Secure SOC 2 · HIPAA-ready AI for business use

One API. A council of AI models. An answer you can defend.

WiseAI Council sends your task to the best-fit models from several vendors, has them review each other's work, and returns one answer with the dissent, the cost and the audit trail attached.

Secure, SOC 2 and HIPAA-ready AI for business use: SOC 2 and HITRUST certified, BAA signed online, PHI lane; a request passes the compliance gate to a council of models from different vendors and comes back as one reviewed answer
In short

WiseAI Council is WiseTREND's multi-model AI service, delivered as a REST API and a self-service dashboard at my.wisetrend.com/wiseai. Each request is classified into a task family (documents, contract review, engineering, architecture debate, operations, document extraction, file processing, research, media), routed to a panel of models from different vendors — Anthropic, OpenAI, Google, DeepSeek, Meta, Mistral, xAI and others — and run in a deliberation mode (solo, draft-critique-revise, panel, debate or map-reduce) so that a second opinion is a different model's opinion. A compliance gate enforces budgets, data-lane rules and provider terms before any model is called; requests carrying protected health information run only on providers under a signed Business Associate Agreement. Tenants sign up in the WiseTREND One portal, create their own API keys, execute the BAA online in minutes, set their own spending ceilings, and see every request, model and charge on the dashboard. Pricing is usage-based, per million tokens, with the provider's model and WiseAI Council's services in one rate.

What WiseAI Council does

WiseAI Council is WiseTREND's multi-model AI service, delivered as a REST API and a self-service dashboard at my.wisetrend.com/wiseai. Each request is classified into a task family (documents, contract review, engineering, architecture debate, operations, document extraction, file processing, research, media), routed to a panel of models from different vendors — Anthropic, OpenAI, Google, DeepSeek, Meta, Mistral, xAI and others — and run in a deliberation mode (solo, draft-critique-revise, panel, debate or map-reduce) so that a second opinion is a different model's opinion. A compliance gate enforces budgets, data-lane rules and provider terms before any model is called; requests carrying protected health information run only on providers under a signed Business Associate Agreement. Tenants sign up in the WiseTREND One portal, create their own API keys, execute the BAA online in minutes, set their own spending ceilings, and see every request, model and charge on the dashboard. Pricing is usage-based, per million tokens, with the provider's model and WiseAI Council's services in one rate.

Key capabilities

A council, not a single model

Every request is seated on models from different vendors and lineages. In draft-critique-revise a second model critiques the first's draft; in panel and debate modes an arbiter merges or rules; map-reduce splits large inputs. Dissent is returned, never hidden.

How WiseAI Council turns one request into one reviewed answer: request, compliance gate, models from different vendors, one answer

Task-aware routing

Nine task families - documents, contract review, engineering, architecture debate, operations, document extraction, file processing, research and media - each with its own seat policy, learned from outcome scores over time.

Every request is classified into one of nine task families and seated by that family's policy

Compliance gate before every call

Budget ceilings, task-family entitlements, data-lane rules and provider terms are checked before a model is called. A refusal comes back with a stable rule code, not a vague error.

Security and compliance layers checked on every WiseAI Council request

PHI lane with a self-service BAA

Execute the Business Associate Agreement in the dashboard - no e-mail, no waiting. Requests flagged as PHI are then routed only to providers under a signed BAA whose deployments WiseTREND has verified.

The Data protection page where the BAA and DPA are signed online

Spending ceilings you control

Daily and monthly ceilings in the dollars you are charged. When a ceiling is reached the next request is refused with a clear code; nothing is ever silently run on a cheaper model.

Monthly and daily spending ceilings set in the dashboard

Named API keys

One key per integration, created and revoked by you, with usage attributed per key. Keys are shown once and stored only as hashes.

One API key per integration, with usage and charges attributed per key

Full transparency on cost

A monthly statement by model and by day, the rate card you are metered at, and every request with its family, mode, status, tokens and charge.

A result showing every model that took part, its tokens, time and charge

Your data, on your terms

Task text, attachments and answers are purged automatically after 30 days (7 for PHI) and can be deleted immediately with a single API call. Accounting and the audit trail survive; the content does not.

Content is kept for the retention period and purged; accounting and audit stay

Failover and audit built in

A seat that fails or is refused moves to the next vendor within the same lane; every routing decision, refusal, call and purge is an append-only audit record.

Models are onboarded, classified, ranked and retired; a failed seat moves to the next approved model

OpenAI-compatible API and DocAI Fabric

DocAI Fabric, the openai SDKs and any OpenAI-compatible tool can call the Council as one model at /v1/chat/completions with your Council API key - JSON in your own schema voted field by field, page images, tool calls and streaming - with the same security checks, metering and retention as the REST API. Connect > DocAI Fabric walks through the setup field by field.

DocAI Fabric, the openai SDKs and OpenAI-compatible tools calling WiseAI Council at /v1/chat/completions

Integrates with your systems

WiseAI Council connects to REST API (council.wisetrend.com/v1), OpenAI-compatible API (DocAI Fabric, openai SDKs), MCP for AI assistants, Anthropic, OpenAI and Azure OpenAI, Google Vertex AI, Azure AI Foundry (DeepSeek, Meta Llama, Mistral, xAI Grok), Perplexity, ABBYY Vantage and FlexiCapture workflows, Webhook callbacks, plus any system with a REST API, SOAP endpoint, hot folder, SFTP drop, or database connector.

What's new in WiseAI Council

Latest

OpenAI-compatible API and DocAI Fabric

  • Use the Council from any OpenAI-compatible tool or the openai SDKs with your Council API key, with five model ids from the full council to one fast model.
  • Ask for JSON in your own schema: three model families answer, every field is voted, and disputed fields go to a tie-break model.
  • Register WiseAI Council as a model in DocAI Fabric for split, classify, extract, redact and the assistant; Connect > DocAI Fabric shows every setting.
  • Sync and async requests now return the same JSON shape, so an integration can switch without changing its parser.
All WiseAI Council release notes
Frequently asked

Questions about WiseAI Council

Answers written for buyers, search engines, and AI assistants evaluating multi-model AI orchestration and HIPAA-ready AI APIs.

What does WiseAI Council do that calling one AI model does not?

It seats several models from different vendors on the same task and has them review each other's work. A single model has no way to know when it is wrong; a second model from a different lineage often does. The Council returns one answer plus the dissent, the confidence and the per-seat record, so you can see how the answer was reached and defend it.

How is WiseAI Council priced?

Per million tokens, in the dollars shown on your dashboard's rate card. The rate covers the provider's model and WiseAI Council's services - routing, peer review, compliance gating, metering and audit - in one package. You are charged only for the models that actually took a seat, and nothing for a failed model call. There is no minimum and no contract; you set your own daily and monthly ceilings.

Can I send protected health information?

Yes, once your organisation has executed the Business Associate Agreement - which you do online, in the dashboard, in a few minutes, with nobody at WiseTREND involved. Requests flagged as PHI are then routed only to providers under a signed BAA whose deployments WiseTREND has verified, and the dashboard shows whether that lane is open today.

What happens to my documents and answers?

Task text, attachments and answers are purged automatically after 30 days (7 for PHI). You can also delete any request's content immediately with one API call. Usage records, charges and the audit trail are kept so your statement stays explainable; the content itself is gone.

How do I get started?

Sign in at my.wisetrend.com with your WiseTREND One account and open WiseAI Council. Your organisation gets a Council account at once. Create a named API key, then follow the Connect page - a single curl command sends your first request.

What happens when I hit my spending ceiling?

The next request is refused before any model is called, with the code budget.client_daily or budget.client_monthly, and every request after it until the window resets. Nothing is ever run on a cheaper model without telling you. Raise the ceiling on the dashboard and requests resume at once.

Does it work with my existing ABBYY workflows?

Yes. WiseAI Council is the AI back end behind WiseTREND's own Vantage and FlexiCapture integrations, and the same REST API is open to your workflows and to AI assistants through MCP.

Put a council behind your next AI request

Sign in with your WiseTREND One account, create an API key and send your first request in minutes. Usage-based pricing, no minimum, no contract.