Question 1

What is the best place to start if we're not sure what to build?

Accepted Answer

Start with AI Strategy & Audit. We audit your operations, data sources, workflows, and current bottlenecks, then identify the AI opportunities most likely to create measurable value.

Question 2

Do we need to come with a specific AI agent idea?

Accepted Answer

No. Some clients come with a clear agent in mind, others only know where the business feels slow or manual. Discovery is designed to turn that operational context into a practical build plan.

Question 3

What do we get after Discovery?

Accepted Answer

You get a prioritized opportunity map, a recommended agent or workflow, the data and integration requirements, the risks to solve before build, and a realistic path to implementation.

Question 4

How long does it usually take to launch?

Accepted Answer

Most production builds take 4–8 weeks after scope is agreed. A focused single-workflow agent can move faster; multi-system agents with approvals, permissions, and complex data usually take longer.

Question 5

What can a Madency AI agent actually do?

Accepted Answer

It can read from your systems, answer questions with source-backed context, draft or execute actions, update records, trigger workflows, and help your team move work forward inside the tools they already use.

Question 6

Are agents custom-built or selected from a catalog?

Accepted Answer

Both. The catalog shows common agent patterns we can build quickly. The actual implementation is still shaped around your systems, your data, your approval rules, and the way your team works.

Question 7

Can the agent work with our existing tools?

Accepted Answer

Yes. We typically connect to CRMs, ERPs, databases, spreadsheets, document systems, internal APIs, Slack, Teams, email, and other operational tools through secure integrations.

Question 8

Can we require human approval before the agent takes action?

Accepted Answer

Yes. Many agents start with draft-and-review workflows. Your team can approve, edit, or reject actions before anything is sent, updated, or triggered in a live system.

Question 9

What data access do you need?

Accepted Answer

Only the access required for the agreed workflow. We map the specific systems, fields, documents, and permissions the agent needs, then connect through approved APIs or secure access patterns.

Question 10

Do you copy or store our business data?

Accepted Answer

Our default approach is to keep your data in your existing infrastructure and connect to it securely. If a project needs indexing, caching, or derived data storage, we define that explicitly before build.

Question 11

Can the agent respect user permissions?

Accepted Answer

Yes. Permissioning is part of architecture. The agent can be designed to limit answers or actions based on roles, source-system permissions, approved data scopes, and workflow rules.

Question 12

Can you work with sensitive or regulated data?

Accepted Answer

Often, yes, but it depends on the requirements. We identify security, compliance, retention, and access constraints during Discovery so the architecture fits the risk profile before implementation starts.

Question 13

What happens after the agent goes live?

Accepted Answer

We can stay on as your AI operations team. That means monitoring, accuracy tuning, failure review, usage analysis, improvements, and expansion as your team learns what the agent should do next.

Question 14

What if the agent gets worse over time?

Accepted Answer

Operations gives the agent an active owner. We review quality, usage, edge cases, retrieval behavior, and data changes, then make agreed adjustments and report on what changed.

Question 15

Can you run several agents over time?

Accepted Answer

Yes. You can start with one high-value workflow and scope additional agents over time. Operations can coordinate monitoring, reporting, and priorities; new agents and major expansions are separate build scope.

Question 16

How do we know whether the agent is working?

Accepted Answer

We define success metrics before launch and report against them after launch. Depending on the workflow, that can include usage, accuracy, response quality, time saved, task completion, or business outcomes.

Question 17

Do you work fixed-price or hourly?

Accepted Answer

Discovery and scoped builds are fixed-price once requirements are clear. AI Operations is a monthly retainer for monitoring, agreed improvements, reporting, and support.

Question 18

Do we need a monthly retainer?

Accepted Answer

Not always. If you only need a build, we can scope that. If the agent is business-critical, changes often, or needs ongoing accuracy and monitoring, an Operations retainer is usually the better fit.

Question 19

What is the first step?

Accepted Answer

Book a free 30-minute call. We'll learn where AI might fit, what systems and data are involved, and whether Discovery, an agent build, or Operations is the right next move.

Questions before we build

Before the first call

What we build

Access, permissions, and control

After launch

How engagements are structured

AI Strategy & Audit

AI Agents

AI Operations

Have a question we didn't answer?