection). Costs are driven by three consumption drivers — compute, transaction/API-call volume, and storage. The monthly total flows straight into the Full TCO Stack.
☁️ Driver model: Tool & API calls are not priced as standalone line items — they are inputs that drive transaction consumption, which in turn drives infrastructure. Every service below rolls up into one of three buckets: Infrastructure (compute), Transaction (API/tool-call volume) or Storage.
Cost Basis
Run is the steady-state month that feeds the TCO stack.
Run Sutra — Full Cost Stack (TCO)
Monthly operating cost across all 6 cost categories per use case. Token cost flows from TCP. RAG cost flows from Context & RAG. Remaining items are entered directly.
Portfolio Run Cost Summary
Run Sutra — Optimization Simulator
Toggle and tune Run Sutra optimization levers.
For estimation purposes each lever contributes up to 5%
portfolio savings, capped at a total 30% optimization benefit.
🚧 Under construction. Savings rates and lever logic are still being refined — treat figures on this page as illustrative, not final.
Optimization Levers
Savings Impact
Scenario Comparison
OSS / Sovereign / Mid-Tier / Frontier — 3-year projection across scenarios with different model mix and usage assumptions.
Map each use case to Agent Foundry capabilities. Shows which Composer / Ops (Monitor · Remediate · Govern) layers activate based on ACP axis scores.
🚧 Under construction. Capability activation is currently a placeholder mapping — treat this page as a preview of the concept, not a finished feature.
Agenomy in Action — Credentialing Cases
Reference benchmarks from live Cognizant engagements. Directional anchors for client conversations. Client Zero = Cognizant's own $4.2M AI budget.
User Guide — Run Sutra Economics Tool
This tool covers Run Sutra (TCP — token cost, optimization, scenarios) and Agent Foundry Ops activation. ACP scoring (Build Sutra) is handled in the dedicated Build Estimator tool.
What This Tool Does
This tool gives you a defensible, structured estimate of the total economics of an agentic AI program — both the cost to build and the ongoing cost to run. It uses two Cognizant proprietary models: Build Sutra (powered by ACP — Agentic Complexity Points) to scope and price the build, and Run Sutra (powered by TCP — Token Complexity Profile) to forecast and optimize the production cost.
Outputs are designed for executive conversations: CFO-ready cost tables, CTO-ready architecture decisions, and CPO-ready business case framing.
Recommended Workflow
Project Parameters — Set currency, blended dev rate, margin, FX
Use Case Portfolio — Add each agentic use case in scope
Agent Details — Add agents under each use case (one or many per use case) and assign an approved model
TCP Profile — Profile token volumes: input, output, context per use case
TCP (Token Complexity Profile) is Cognizant's proprietary run-cost model. It establishes a baseline cost from 7 production factors, then applies optimization levers to model achievable savings.
Model tiers:
Tier
Cost / 1M tokens
Examples
Frontier
$15–75
GPT-4o, Claude Opus, Gemini Ultra
Mid-Tier
$0.75–4
GPT-4o-mini, Claude Haiku, Gemini Flash
OSS / Sovereign
<$0.50
Llama 3.3, Phi-4, Mistral Large 2
Achieving 3–4 optimization levers simultaneously typically reduces run cost by 70–85% from TCP baseline.
Save, Export & Share
💾 Save — Stores the full project state in your browser's local storage. Reload the page and your data persists.
⬇ Export CSV — Downloads all use cases with ACP, TCP, build cost, run cost, and scenario columns. Named agenomy_<project>.csv.
📊 Dashboard — Jump directly to the executive summary. Recalculates all figures before display.
Tip: Use Scenario Comparison to show the client the range — OSS/Sovereign (aggressive optimization) vs Frontier (all Frontier models, no caching) — before settling on a recommended baseline.
Literature — Agenomy Framework Reference
Key materials underpinning the Build Sutra · Run Sutra · Agent Foundry methodology. For working sessions and client conversations.
Agenomy Framework
Agenomy is the umbrella discipline for enterprise agent economics. It provides a structured methodology for costing, governing, and optimizing agentic AI programs across their full lifecycle.
Lifecycle: Design → Build → Deploy → Run → Govern → Optimize → Value
Deck: Cognizant_Agenomy_Deck_v11.pptx (22 slides, July 2026) Audience: CFO · CTO · CPO · Procurement · AI Program Office Contact: AIMUDeals@cognizant.com
Build Sutra — ACP Methodology
Build Sutra is powered by ACP (Agentic Complexity Points) — a Cognizant proprietary scoring model that converts agent complexity into a defensible build estimate and commercial model.
4 weighted dimensions: Orchestration · Data · Integration · Governance
1 ACP = 4 person-hours of delivery effort
Direct mapping to T&M / Fixed Price / Milestone / Phased SOW
ACP is a Cognizant IP. Do not share scoring formula externally without DCC clearance.
Run Sutra — TCP Methodology
Run Sutra is powered by TCP (Token Complexity Profile) — a Cognizant proprietary model that establishes a production cost baseline from 7 factors and quantifies optimization potential.
Observability and control plane. Monitor → Remediate → Govern. FinOps, HITL, guardrails, agent registry, eval suites.
Agent Marketplace
Enterprise catalog for agent reuse and distribution. Persona-driven UI. AgentOps integration.
Credentialing Cases & Client Zero
Global Insurance Group — Claims & Compliance AI
3-agent program · ACP 104 (Phased SOW) · 63% run cost reduction · $1.1M annual saving vs TCP baseline. Semantic caching + LLM routing on 280M tokens/month.
Tier-1 Retail Bank — Customer Onboarding
4-agent graph · ACP 139 (Phased SOW) · 71% token cost reduction · £2.1M annual run cost vs £7.2M projected. 34% → 9% digital drop-off.
Cognizant Client Zero — Agent Foundry Ops on Our Own AI Budget
18 internal products · 600+ team · $4.2M annual AI API spend. First live deployment of Agent Foundry Ops. 34% cost down · 62% OSS routing · 41% cache hit · $0 budget surprises. We eat our own cooking before recommending it.
Practice: Analytics & AI · $100M+ · 600+ team Contact: AIMUDeals@cognizant.com