Custom workflow orchestration
Some workflows do not fit the standard playbook. Run them anyway.
Claims intake. Multi-region subscription cancellations. Dealer triage across OEM brands. Compliance routing across languages. Document processing with attachments and exception handling. These are not edge cases. They are the operational work that does not fit a packaged template, but still needs to run correctly, at scale, without a team managing it by hand. Ortoo Orchestrator handles each of these as one defined workflow inside Salesforce.






Ortoo Orchestrator runs the workflows that do not fit a template.
It handles that process as one workflow: unstructured intake becomes a structured record, AI interprets where it helps, deterministic logic decides and acts, and every step is logged in Salesforce. You get a workflow that holds up at volume, without a team managing it by hand.
The problem is not the individual tools. It is the absence of a single execution model that coordinates everything from intake to resolution. Ortoo Orchestrator provides that model.
The failures repeat, even when the workflow is one-of-a-kind.
The failures happen in the spaces between the tools: between unstructured intake and structured records, between AI outputs and downstream actions, between approval and execution, between systems and teams. Work gets stuck there, someone steps in to push it forward, and the team ends up managing the workflow instead of the workflow handling the work.
Problem
Workflows span Salesforce, external systems, AI tools, and teams, with logic scattered across Flows, Apex, rules, and integrations, so they break in the handoffs.
Fix
One workflow definition coordinates every step across every system, and existing Flows and rules become components inside it.
Problem
Inbound work arrives unstructured: emails, forms, attachments, free-text fields.
Fix
AI extraction and classification turn unstructured input into structured records at intake.
Problem
AI outputs need follow-up coordination that nobody owns, so they sit or get chased by hand.
Fix
AI runs as a workflow step with the next step already defined, so the workflow continues automatically and nothing waits to be noticed.
Problem
Approvals, escalations, and exceptions depend on people watching the work.
Fix
Configured criteria fire automatically, and human review is a workflow step, not a bottleneck.
Problem
Each new requirement means another Flow or integration, and volume growth forces headcount growth.
Fix
New steps are added to the same workflow definition, and throughput scales with configuration, not team size.
Problem
Operations cannot see how the workflow actually executed end to end.
Fix
Every step logs on the Salesforce record, so workflow execution becomes traceable.
Structured intake. One workflow definition. Ops in control.
Ortoo Orchestrator runs the whole sequence as one workflow definition: AI interprets where it helps, deterministic logic decides and acts, and operations changes the rules without code.
Unstructured intake becomes structured records.
The workflow runs as a defined sequence: intake, classification, enrichment, routing, action, follow-up. Each step is handled by a specialist agent. AI is applied where interpretation helps, extracting intent from an email, classifying a document, matching an account from a URL or domain. Deterministic logic handles decisions and actions. Every step is traceable.
How Ortoo Orchestrator worksEverything in one workflow definition.
The workflow uses existing Salesforce data, Flows, queues, and security model. External systems (ERPs, billing platforms, data lakes, messaging tools, AI providers) are called inside the same workflow definition. No middleware bolted on. No second source of truth. Logic that is currently scattered across Flows, Apex, assignment rules, and integrations is brought into one place. Existing components are not replaced; they become steps inside the orchestration.
Operations teams control the workflow without engineering cycles, all from the product UI: no Setup access, no code, no IT queue.
Steps, rules, and AI behaviour are configurable without a development sprint. Workflows version, branch, and adapt as conditions change. When the business changes, the team that owns the work adjusts the workflow. Not the IT team. Not after a six-week change request.
What changes day to day.
Workflows ship in weeks, not quarters.
Start with the one that breaks most, expand from there.
Operations owns the workflow, not engineering.
Steps, rules, and AI behaviour are configured without code.
Unstructured intake stops blocking work.
Emails, forms, and attachments become structured records automatically.
Throughput scales without headcount.
Volume grows on configuration, not team size.
Every step is traceable.
Each action logs on the Salesforce record.
Cost stays predictable.
Pricing follows work completed, not token usage.
One way to handle the work, whatever the workflow.
Service and revenue workflows follow recognisable patterns. Claims intake, compliance triage, dealer support across OEM programs, document classification, subscription operations: each is different. But every one runs across Salesforce data, external systems, AI steps, approvals, and people. Ortoo Orchestrator coordinates all of it inside one workflow definition. The shape of the work changes; the way it is executed does not.
Execution layer
One workflow definition
Start with the workflow that breaks most. Expand from there.
Begin with the process that costs the team the most hours or fails the most often. Validate on real production volume. Then expand step by step. No replatforming. No replacement of what is already in place. The first workflow is live in weeks, not quarters.
Read this inbound email, extract the account, and route the case to the right pod.
Salesforce-native. Cross-system by design.
Ortoo Orchestrator runs natively inside Salesforce on the existing data model, security model, and automation layer. No external workflow engine. No data duplication. Every step in every workflow logs on the Salesforce record. Where workflows reach external systems, ERPs, billing platforms, data lakes, AI providers, messaging tools, Orchestrator handles those callouts inside the same workflow definition. The result is one execution path that spans the full operational stack, with full visibility from end to end.
Coordination layer
Ortoo Orchestrator
Salesforce
Objects (standard and custom), Flows, queues, rules, security model
External stack
ERPs, billing systems, data platforms, messaging tools, AI providers, MCP servers
AI does the interpretation. Deterministic logic does the rest.
Custom workflows almost always arrive with unstructured input: emails, attachments, free-text fields, multi-language content. That is where AI earns its place, interpreting what the content is, what it contains, and what should happen next. Once the content is structured, deterministic logic takes over. Routing rules, business policies, escalation criteria, system actions. Each AI capability runs on the categories the operations team configured. None runs as a default applied to everything. And every AI decision logs on the Salesforce record, so the audit trail covers not just what the workflow did, but why.
Extract the key fields from this email and attachments.
Ortoo Orchestrator supports any LLM provider.
Customer in action
Custom orchestration in action: Cars.com
Cars.com's Dealer Inspire division supports dealers and OEM partners across Ford, GM, FCA, Toyota, Honda, BMW, and Mercedes-Benz. 14,000+ cases arrive monthly with account information buried in unstructured email content, and routing requirements that span POD groups, MAG groups, agent skills, capacity limits, and AWS Connect availability. Standard Service Cloud routing could not reach the decisions the operation needed.
Operational hurdles
25 senior agents spending 60% of their time, 2,800 hours monthly, on manual case triage and assignment
Account information buried in unstructured email content; URL formats varied, signatures contaminated extraction
Multi-dealer emails requiring automatic case splitting with parent-child relationships
Routing logic spanning POD, skills, capacity, business hours, and AWS Connect agent availability
Subscription cancellations and privacy requests requiring manual data extraction across multiple workflow steps
Measurable results
33,600 annual hours reclaimed, 16 FTE equivalents released from administrative routing
Near 100% first-touch routing accuracy across 6 production use cases
100% coverage of revenue retention signals in cancellation workflows
Subscription cancellation processing reduced from hours to minutes
Manager self-service control of routing logic, AI rules, and escalation thresholds, without IT involvement
When operational workflows do not fit a packaged template, the requirement is not a different product. It is a different way to run the work.
A Property Group
Real estate sales orders routed by territory rules, with notifications and downstream Flow calls running inside the same workflow.
Read the case studyByMyCar
Multi-source lead intake for a dealership network. Up to 100 Salesforce fields auto-populated in one step before assignment.
Read the case studyOppFi
Loan application workflows with custom assignment logic that keeps queues balanced and credit decisions moving without manual triage.
Read the case studyRelated pages
Service workflow orchestration
For case, escalation, and resolution workflows that follow a recognisable pattern.
Read moreRevenue workflow orchestration
For lead, opportunity, and territory workflows.
Read moreInsurance claims orchestration
Claims intake, classification, evaluation, and resolution as one workflow.
Read moreOrtoo Orchestrator
The system that runs every workflow on this page.
Read moreWhy orchestration, not more automation
The strategic case for orchestration over additional automation layers.
Read moreCommon questions from operations and platform leaders.
Our workflow does not look like anything on your service or revenue pages. Does that matter?
No. Service and revenue workflows are the most recognisable shapes, but the system runs operational work regardless of shape. Customer workflows in production today include compliance triage across 16 teams and multiple languages, claims processing with attachment-driven routing, subscription operations with ERP integration, dealer support across OEM programs, document classification at intake, and multi-language case intake with custom capacity logic. The operating model is the same across all of them. The steps are configured to the specific work.
How does this differ from building more Salesforce Flows?
Flows automate individual steps. Orchestration defines how steps connect: intake, AI interpretation, routing decisions, system actions, approvals, escalation, follow-up, all as one execution path with a full audit trail. Existing Flows do not get replaced; they become components inside the orchestration. The orchestration layer is what was missing.
We already have AI tools and assignment rules. Why add orchestration?
AI tools and assignment rules each do one thing well. They do not coordinate what happens next. Orchestration is the coordination layer: AI runs as a step, its output triggers the next step, deterministic logic handles decisions, escalations fire on configured criteria, and the whole execution is visible end to end. Without that layer, AI outputs and routing decisions sit in isolation and somebody has to chase them.
How does this work with our existing external systems, ERPs, billing platforms, data lakes, messaging tools?
External system calls happen inside the workflow definition. Ortoo Orchestrator handles the callout, processes the response, and feeds it into the next step. MCP tools, REST APIs, and platform-native integrations are all supported. There is no separate integration layer to build or maintain.
How long does it take to get one custom workflow running?
Most teams start with the workflow that breaks most often or costs the most hours, usually the one the team has been manually managing the longest. Validation on real production volume typically takes weeks, not quarters. Cars.com went live with their first phase in Q4 2024. Assent Compliance was processing cases from day one of go-live and reached 120,000+ records within three months. No replatforming. The workflow runs natively on the existing Salesforce instance.
What AI is included, and how is it governed?
AI capabilities cover content extraction, classification, document understanding, account and record matching, and missing-information handling. Each runs on categories the operations team configured; none runs as a default. All AI decisions log on the Salesforce record. Ortoo Orchestrator supports any LLM provider.
See where your trickiest workflow breaks down, and how to fix it.
Book a 30-minute session. Bring the workflow that breaks most often, costs the team most hours, or does not match anything in a packaged template. We will walk through how it runs today and where one connected workflow would change the operational picture.
Installs natively into Salesforce. Start with one workflow, expand as the team is ready. Pricing follows work completed, not token usage.