RAISE ’25 brought together some of the most influential voices in AI, financial services, cybersecurity, and enterprise automation. The discussions were candid, future-focused, and deeply practical—highlighting not just where AI is going, but how organisations can actually capture value today.
Below are my distilled insights across the most impactful sessions and product showcases.
1. The Future Is Not AI Replacing Humans — It’s AI Choreographing Work
Siavash Alamouti (Mimik) emphasized a crucial shift:
AI systems must be choreographed, not orchestrated. Rules evolve, contexts shift, and static workflows fail. Zero-trust architecture and trust-centric design remain fundamental, but the north star is clear—AI should enhance human relationships, not replace them.
2. Enterprise AI Agents Are Moving from Assistants to Full Task Owners
Sanjay Krishnan (Concentric AI) laid out a simple but powerful contrast:
-
Assistants answer questions
-
Agents complete tasks end-to-end
Enterprise value emerges only when full delegation is possible—with guardrails on both data and human oversight. Most real-world systems now run as multi-agent pipelines, where 10+ micro-agents coordinate towards one outcome.
Examples included lending, document processing (OCR → VLM → extraction), and recommendation agents powered by heavy context engineering.
Cost efficiency is already visible: AI-enabled underwriting costs can drop as low as $0.3 per loan application.
3. Coding Without Coders: The Rise of AI-First Software Development
Dr. Sriram Rajamani showcased how AI is redefining software creation:
-
Non-programmers building production apps
-
Voice-to-workflow automation (WhatsApp-driven apps for NGOs)
-
Spreadsheet agents predicting formulas, analyzing sentiment, and generating macros
-
Context-driven Excel copilots that treat spreadsheets as full “programs”
The message was unmistakable: LLMs reduce the dependency on scarce engineering talent and institutional code knowledge.
4. Pitch Point: The Startups Redefining Enterprise AI
spoofsense.ai
Advanced anti-spoofing using micro-texture detection, skin-light interaction models, and ensemble vision systems.
Backed by Nvidia and Google for Startups.
Codemate.ai
A full-stack coding agent automating the SDLC with guardrails. Hybrid LLM/SLM mode, on-prem ready, with “VIBE coding” for product teams and engineers alike.
Arrowhead.ai
One of the most human-like voice bots in the market. Every POC has converted to a contract.
Key differentiators: emotional intelligence, latency reduction, and hyper-personalized outbound calling.
Byteflow.bot
Open-source AI automation platform with MCP connectors, agent orchestration, document parsing, and API-to-MCP converters. Focused on enterprise agent deployment with token-based monetization.
Neurofin.ai
Focused on lending workflows—CIBIL reading, Aadhaar/KYC verification, JSON extraction pipelines, and LOS integration.
5. Cyber-Resilient Data Strategies Are Becoming Non-Negotiable
Cohesity outlined the architecture required for AI-ready, regulation-friendly data ecosystems:
-
Immutable copies
-
Multi-media backups
-
Vector-indexed sensitive data detection
-
On-prem “Private GPT” options
-
Redaction pipelines for PII
-
Time-series digital twins for AI data lakes
The message was clear: no single point of failure—no single point of deletion—ever.
6. AI Regulation: Trust, Ethics & Accountability Take Center Stage
Financial services leaders stressed:
-
Understandable and explainable AI by design
-
Human accountability in every loop
-
Deep literacy and governance committees across functions
-
Guardrails that evolve with model failures
-
A shift from “compliance” to customer confidence
Case studies like Air Canada’s chatbot legally committing discounts underscored the operational risk.
Key takeaway: “The model made me do it” will never stand in BFSI.
7. How AI Will Reshape Financial Services in 5 Years
Major insights included:
-
70% of white-collar workers already use AI, but few organisations connect it to revenue
-
Voice agents deliver 60% cost savings in BFSI
-
Agentic AI will dominate compliance workflows
-
India imports 90% of AI chips; vertical stack control will be strategic
-
LLMs depreciate in under nine months
-
Core roles that won’t disappear: customers, teachers, politicians
The overarching message: This isn’t a race—there will be multiple winners if India scales responsibly.
8. Customer Service, Portfolio Management & Collections Will Be Agent-Led
As Gnani.ai demonstrated:
-
Non-coders can build domain-specific agents in 2 hours
-
60% guardrails are ideal for responsible enterprise agents
-
RAG on customer data will fuel hyper-personalized cross-sell
-
Voice AI agents will become emotional, contextual, and avatar-powered
-
Settlement negotiations and customer conversations will be largely agent-to-agent
AI is not just for “collect”—it is for care: high-speed, high-sensitivity engagement at scale.
Final Thoughts
Across 3,000+ in-person attendees and thousands more online, one message echoed through RAISE ’25:
AI is no longer experimental.
It is a productivity multiplier, a business model enabler, and a competitive necessity.
Those who don’t adapt—won’t remain relevant.