Interview:. Adeptia automates the handling of data exchanged between businesses using, for example, Electronic Data Interchange (EDI) methods. It calls this first-mile data and speeds adoption and processing. Co-founder and chief innovation officer Deepak Singh contacted B&F to argue that the idea AI agents could replace SaaS applications is misconceived because it does not recognize how things like EDI-anchored SaaS apps work deep in an organization’s infrastructure and they can’t simply be uprooted or replaced. We asked him some questions to find out more.
Blocks & Files: What do you think of the idea that AI agents will replace SaaS apps? Where did it come from? Is it feasible?
Deepak Singh: This idea reflects a fundamental misunderstanding of what enterprise SaaS actually does. This idea of AI agents disrupting and replacing SaaS applications comes from a recent interview in which Microsoft CEO Satya Nadella introduced this possibility. But we believe that Satya’s comments come from Microsoft’s strategic positioning – they have a perceived AI advantage through OpenAI and want to create market urgency. But it’s oversimplified.
For example, with regard to CRUD (Create, Read, Update, and Delete). The reality is that SaaS isn’t just “CRUD with a UI.” In fact, it is decades of domain expertise, compliance frameworks, and complex multi-party orchestration codified into reliable systems. At Adeptia, we process millions of transactions daily from thousands of partners in hundreds of formats.
An AI agent might help query that data, but it can’t magically handle a malformed EDI file from a supplier or ensure HIPAA compliance in healthcare data processing. This is a five to ten-year evolution, not a two-year revolution. AI will transform how we interact with SaaS, not replace it.
Blocks & Files: What do SaaS apps do as well as handle business operations (CRUD and logic)?
Deepak Singh: Modern SaaS platforms are sophisticated orchestration engines for real-world complexity. Beyond basic operations, they provide:
- Domain Intelligence: Years of industry-specific knowledge embedded in the platform.
- Trust and Compliance: Audit trails, security certifications, regulatory frameworks.
- Multi-Party Coordination: Managing relationships between dozens of stakeholders.
- Exception Handling: Dealing with the 20 percent of cases that break standard rules.
- Data Quality Assurance: Ensuring information accuracy before it impacts operations. At Adeptia, we don’t just move data but we understand that a date format difference between partners can break an entire supply chain, and we handle it automatically.
Blocks & Files: Do SaaS apps need to ingest partner EDI files, validate compliance rules, and manage exceptions? Can you provide examples?
Deepak Singh: Absolutely. This is the unglamorous reality of enterprise operations that AI evangelists often ignore. Examples from our customers include insurance. A major carrier receives claims from 500-plus brokers, some via modern APIs, others through EDI X12 837 files, many as PDFs. Each must be validated against state regulations, coverage limits, and fraud patterns. Manufacturing: auto parts suppliers send advance shipping notices in dozens of EDI variants. One character wrong in a part number can shut down a production line.
Take healthcare: patient data arrives from labs, providers, and pharmacies in HL7, proprietary formats, and faxed documents. HIPAA compliance isn’t optional – it’s federal law. These aren’t edge cases but they’re daily operations for thousands of enterprises.
Blocks & Files: Could agents do these things?
Deepak Singh: AI agents excel at natural language queries and pattern recognition, but they struggle with the deterministic requirements of business operations. Here’s why:
- Accountability Gap: When an AI agent mis-processes a million-dollar order, who’s liable?
- Compliance Black Box: Regulators require explainable, auditable decisions, not AI probability scores.
- Format Brittleness: AI might read a PDF brilliantly but fail completely on a malformed EDI segment.
- Exception Handling: Business exceptions often require human judgment within regulatory constraints. Could an AI agent learn to process EDI data? Perhaps. Would you trust it with your supply chain? That’s the real question.
Blocks & Files: Tell me about your idea of a three-layer SaaS app model.
Deepak Singh: The future of SaaS is a three-layer architecture that combines the best of AI innovation with enterprise reliability:
- Layer 1 – Intelligent Interface: AI agents and natural language interfaces that make software accessible to everyone. Ask questions, get insights, trigger workflows – all through conversation.
- Layer 2 – Orchestration & Governance: The SaaS platform that ensures AI decisions align with business rules, compliance requirements, and operational constraints. This is where domain expertise lives.
- Layer 3 – Execution & Integration: The reliable backbone that actually moves data, processes transactions, and connects systems. This handles the messy reality of thousands of formats and protocols.
Blocks & Files: Where do agents play in this and what do they do?
Deepak Singh: AI agents are powerful in Layer 1 and assist in Layer 2, but they don’t replace the foundation. They:
- Simplify Interaction: Users can say “Show me delayed shipments from Supplier X” instead of navigating complex UIs.
- Surface Insights: Identify patterns humans might miss across thousands of transactions.
- Suggest Actions: Recommend optimizations based on historical data.
- Automate Routine Decisions: Handle the 80 percent of cases that follow standard patterns. But, critically, they operate within the guardrails set by the platform. They’re copilots, not autopilots.
Blocks & Files: What does Adeptia bring to the table?
Deepak Singh: Adeptia represents the future Satya’s missing. We are already AI-enabled AND enterprise-reliable. Our platform:
- Handles Reality: We process any data format from any source with 99 percent accuracy, whether it’s a modern API or a coffee-stained fax.
- Embeds AI Intelligently: Our AI learns from patterns to suggest mappings and catch anomalies, but always within business constraints.
- Enables Humans: Business users can onboard partners without coding, while IT maintains governance.
- Scales Trust: From 10 to 10,000 partners, maintaining compliance and reliability, we’ve spent 20 years encoding the complexities of real-world data exchange. AI enhances our platform; it doesn’t replace it. That’s why Fortune 500 companies trust us with their mission-critical operations. The future isn’t agents replacing SaaS but it is intelligent SaaS platforms that leverage AI while respecting the realities of enterprise operations. We’re not waiting for that future. We’re delivering it today.
