Available nowFor companies, operating teams, and their investors
Know what your companies spend on AI.
Connect AI spend to business results.
Set spending limits before money leaves the account. See which company and initiative used AI, what it cost, and what result management can support. Each company keeps control of its accounts and data; authorized portfolio teams get one consistent view.
Stop spending at the limit See cost by company and initiative Review the evidence behind reported results
Operating review · 30 daysAI value and control coverage
Guided example · fictional values Illustrative dataObserved AI cost$72.4k 8.2% vs. prior
Spend under a hard limit96%6 participating companies
Spend with result evidence71%4 management baselines
Company evidence positionControl coverage →
Evidence gapReview readyDecision requiredControlled1234
1Atlas94% / 82%2Harbor88% / 61%3Juniper76% / 74%4Meridian91% / 46%
OpenAIAnthropicAzure OpenAIGeminiBedrockCompatible APIs
The product in plain English
Put one set of rules between your business and the AI services it uses.
Your existing software keeps working. AI Gateway HQ checks who is asking, whether the work is allowed, and whether money remains in the budget. Approved work continues; blocked work stops before the bill.
Your business usesApps · tools · agentsNo new chat interface
AI Gateway HQCheck rules and budgetOne control point
Approved destinationAI serviceOnly when allowed
It does not replace ChatGPT, Claude, Codex, or your applications. It gives the business one place to control and understand AI use.
Individual or small team
Try the controls without becoming an AI expert.
Use the no-cost Test Lab first. No card or provider account is needed until you decide to connect real work.
- Guided setup
- Free simulation
Company or technical team
Set limits once, then use them across the company.
Connect provider accounts, set who may use them, choose approved services, and decide how much each workload can spend.
- Company identity and roles
- Limits checked before spend
Private-equity portfolio
Compare participating companies without taking away their control.
Review cost, approved business results, and open management questions across the portfolio, with a permissioned drill-down for each company.
- Company-approved reporting
- Evidence behind reported results
One place to operate AIInvisible to applications.
Invisible to applications.
Obvious to operators.
Replace scattered keys and one-off proxy rules with one operating surface for security, platform, finance, and application teams.
Policy-aware routing
Choose eligible targets by API compatibility, configured cost, priority, weight, circuit availability, identity, environment, risk signals, and signed workload attributes.
Explore routingWorkloadPolicyEligible target
Budget at decision time
Atomically reserve gateway credit and configured provider cost before inference, then reconcile non-streaming token usage.
Control costGovernance with explicit modes
Use Observe, Shadow, or Enforce, inspect reason codes, and block local jailbreak signals before provider credentials are used.
See governanceEncrypted BYOK baseline
Use tenant-bound KMS encryption in the AWS production profile. Customer-VPC deployment and customer-managed-key options are planned.
Review deploymentA stable developer surfaceChange the route.
Change the route.
Not every application.
Use compatible model APIs while administrators rotate credentials, change route pools, cap usage, or contain an upstream outage centrally.
OpenAI Responses, chat, embeddings Anthropic Messages and streaming JSON payload pass-through with model and token-cap enforcement
Read the quickstartfrom openai import OpenAI
client = OpenAI(
base_url="https://api.aigatewayhq.com/v1",
api_key=os.environ["AIGHQ_API_KEY"]
)
response = client.responses.create(
model="company-approved-fast",
input="Summarize the incident timeline"
)200 governedroute_cost-saverpolicy.allow
Built for shared ownership
Everyone sees the control they need.
The customer console brings gateway configuration, recent cost allocation, and request evidence together. The permissioned portfolio view adds executive comparison across participating companies. A company can separately approve a bounded provider statement subtotal and variance rollup without sharing invoice files, payment details, or provider credentials.
Start in a test environment
Follow the safe-start path Observe first. Enforce after validation.
Connect a restricted test credential, inspect policy and guardrail decisions, and validate operational behavior before routing sensitive or production traffic.