Pricing
Pay for the work your agents actually do.
AlphaAgent is priced on input-token consumption, and the unit price falls as you scale. You run everything in your own AWS account and pay AlphaAgent only at runtime, when your team chats with an agent or runs a workflow. No output-token fees. No setup fees. No per-seat licensing.
How billing works
Two costs, both transparent.
AlphaAgent runs in your own AWS account, so you pay AWS directly for the infrastructure and the Amazon Bedrock model usage your workloads consume. On top of that, you pay AlphaAgent a licensing fee for the operating layer — the intelligence, orchestration, governance and runtime that turn a raw model endpoint into production-grade agentic work.
Paid to AWS · your account
Infrastructure & model usage
A fixed infrastructure baseline (compute, networking, storage) plus variable Amazon Bedrock token inference — all at public AWS list prices, in your environment. Sizing tiers and worked figures are in the docs.
Infrastructure cost & sizingPaid to AlphaAgent · the operating layer
Consumption licensing
A single, declining rate per million input tokens — billed only at runtime. Building the production layer on top of a raw model is the hard, expensive part. This is the engineering you don’t have to do, priced against the tokens you actually run.
You only pay at runtime
Creating agents, connectors and Augmented Multi-Resolution Property Graphs carries no AlphaAgent charge. We bill only when your agents do work — in chat or in a workflow.
Input tokens only
AlphaAgent is priced on input-token consumption. We do not charge an output-token fee, so the cost reflects the reasoning, retrieval and orchestration a task actually needs.
The price falls as you scale
Your unit rate declines with volume. The more agentic work you run, the less each million tokens costs — so growing usage never means a runaway bill.
Nothing to pay while you evaluate
During your evaluation period you pay no AlphaAgent charge at all. We measure your real usage first, then agree a fair rate based on how you actually work.
The rate card
Your rate declines with monthly volume.
AlphaAgent licensing per million input tokens. These are published reference rates — the fair starting point for every conversation. Amazon Bedrock and AWS infrastructure are billed separately through your own account.
Pilots and single teams
A team in production
Multiple teams
Production scale across a function
Enterprise-wide deployments
Rates are indicative and agreed per customer after an evaluation period. Large or committed-volume deployments are priced bespoke.
What it costs, in practice
Tokens are hard to picture. Workflows aren’t.
So here is what real agentic work costs across the industries we serve — the price of a single run, and what it looks like once a team is running at scale. A single deep workflow is inexpensive; the economics are defined by how many you run, which is exactly what walks you down the rate card.
Capital Markets
Markets, portfolios, research and risk
Signal generation (full desk run)
~10M tokens
~$160per run
Investment research memo (evidence-backed)
~4M tokens
~$64per run
Portfolio construction & scenario analysis
~5M tokens
~$80per run
At scale
A research desk running ~50 deep runs a day ≈ ~5B input tokens / month, at $11 / 1M.
~$55k / monthAlphaAgent licensing
Banking
Personalization, relationship management and operations
Deep KYC / KYB / enhanced due diligence
~1M tokens
~$16per run
Credit & lending analysis
~2M tokens
~$32per run
Fraud investigation
~2M tokens
~$32per run
At scale
An onboarding team clearing ~3,000 deep KYC reviews a month ≈ ~3B input tokens / month, at $11 / 1M.
~$33k / monthAlphaAgent licensing
Insurance
Underwriting, claims, distribution and policy intelligence
Underwriting analysis
~3M tokens
~$48per run
Claims triage (robust)
~1M tokens
~$16per run
Fraud & subrogation investigation
~2M tokens
~$32per run
At scale
A claims unit triaging ~5,000 claims a month ≈ ~5B input tokens / month, at $11 / 1M.
~$55k / monthAlphaAgent licensing
Payments
Merchants, networks, risk and reconciliation
Merchant onboarding
~1M tokens
~$16per run
AML / sanctions / dispute workflow
~1.5M tokens
~$24per run
Fraud investigation
~2M tokens
~$32per run
At scale
A risk operation clearing ~8,000 cases a month ≈ ~12B input tokens / month, at from $7 / 1M.
~$84k / monthAlphaAgent licensing
Illustrative only. Per-run figures use a reference rate of $16 per 1M input tokens; at-scale figures apply the volume rate for that monthly usage. Token counts vary with task depth. Amazon Bedrock and AWS infrastructure are billed separately through your own AWS account. Your rate is agreed after an evaluation period and reviewed down as your volume grows.
Fair by design
You don’t pay until we know what fair looks like.
We would rather price you on how you truly work than lock you into a number before either of us knows the shape of your usage. So the evaluation period is free of AlphaAgent charges, and pricing follows the evidence.
Extended evaluation
You get an extended evaluation period — up to 90 days, or longer where it makes sense — with no AlphaAgent charge. You build your agents, connectors and AMPGs and run real work, paying only your own AWS.
We measure real usage
Over that period we see how your team actually uses AlphaAgent — which workflows, at what depth, at what volume. Your pricing is grounded in evidence, not a guess made on day one.
A fair rate, reviewed down
We then agree a rate tailored to your usage and lock it in. As your volume grows, we review it down — so scaling up on AlphaAgent only ever makes each unit of work cheaper.
Build the operating layer first.
Pay AlphaAgent when it works.
Early access is guided: demos, workflow design, and hands-on setup with our team. Tell us the workflows that matter and we’ll show you exactly what they cost.