Asked before every demo.

An AI workforce for a company that has the jobs but not the people: six modules on one account, one ledger and one agent layer, with your people keeping the decisions and the sign-off.

A twelve-person company with the same roles as a large competitor and three of the people. Also a department inside a larger organization that wants to run one brand this quarter without opening a procurement file.

No. A chat box bolted onto software was the previous generation. Here the intelligence is the product: agents with a job, a schedule, a budget and a data scope, working on your own data, every run on a ledger.

If you want a direct chat with the models, the Playground is there for that.

No. Take one. The others are there when you want them and cost nothing until you use them. There is no bundle to unlock and no tier to climb.

A brand is whatever you treat as a separate business: a product line, a market, a subsidiary. One account carries as many as you run, each with a context of its own enforced in code, and there is no per-brand fee.

The one closest to the job that hurts most. Most companies start with Marketing or Business Intelligence. If you are not sure, the Intelligence audit measures where you stand and says which function to switch on first.

No, and we say so. The catalog lists every agent as built, partial or not built, and each module page carries a roadmap band for what is not for sale yet. We would rather lose a deal to honesty than win one and explain later.

BearingBridge, run by people who spent two decades on the agency side. Named people, independent of any model provider, and no resale of your data.

You put money in a wallet, in dollars. Everything you use draws on it: tokens for the models, storage for what you keep, and paid data where a source charges for it. You see what an action spends before or as it runs, and when the wallet runs low, you add more.

No subscription, no seat fee, no minimum spend, no minimum term.

No. The wallet holds dollars, and it pays for what you use: tokens for the models, storage for what you keep, and paid data such as search results and rankings. Nothing of ours sits in between, so there is no pack to buy and no exchange rate to learn.

Nothing, at every headcount. Put the whole team in. A person reading reports costs you nothing; cost starts when somebody asks an agent to do something.

There is none, on purpose. Model prices move, sometimes overnight, and a table on a marketing page is wrong within a month. The price lives where the decision is made: image and video quote the exact figure before they run, and everything else prices itself on screen as it runs.

For the figures on your own workload, ask on a demo call.

No. You paid it in, so it is yours until you spend it. A month or three years.

The machine’s own mistakes. Retries, runs that fail halfway, and tools an agent reached for on its own initiative are our cost. A watcher that finds nothing changed does not charge you either.

There is nothing to trial because there is nothing to subscribe to. The account is free, the seats are free, and the balance stays at zero until you put something in it. Fund a small amount, run one module on one brand, and judge the output against your own numbers.

No card is needed to request an account. When you are ready to run work, you fund the wallet by card through Stripe. Your balance only increases from a signed payment event we verify, recorded once, so a retry cannot charge you twice. Nothing renews on its own unless you switch on auto top-up.

Each of the eight services is a fixed price, quoted before it starts, and none of it draws on your wallet. Onboarding, the audit and custom builds are paid once; managed operations and Premium Support are monthly with no minimum term. Start within thirty days of the audit and its fee comes off onboarding.

Four caps and a balance. A daily cap per person, a monthly cap for the organization, a draw limit per person on a shared wallet, and a monthly budget per agent. Every one is a number you set, and nothing spends past them, including the agents.

The work stops and says so. A person at their daily cap waits for tomorrow or asks an admin to raise it. An agent at its monthly budget stops until the month turns or the budget is raised. Nothing is billed past the line.

New AI work pauses until you top up. Nothing else changes: every report, ranking, article, deal and record stays there to read and export, and everyone can still log in. We do not hold your own data hostage over a lapsed balance.

Only if you switch auto top-up on. It comes with three brakes: a cooldown between top-ups, a daily limit, and a latch that stops after an error, so a runaway job cannot sit there charging your card.

Yes. Every run lands in the ledger with who ran it, on which brand, what it did, when, which model, tokens in and out, and the exact cost. Admins see the whole organization, which is how spend gets charged back to the brand that used it.

Image and video quote the exact figure first, and nothing is charged until you press the button. Everything else shows its cost on screen as it runs. Past that, the caps are there for the accident you have not thought of yet.

A worker defined inside the product with four controls that cannot be left blank: a job, a schedule, a budget in dollars, and a scope of data it may read. It runs overnight, weekly, on a trigger or on demand.

Three kinds exist today: watchers that look for change, drafters that produce work, and Ask Intelligence, the analyst on call.

No. Nothing an agent produces is published, launched or spent without a person with the right role signing it. Drafts wait. Ad campaigns arrive built and pushed paused. An article is refused until a person validates the outline.

Nothing. A watcher compares snapshots first and only calls a model when something moved, so a quiet week costs close to nothing.

It is a role, not a default. Some roles can define agents and set their schedule, scope and budget. Some can run agents that others built. Some can only read what agents produce.

No. Scope is enforced in code, not by a prompt. An agent built for Sales does not read Marketing’s drafts, and a draft for one brand cannot borrow the voice of another.

It is built not to. Ask Intelligence reads your data through fifteen read-only tools filtered on the active brand, quotes the real figure and names its source, or says the data is not there. When it explains why a number moved, the causes are computed in code from your own data before the model sees them.

Yes. A connector, a step in your trade, a report for the board or a watcher for your own source is a custom build: scoped, quoted before it starts, and running under the same caps, ledger and approval gate as everything else.

Seventeen text models across five providers, thirteen image engines and three video engines, all behind one gateway. Your admin chooses the model per module, per organization, and can restrict the list. If we stop supporting a model, the module falls back to the default.

The providers are named on the security page, where an evaluator needs to know where a prompt can physically go.

Yes, in the Playground. It holds every model your administrator allows, sends one message to up to three of them side by side, takes files in and gives documents out, and prints the price under every answer. It sees no brand data unless you paste it in.

No. Calls go to the providers on their API terms, which exclude training on API traffic. We keep the ledger line, not the transcript.

Yes. Markets and languages are set per brand, every keyword and ranking carries a language, and Ask Intelligence answers in the language of the question. Training sessions run in English or French.

Model calls leave through one gateway, to the providers your admin allowed, and nothing else leaves. The database is reached only from the server with a server-only key, and organization and brand scope is enforced in the application code on every query.

No. Each organization is its own tenant. Agents read only the organization they belong to, and there is no shared pool.

Not yet. The audit starts when customers need the report, and the security page changes that line the day it lands. The same page lists the other open items plainly: no two-factor sign-in and no single sign-on today. If your policy requires one of them, say so before you buy, not after.

The shared platform lives in a single region, named in the contract on request. If a residency rule is part of your decision, dedicated hosting gives you an EU or US region with backups that stay in it, provisioned in about two weeks, with a written uptime commitment.

Yes, at any balance: reports, rankings, articles, deals and records are readable and exportable. There is no subscription to cancel, so leaving is stopping. A dedicated instance comes with a full export written into the contract.

Visitors are identified by a salted hash, not by name. The assistant answers only from the knowledge base you curated, says so when it cannot, logs every exchange for you to read, and runs under a daily cost cap so a traffic spike is not a billing event.

Search positions and competitor pages for the keywords, markets and languages you track. What ChatGPT, Claude, Gemini and Perplexity say about you. Ad performance synced back from Google Ads. Everything else in the context is your own material, and only what you chose to put there.

No. Rankings are measured by the product itself for the keywords you track. A Search Console and GA4 connection is on the Marketing roadmap, and the catalog says so.

Not out of the box. Today an approved article leaves the content engine for you to publish. CMS auto-publication and the social publishing APIs are on the roadmap. If you need it now, a connector to your CMS is a custom build, quoted before it starts.

Yes, through the data integration service: ERP, CRM, PLM, PIM, MES, HR, finance, ad platforms and spreadsheets, with one written definition per metric and lineage on every figure, so Ask Intelligence can name the table and the row. Sales also ships with its own pipeline if you would rather start there.

Yes, that is the point of one layer. Marketing knows what Sales is closing. A watcher’s finding becomes a card on the Project Management board in one click. The weekly operating review reads every module. Nothing is re-explained.

A week. Within one business day of your request a person answers by name. Then a ninety-minute session configures your brands, markets and caps with you, the product reads your approved content and your competitors’ pages, and drafts start arriving. Five days later you are running it yourself.

The product runs without any service. Most companies buy two of the eight and never need the rest.

A conversation about your business, not a slide deck. Say which module you care about and we run it on your case. If you want the figures for your own workload, that is where you get them.

The shared onboarding channel stays open for thirty days. After that, support continues under your plan, or under Premium Support: monthly, fixed by the functions running, no minimum term, and it does not draw on the wallet.

Managed operations. A named operator runs the modules with you until your team takes them over. Monthly, fixed by how many functions are running, and you can stop with a month’s notice.

Yes. The partner program pays a commission on the companies you introduce that become customers. We handle the demo, the onboarding, the support and the billing. Partners also deliver the services for their own clients.