Frequently asked questions
Short, plain answers first. Open “Full detail” on any question for the technical version. New to AI, or wary of it? Start with Straight answers.
What is CONTROL?
A system that keeps track of what’s happening in your business: what was promised, what changed, what’s overdue, and where each fact came from. It shows you what needs attention and leaves the decisions to you. It never changes your files and never sends anything for you.
Full detailHide detail
CONTROL is a Continuous Operational Intelligence platform from Gateway Data. It continuously captures operational signals from the work happening across an organization, including email, meetings, tasks, decisions, and documents. It connects those signals to the right people, projects, and history, verifies their source and confidence, and makes them available for permitted action.
How is CONTROL different from AI summarization tools?
A summary makes things shorter, and the detail you needed often gets cut. CONTROL keeps the detail and remembers where each piece came from, so you can check it.
Full detailHide detail
Summarization tools compress information: they make large, noisy inputs shorter. What they drop is often what operations teams need most: the anomaly, the dependency, the change that actually matters, along with where it came from and who is allowed to see it. CONTROL is built to extract that signal and keep its provenance and security context intact.
Why not just connect a regular AI tool to my email and files?
As your business grows, so does that pile of information. Eventually there’s more than the AI can read at once (its ), and some of it quietly gets left out of the answer.
CONTROL picks out the important details once, as they arrive, and keeps where each one came from. Answers come from what it already knows, with a link to the exact email or document behind them.
Full detailHide detail
A large language model LLM can only consider a limited amount of text at once: its context window. Tools that answer by re-reading your sources must fit everything relevant into that window on every question. As email, meetings and documents accumulate, the volume eventually exceeds it, and reference material is truncated or excluded without warning. Each question also repeats the time and cost of reprocessing the same sources, and the answers vary from run to run.
CONTROL extracts operational signals once, at capture, and preserves them as intelligence artifacts with their source, time, entities and provenance. Every later question is answered first by retrieving and assembling known intelligence artifacts. There is no re-querying of all your data sources. Original sources are reconsumed only when an analysis specifically requires it, and then only the specific citation is retrieved from the source-of-truth document.
Your history keeps growing. The amount the AI must read to answer any one question doesn’t.
Is CONTROL a dashboard or a chatbot?
Neither. A only knows what you type into it. CONTROL is the careful process underneath: checking sources, permissions and evidence. You can ask it questions or view it on screen, but that’s just the front door.
Full detailHide detail
Neither. CONTROL is an architecture for operational intelligence. A chat interface or a dashboard can sit on top of it, but the core of CONTROL is governed signal extraction, evidence, and permission-aware delivery.
Do I need to replace my existing software?
No. Keep everything you use now. CONTROL works alongside it.
Full detailHide detail
No. CONTROL does not ask you to replace the software your business already uses. Keep your email, calendar, customer records, project tools, documents and specialized systems. CONTROL adds a governed intelligence layer around them, capturing the signals they create and connecting them into one operational picture. There is no rip-and-replace migration.
Does CONTROL write back to my systems?
No. It only reads. It can’t change or delete anything in your email, files or other software. It can prepare email drafts for you to send yourself, and add calendar entries if you allow it.
Full detailHide detail
No. CONTROL doesn’t change the systems it learns from. It is read-only by design.
Third-party sources such as Slack, Notion, Google Drive, Docs and Sheets are connected read-only. Client source documents, including records stored in the Records Vault, are never modified. Extracting a signal from a source never gives CONTROL the ability to change or transform the original. The only things CONTROL adds outside itself are ones you approve: calendar entries, with your permission, and Gmail drafts for you to review and send yourself. CONTROL has no authority to send email, text messages or any other message on your behalf.
That’s possible because CONTROL builds and manages its own intelligence packets, which hold references and pointers back to the original sources. There’s nothing to sync, no two-way coordination, and none of the security risk, data-contamination concerns, maintenance cost or overhead of read-write integrations. Your existing systems stay exactly as they are.
Is my data used to train AI models?
No. Your business information isn’t used to train the CONTROL uses.
Full detailHide detail
No. Customer data is not used to train third-party foundation models. Model access is governed by policy, and model-assisted actions are recorded for audit. See our Security page for more.
Do I need my own AI subscription?
No. The AI is included in your subscription, so you don’t need ChatGPT or any other AI service. You can choose a different AI model for any task, or use your own AI account if you prefer.
Full detailHide detail
Every CONTROL subscription includes a bundled token allocation that pays for the AI it uses. CONTROL routes each task to a suitable model under your workspace policy, and the AI Model Router MMR lets you choose the model for a task, see a cost estimate before each run, and switch providers mid-task.
Already pay for an AI service and use CONTROL heavily? The optional bring-your-own-key setting lets CONTROL use your own AI account instead. It is never required.
Does CONTROL respect existing permissions?
Yes. Each person only sees what they’re allowed to see.
Full detailHide detail
Yes. CONTROL is permission-aware by design. It applies role-based access control RBAC and visibility rules, and it releases intelligence only through authorized views. You never see signals you are not permitted to see.
Can I subscribe to CONTROL now?
Yes. Job Search Basic is open to everyone now. Job Search Standard, Job Search Professional and CONTROL Business are available through the Early Access Program.
Full detailHide detail
Yes. CONTROL Job Search Basic is available to everyone as a paid subscription. See the job search screens →
Job Search Standard and Job Search Professional, which include Job Search, basic task management and skills, are in the Early Access Program, as is CONTROL Business (up to 10 concurrent users).
Subscription sign-up and secure payment by Stripe happen on the CONTROL overview page.
How do I get access to CONTROL?
For Job Search Basic, just sign up. For Job Search Standard, Professional or CONTROL Business, request Early Access. For Business, tell us briefly what you’d like it to help with.
Full detailHide detail
CONTROL Business (up to 10 concurrent users) is currently in the Early Access Program. To apply, share a brief use case and how you see CONTROL providing a solution; a limited number of businesses with 1–10 users are selected to represent a cross-section of industry segments. Job Search Standard and Professional are also in Early Access, with no use case needed. Job Search Basic is open to everyone: sign up here. Request Early Access and our team will review your request.
Is there a free trial?
No. CONTROL learns about your business over time, so a short trial would show you very little. Instead: a 3-month commitment to get started, but renewals require your explicit request. If it isn’t worth it, do nothing and it ends.
Full detailHide detail
No. CONTROL uses a full-feature, subscriber-driven renewal model, and there are no free trials. CONTROL builds intelligence over time. A short free evaluation would accumulate very little of it, so it couldn’t show what CONTROL actually does for your business.
Free trials and limited evaluation editions rarely reflect what a fully implemented, integrated and populated platform looks and feels like. The gap tends to surface later, as add-on modules, upgrades or higher tiers needed to reach the performance you expected at the start.
CONTROL subscriptions are all-in. No limited-feature trials, no evaluation edition that bears little resemblance to the real thing, and no bait-and-switch. What you start with is the full product. You get every feature, sample data and a demo, help built in everywhere, and GENESIS guided onboarding, so you can onboard yourself and get hands-on with every module, with the flexibility to manage peak-consumption months.
Subscriptions aren’t limited and features aren’t paywalled. Every subscription includes a bundled token allocation. The reasoning layer ARGO, GENESIS, Marketing Intelligence and the Artifact Factory are advanced, high-token-consumption features, but most businesses don’t use them every month. They consume more tokens than standard modules like MAILstrom, Job Search and Task Management, and they are always available to explore and use.
For the occasional month when use of these modules spikes, you can buy a Bump or Token Index Paks to cover it. This keeps the CONTROL subscription price lower than it would be if every plan bundled enough tokens for continuous high consumption that most months never need. For consistently high use, there is also a bring-your-own-key option, where CONTROL uses your own AI account.
A 3-month commitment to get started, but renewals require your explicit request. At the end of that period, your subscription is suspended unless you choose to renew. We don’t lock you into an annual subscription because you forgot to cancel, and we don’t trap your data in a proprietary format you can’t get back out.
If CONTROL has added value and helped you run and understand your business, you reach out and renew. If it hasn’t, you do nothing and the relationship ends. Your profile and its intelligence packets are deleted from our systems after the 60-day cancellation period.
A subscription only renews or auto-renews if you explicitly ask for it. Merit-based subscriptions: it’s how we think all SaaS should earn its fees.
Who makes CONTROL?
CONTROL is built by Gateway Data. You can reach the team at control@gatewaydata.ca, or contact the Early Access team at eap@gatewaydata.ca.
What is Continuous Operational Intelligence?
Our name for AI that keeps an up-to-date picture of your business as things happen, shows where everything came from, and stays inside the rules you set.
Full detailHide detail
Continuous Operational Intelligence is a business intelligence capability that continuously captures operational signals, connects them to the right entities and history, verifies their source and confidence, and makes them available for permitted action across the organization. Read the full definition.
What is an intelligence packet?
A tidy record CONTROL keeps about one thing, like a customer or a job: what’s happened, where each fact came from, how sure it is, and who’s allowed to see it.
Full detailHide detail
An intelligence packet is the unit of information CONTROL produces. It combines an entity (a person, project, customer, or event) with its history, supporting evidence, a confidence level, and visibility rules. Packets move through authorized views to support actions such as explaining, alerting, recommending, routing, or approving. The packet travels; private profiles do not.
What is the permission check GATEWAY?
The part of CONTROL that checks permission before anything happens, and keeps a record of each decision.
Full detailHide detail
The permission check is CONTROL’s decision and governance layer. Before CONTROL performs a governed operation, the permission check determines whether it is permitted in the current identity, workspace, security, privacy and policy context.
That can include checking permissions, information classification, source authority, subscription entitlements, AI-model eligibility, funding, usage limits and required approvals.
The permission check is different from the search layer QIL. The search layer determines what evidence is needed to answer a question; the permission check determines what the search layer, or any other CONTROL subsystem, is authorized to access or do.
This allows CONTROL to retain evidence not only of what happened, but of the authority and policy under which important operations were permitted or denied.
Is the permission check just about security?
No. It checks more than who you are. It also checks what information is involved, what you’re trying to do, and whether it’s allowed right now.
Full detailHide detail
No. Authentication and permissions are part of the permission check’s decision context, but the permission check is broader. A governed decision may depend on:
- who the user is;
- which workspace they are operating in;
- what role and permissions they have;
- which information is involved, and its classification or privacy requirements;
- which entity it concerns;
- the current workflow state;
- subscription and commercial policy;
- AI-model eligibility;
- funding and usage limits;
- and whether additional approval is required.
The permission check evaluates the controls that apply to the specific operation, rather than assuming that a successful login or broad role automatically authorizes everything that follows.
Does the permission check control AI?
It decides whether AI may be used for a task, and with which information. The AI never decides that for itself.
Full detailHide detail
It governs AI execution, but it is not an AI model or a model router. When CONTROL proposes an AI operation, the permission check determines whether that execution is permitted under the applicable security, privacy, model, provider, funding and commercial policies.
Model selection and routing then happen inside that governed boundary. CONTROL can use different models without allowing the model itself to decide what information it may access or what it may do.
What’s the difference between the search layer and the permission check?
One part works out what information is needed to answer your question. The other decides whether you’re allowed to see it.
Full detailHide detail
The search layer determines what evidence is needed to answer a question. The permission check determines what evidence and operations are permitted.
For example, the search layer may determine that a question requires meeting evidence about a particular person from the last seven days. The permission check determines whether the current user, workspace and operating context are authorized to access those sources and disclose that person’s information. The search layer performs the query inside that authorized boundary.
How do CONTROL’s intelligence systems work together?
Seven parts, each with one job: capture, remember, identify, find, check permission, think things through, and keep a record. They work together in a loop as new information arrives.
Full detailHide detail
CONTROL separates the responsibilities required to turn information into trustworthy operational intelligence. The capture layer CURT preserves captured source information. The memory layer MOM provides persistent operational memory. The identity layer EIL establishes the identities and relationships involved. The search layer turns questions into bounded evidence queries. The permission check determines what each operation is authorized to access or do. The reasoning layer uses governed evidence for higher-order analysis and planning, and the record layer MARS preserves evidence of important decisions and activity.
These systems continually interact rather than operating as a simple pipeline. New information changes operational context, which changes later queries, decisions and actions: a continuous cycle of capture, resolution, identity, context, query, governance, reasoning, action and audit.
Which AI models does CONTROL use?
Whichever suits each task, within the rules you set. CONTROL isn’t tied to any one AI company.
Full detailHide detail
CONTROL is model-agnostic. It routes each task to an appropriate model under your organization's policy, weighing quality, speed, cost, and data boundaries. Because models change quickly, CONTROL is designed so no single model is a permanent dependency.
How is CONTROL different from personal AI memory assistants?
Those remember what one person saw. CONTROL builds a checked, shared picture the whole business can rely on.
Full detailHide detail
Personal context assistants remember what one person saw, heard, or opened. That makes them useful, but it is personal and can be incomplete. CONTROL converts permitted signals into shared, verified artifacts that the whole organization can rely on and act on.
Job Search Basic is open to everyone now. Job Search Standard, Job Search Professional and CONTROL Business (up to 10 concurrent users) are in Early Access; for Business, apply with a brief use case. See the job search screens →