A Bengaluru startup barely a year into its formal existence just secured ₹32 crore (roughly $3.8 million) to automate the exhausting machinery of customer engagement at Indian banks and insurers. Desible.ai announced the seed-plus round Monday, led by Prime Venture Partners with backing from existing investor Invention Engine, though the company sidestepped questions about its valuation.
The funding will bolster what the startup calls "agentic AI capabilities"—software that doesn't just respond to customer queries but orchestrates entire workflows across voice, WhatsApp, SMS, and email without human handholding. Desible, incorporated on December 14, 2023 under the legal name Basal Analytics Private Limited, plans to deepen compliance infrastructure and accelerate product development for financial-services clients, according to an October 7 statement.
The company's pitch is straightforward. Financial institutions need to automate repetitive customer touchpoints—policy renewals, collections calls, welcome messages, lead qualification—and Desible claims to offer more than 25 pre-built workflows spanning revenue, risk, underwriting, service, and claims functions. Inc42 reported that the platform integrates with familiar CRM systems like Salesforce, Zoho, HubSpot, and SAP, as well as telephony providers including Twilio and Ozonetel.
"Voice AI is our entry point, not our destination," cofounder Uttam Tiwari told ETEntrepreneur, hinting at broader ambitions beyond call automation.
Desible's website displays SOC 2 Type II and ISO 27001 certifications, standard badges in enterprise software, though the company didn't furnish independent verification documents when asked. It also claims compliance with Indian regulatory frameworks: IRDAI for insurance, TRAI/DLT for telecom, RBI for banking, and the country's Digital Personal Data Protection Act. Whether these claims hold up under scrutiny at scale remains to be seen.
The founders bring a mix of enterprise software and edtech exit experience. Tiwari, who handles investor relations and growth, spent 12 years in enterprise sales and investment banking and previously cofounded an edtech venture he later exited. Cofounder Omkar Raikar leads technology and product; his background includes a dozen years in machine learning and enterprise software at CRISIL and Edelweiss, plus another edtech exit. Both appear as directors on MCA aggregator TheCompanyCheck, with records last refreshed in August.
Desible reports processing more than 10 million customer engagements monthly and interest from over 40 financial institutions, according to company statements, though it hasn't named clients publicly. LinkedIn lists the team at between two and 10 employees, a figure that may lag reality given typical hiring patterns at funded startups.
The competitive landscape is dense. Navana.ai, another voice-AI player targeting financial workflows, raised ₹40 crore (about $4.2 million) in a Series A round led by Ronnie Screwvala in September, according to Seedtable. Older competitors include Skit.ai, which pulled in a $23 million Series B in 2021 for voice automation, PRNewswire reported at the time.
Gaurav Ranjan, a principal at Prime Venture Partners, framed the investment as a bet that generic AI models won't suffice in regulated industries. "AI adoption in financial services won't be driven by generic models alone," he said in the announcement—a recognition that verticalized tools built for compliance-heavy environments may have staying power even as foundation models proliferate.
For Prime Venture Partners, the Desible round extends a pattern of backing vertical AI tools tailored to regulated sectors. Invention Engine's earlier stake in the company—size and timing undisclosed in available coverage—signals the firm saw enough traction to double down.
Whether Desible can carve out durable differentiation in a market crowded with automation vendors will depend on execution. Financial institutions move cautiously, and enterprise sales cycles stretch long. The startup has capital and a clear wedge. What it doesn't yet have is public proof at scale.
