FerrataAtlas

The operating model

Your receivables operation, modeled as AI-native functions.

Atlas turns the way your team runs receivables into something you can see and change: agents that each own a kind of work, the rules that move accounts between them, and the context they work from. You pick options; nothing is built by chat.

Your operation, modeledEvery open account in exactly one agent
15 days past duePromise madeDispute · short payBroken · card declined90 days past duePaid in fullUnresolved at 90 daysStandard follow-up4 open · $87,360Collections3 open · $355,600Promise to pay3 open · $220,850Dispute & deductions2 open · $30,630Charge-off & Legal1 open · $19,870
Illustrative data

An agent

Each agent owns one kind of work.

An account sits with exactly one agent at a time. Triggers move it; paying in full takes it out.

Triggers
The events that bring an account in: due soon, 15 days past due, a promise, a dispute, a declined card.
Priority
When two triggers fire, the higher-priority agent takes the account.
Sequence
The steps, by day and channel, at most one touch per customer per day.
Channels and tone
Email, text and voice, in the tone you choose: friendly, direct, firm or formal.
Contact hours
When the agent may reach out, inside the 8am to 9pm local window for calls and texts.
Hand-offs
What sends the work to a person: a request for one, a legal mention, a second broken promise.

Context, layered

Every step works from the right context, in order.

  1. 01

    Ferrata

    The shared capabilities every company gets, such as the AI Credit Rating.

  2. 02

    Your company

    Who you are, how you speak to customers, and what your team has taught Atlas.

  3. 03

    Atlas and the agent

    Your receivables goals and guardrails, then the agent's own instructions.

  4. 04

    Locked rules

    Hard stops no setting can change: contact hours, consent, disclosure.

  5. 05

    The transaction

    The account's invoices, payments, promises and history, read exactly from the record.

Change with a preview

See every account that would move before you save.

Edit an agent in place and Atlas shows every changed field and every account that would join, leave or be replanned. Each save is a new version, and accounts already in the agent follow the new settings from their next step without repeating one.

One call, routed by what the caller needsInbound and outbound
  1. Caller identifiedNumber matched to the account and contact
  2. VerifiedNo invoice detail until the customer is verified
  3. Context loadedBalance, invoices, promises, disputes, rating, history
  • A date to payPromise to pay confirms it in writing
  • Pay nowA secure payment link by text
  • Something is wrongDispute & deductions opens it with the evidence
  • Wants a personHanded to your team with a summary
Illustrative data

What runs without anyone

Atlas does the work. You decide what waits.

Once modeled, the operation runs. You decide which steps wait for a person.

Atlas runs on its own. You flag any step for review, and flagged work waits in the Approval Queue for an Approver, Admin or Owner. Charge-off and legal waits for a person on every step.

On its own

  • Moving accounts between agents as triggers fire
  • Drafting and sending each step in its sequence
  • Rating every account nightly from its own history
  • Holding to contact hours, consent and one touch a day

Waits for a person

  • Any step you flag for review
  • Every step in Charge-off and legal
  • Hand-offs your agents are told to make

FAQ

Modeling your operations, run end to end.

How does Atlas decide which agent works an account?

Each Atlas agent has triggers and a priority. When an event fires, such as a promise or 15 days past due, the highest-priority matching agent takes the account. An account sits with exactly one agent at a time, and paying in full takes it out.

Can we build our own agents in Atlas?

Yes. Atlas agents are built by picking options, never by chat: triggers, channels, tone, cadence, contact hours, hand-offs, review and priority. A test drive previews every channel before the agent can be activated.

What happens to accounts when we change an agent?

Atlas previews the change first: every changed field and every account that would join, leave or be replanned. A save becomes a new version, and accounts already in the agent follow the new settings from their next step without repeating one.

What context does an Atlas agent use when it writes to a customer?

Atlas layers context in a fixed order: Ferrata's shared capabilities, your company, your receivables settings, the agent's own instructions, locked rules, and the account's own invoices, payments and history, read exactly from the record.

Which parts of the operation run without a person?

Atlas runs on its own: it moves accounts, sends each step and rates every account. Any step you flag waits in the Approval Queue for an Approver, Admin or Owner, and every step in Charge-off and legal waits for a person.

Can our team teach Atlas how we do things?

Yes. Your company's Owner can add learnings in Atlas, and they apply to your company only. Atlas uses them in the company layer of every agent's context, alongside the rules and history it already reads.

See the platform work on your own receivables.

Thirty minutes on your own aging report. You leave with a view of how Atlas would work it, and an honest answer if the sequencing is wrong and you should fix something else first.

Your credit policy, terms, tone and escalation rules are your advantage, and they stay yours. Atlas is configured to the way you run receivables. You are not handing over a playbook or fitting into a template.