Voice AI just landed $61M: What customer support leaders need to know
Giga, a San Francisco startup building voice-based AI agents for enterprise support, raised $61 million in Series A funding led by Redpoint Ventures, with participation from Y Combinator and Nexus Venture Partners. The valuation wasn't disclosed. The founders, IIT Kharagpur alums and Forbes 30 Under 30 honorees Varun Vummadi and Esha Manideep, say the new capital will push deeper into Fortune 100 deployments and team growth.
The voice AI field is packed-from specialised players like ElevenLabs and Vapi to big tech such as Amazon and Microsoft-chasing a market expected to grow from $3.14B in 2024 to $47.5B by 2034. The goal is simple: automate the routine work your team handles on phones every day without tanking CSAT.
Why this matters for support teams
Giga claims it can spin up production-grade voice AI for enterprises in under two weeks. Teams upload call transcripts and policies, and the system builds from what already works. That speeds up time-to-value and reduces the lift on your ops and engineering teams.
The company's core edge is a real-time orchestration layer that can listen, interpret intent, decide next steps, hit internal tools or databases, and reply-all in under half a second. That sub-second loop keeps conversations natural and cuts awkward gaps that frustrate customers.
A live example at scale
Giga is working with DoorDash. If a Dasher can't finish a delivery, the AI keeps a live thread with the Dasher, calls the customer to verify the address, checks policy compliance, and resolves the case in real time. DoorDash reports fewer escalations, faster resolutions, and smoother workflows across teams-important when operating in 40+ countries and serving nearly 50 million people each month.
Language and accent barriers aren't a side note
Voice AI has struggled with nuance, accents, elderly speakers, and people with speech impediments. That's not a minor bug-those issues break trust and blow up handle times. Giga's approach: multilingual LLMs, native-language options, and stored user preferences so each call gets easier over time.
If you serve a global base or older demographics, this is where pilots succeed or fail. Plan a language strategy up front and test with real customer audio, not polished demos.
Security and compliance: healthcare and finance
For regulated industries, Giga deploys on the client's cloud with open-source models, keeping customer data out of Giga's hands. In financial services, it's already automating compliance checks-flagging unusual transfers, collecting confirmations, and maintaining traceable records.
They also cross-reference external data-like verifying property details via third-party databases-to reduce fraud. If your risk team is pushing for controls and audit trails, that's a meaningful checkbox.
How to evaluate voice AI for your operation
- Choose the right call types: Start with routine, high-volume intents (order status, appointment changes, address verification). Avoid emotionally loaded or high-stakes calls at first.
- Require sub-second latency: Anything slower tanks call flow and increases repeats. Ask for live latency dashboards.
- Demand real integrations: CRM, order systems, scheduling, payments, identity verification. No "CSV export" hacks.
- Lock in escalation logic: Clear thresholds for handoff to humans based on sentiment, confidence, or policy triggers.
- Language plan: Support native-language options, accent coverage, and preference memory. Test with real customer accents.
- Compliance and privacy: On-cloud deployment options, PII handling, redaction, audit logs, and data retention policies.
- Operations fit: Who owns tuning and QA? How are new intents added? What's the change management plan for agents?
- Transparent pricing: Understand cost per minute, model usage, and integration fees so ROI is clear.
KPIs that actually matter
- Containment rate: Percentage of calls resolved without human handoff.
- First-contact resolution (FCR): Don't trade speed for repeat calls.
- Average handle time (AHT): Track with and without AI; target a noticeable drop.
- CSAT/NPS on AI-handled calls: Compare to human benchmarks.
- Escalation rate and reasons: Use this to guide training data and policy tweaks.
- SLA adherence: Especially important for regulated workflows.
- Agent experience: Post-handoff efficiency and burnout indicators.
- Error rate: Wrong actions, policy missteps, or compliance misses.
Rollout plan that reduces risk
- Pilot one call type with clear success criteria and a fixed timeline.
- Feed the system high-quality transcripts, policies, and real outcome labels.
- Enable dual recording and run human-in-the-loop QA for the first weeks.
- Route tricky calls to senior agents; capture edge cases to retrain weekly.
- Publish an escalation map agents can rely on; no guesswork.
- Expand only after you see stable containment, CSAT parity or better, and clean compliance logs.
Market snapshot and what to watch
This is a crowded market with specialist startups and tech giants vying for enterprise voice. Giga's pitch is speed, sub-second orchestration, multilingual depth, and regulated-industry deployment. If they keep those promises at scale, the impact on queue lengths and payroll could be meaningful.
Before you sign anything, validate with your own call mix. Ask for a two-week pilot, real integrations, live metrics, and a clear plan for language and escalation. A small, well-chosen rollout will tell you more than any demo.
Useful links
Who's behind Giga
Giga was founded by Varun Vummadi and Esha Manideep, both IIT Kharagpur graduates and Forbes 30 Under 30 alums. The Series A was led by Redpoint Ventures with participation from Nexus Venture Partners and Y Combinator.
Your membership also unlocks:
