Common questions.

Getting started

What kind of companies does DividendAI work with?

We work with industrial operators: manufacturers, energy companies and utilities, mining and resources businesses, infrastructure and construction firms, and aerospace and defense organizations. What they have in common is document-heavy operations, scarce expert time, and years of production data that has never been properly interrogated. If your business runs on physical things, whether components, plants, assets or sites, we are built for you.

We work with organizations that run on documents and data: logistics and supply chain, financial services, professional services, healthcare and life sciences, technology and the public sector. What they have in common is document-heavy operations, scarce expert time, and years of records that have never been properly interrogated. If your people spend their time reading in order to decide, we are built for you.

How quickly can we get started?

Faster than any comparable engagement. Our Amplify AI platform is already built. Day one is spent choosing the process and agreeing what improving it is worth. By the end of week two, a working agent has been tested against your own historical data. By week five, it is live in your environment.

What do we need to provide to get started?

Three things. Access to the records the process depends on, whether work orders, maintenance logs, drawings or recipes. One expert who owns the decision and the judgment calls, to work alongside us. And your IT environment for deployment. We handle everything else, including data structuring, agent development and platform configuration.

What does Phase 1 cost, and what are we committing to?

Phase 1 is scoped to be small enough to walk away from. You are committing to a short, defined engagement. We agree the scope and the cost up front, tied to what improving the result is worth to your business. If the agent does not prove accuracy against your own history by the end of week two, you can walk away. If it succeeds, we move to Phase 2 and deploy in your environment.

The technology

Do we need to replace your existing systems?

No. DividendAI works with the records and systems you already have. We structure your existing data and deploy agents that operate within your environment. We build on the technology investments you have already made. Nothing is replaced, and nothing is migrated to a vendor cloud. Your data stays where it is.

What types of documents and data can agents work with?

Maintenance work orders, calibration certificates, engineering drawings including marked-up revisions, furnace and process charts, procurement records, production schedules, inspection reports and quality records. If your team reads it to make a decision, the agent can be trained to read it too, at a scale your people do not have time to reach.

Contracts and rate cards, invoices and remittances, policies and procedures, ledgers and reconciliations, reports and board packs, correspondence and case files. If your team reads it to make a decision, the agent can be trained to read it too, at a scale your people do not have time to reach.

What does "human in the loop" mean in practice?

During weeks three and four the agent works live as advice. It handles each case and your designated expert confirms or corrects every call. That confirmation data refines accuracy before the agent moves to production. After go-live the agent works autonomously within defined boundaries, and exceptions are routed to your team. Agents amplify the capabilities of your people, freeing them for judgment and decisions rather than activities and analysis.

What large language model providers do you use?

We are model-agnostic. Six model vendors are available, enabled per organization and lockable to a single model per agent once it is proven, so you can match capability and token cost to each task. Inference runs under zero-retention and no-analytics agreements: nothing is kept after the call and nothing trains a model. The platform is deployed inside your enterprise environment, and you retain full control over which providers are approved.

Can the platform handle unstructured data, such as scanned documents, handwritten notes and legacy formats?

Yes. A significant part of every engagement is structuring data that has never been properly organized. Scanned drawings, handwritten calibration certificates and legacy file formats are common. An agent is only as good as the data it can reach, so clean data is the first layer of every deployment, not an afterthought.

Yes. A significant part of every engagement is structuring data that has never been properly organized. Scanned contracts, annotated statements and legacy file formats are common. An agent is only as good as the data it can reach, so clean data is the first layer of every deployment, not an afterthought.

Security and data governance

Where does our data go?

Nowhere it does not already live. Amplify runs in the cloud, inside your own environment and your own tenant, with data residency, security and governance to enterprise standard. Data does not move to a vendor cloud. Calls are transient and leave nothing behind, and nothing is used to train a model. Your data is yours. We do not offer air-gapped or edge deployments today.

How is access to the platform controlled?

Through your own identity system. We integrate with your enterprise single sign-on and enforce role-based access control, so each user sees only what their role permits. You govern access. We do not.

What audit trail is available?

Full auditability is part of the product. Every agent action is logged under your own audit controls: what it accessed, what it recommended, and which person confirmed or overrode it. You can answer any compliance question about what the agent did and why.

Are you compliant with our industry's regulatory requirements?

Security and compliance controls are built into the platform from the start, not layered on at the end. Because agents run in your environment under your governance, your existing regulatory controls apply. We work with your security and IT teams during deployment to confirm requirements are met. Specific certifications and controls are discussed during scoping.

Results and value

What results have you seen in production?

Across more than 30 enterprise deployments in more than 15 countries: 87% reduction in production planning time, 80% shorter recipe optimization cycles, 85% time reduction in drill and blast fragmentation analysis, 75% reduction in material strength optimization cycles, and 75% lower token costs through platform efficiencies. Every engagement begins with a written hypothesis and a number, and we agree what it must prove before we start.

Across more than 30 enterprise deployments in more than 15 countries: 87% reduction in production planning time, 85% reduction in contract review time, $4.2m+ in pricing exposure identified per review cycle, 90% less cash flow forecast preparation time, and 75% lower token costs through platform efficiencies. Every engagement begins with a written hypothesis and a number, and we agree what it must prove before we start.

Can you guarantee outcomes?

We write down what the engagement must prove before we begin. If the agent does not demonstrate accuracy against your historical data within two weeks, you can walk away. Every phase is scoped to earn the next one. You are not buying a technology deployment. You are buying a proven result, or the ability to stop.

How is return on investment measured?

We align on the metric before we begin: the right answer found on historical data and on live cases, time from trigger to confirmed answer, units saved per event, and the annual value of the outcome. Your expert's assessment is part of the measurement. We track adoption, not deployment. An agent that runs but is not used is not a success.

Does the value compound over time?

Yes, and it is a large part of why clients stay. The data structures, integrations and patterns proven on the first process carry into the second, so the second costs less to build than the first. Over time you develop a growing estate of agents, each benefiting from the infrastructure already in place. Our customer retention rate is 100%.

Working with DividendAI

What does "forward-deployed engineers" mean?

It means our engineers work alongside your team, in your environment. They learn how your operation actually works, rather than how a software manual says it should. Every deployment is built for your operating reality: physical constraints, data availability, climate, equipment age. All of it is taken into account.

What does the ongoing relationship look like after go-live?

Our engineers stay with you. They monitor accuracy and drift and keep building in rolling sprints, applying patterns proven on one process to the next, so each process costs less to build than the one before it. Support levels are set in the contract rather than assumed. See how we deploy.

That is your choice, and we agree it during scoping. Most clients keep a part-time allocation of the same engineers, who monitor accuracy and drift and keep building in rolling sprints, applying patterns proven on one process to the next. Some train their own engineers on the platform and take it in house. Support levels are set in the contract rather than assumed. See how we deploy.

How is DividendAI different from a consulting firm or a systems integrator?

Consulting firms charge for time and deliver recommendations. Systems integrators charge for implementation and deliver configured software. DividendAI does both, and owns the outcome. We write down what the result must be before we start. We bring the platform pre-built. We deploy it in your environment. And we measure success by whether the agent delivers the right outcome, not by whether the software was installed.

Is DividendAI a software product or a services company?

Both, and you can take either. Amplify is our secure, pre-built AI agent platform, and it stands on its own if your engineers want to build with it. Most clients take the platform together with our engineers, who deploy it and build the first agents against your own records, because that is the fastest route to a result. The usual order is engineers first and your own team second, and you are not obliged to keep us to keep the platform. See the platform and how we deploy.

Why is DividendAI powered by Insight Partners?

Insight Partners is one of the world's leading software and technology investors, with a portfolio of more than 700 companies across enterprise software, cyber security and data infrastructure. Its investments include Wiz, SentinelOne, Armis, Databricks, Datadog and Veeam. Being powered by Insight gives DividendAI decades of enterprise software experience to draw on, access to that global network of portfolio companies, and the institutional support to sustain long-term client relationships. It is also why the platform was already built on the day we started.