Big Ideas 2026: Voice Agents and High-Stakes Trust

Voice agents are being deployed at scale across healthcare, banking, and recruiting. Healthcare sees agents handling patient-facing calls including post-surgery follow-ups and psychiatry intake. Financial services adopt voice AI because it outperforms humans on compliance tracking. The technology has achieved sufficient accuracy and latency improvements that some companies now slow down agents to sound more human.

• •Healthcare deploys voice AI for scheduling, post-surgery follow-ups, and psychiatry intake calls due to high turnover and staffing difficulties
• •Banking and financial services use voice agents because they never violate compliance rules and performance can be tracked over time
• •Recruiting uses voice AI for instant interviews across retail, entry-level engineering, and mid-level consulting roles
• •Voice AI excels at multilingual conversations and heavy accents, often outperforming human transcription accuracy
• •Some geographies still have human labor cheaper than best-in-class voice AI, creating regional variation in adoption

The voice AI market is structured as an industry rather than a single market, with winners emerging at every layer of the stack. BPOs and call centers face varied threats - some will transition by adopting AI to offer cheaper prices and higher volume, while others may face harder disruption. End customers often prefer buying solutions rather than implementing technology themselves.

• •Call centers using AI will survive by offering lower prices and higher volume, while those that don't adopt face existential threats
• •Voice AI is an 'industry not a market' - expect winners across and at every layer of the stack
• •Government services (DMV, non-emergency calls) represent untapped opportunity if 911 calls can be handled with voice AI
• •Consumer voice AI for health and wellness is emerging in assisted living facilities as companions and wellness trackers
• •Platforms like Eleven Labs enable founders to test voice creation and agent capabilities to understand what's possible

A new customer segment called 'Healthy Mouse' is emerging - healthy individuals who engage frequently with the healthcare system for proactive monitoring, contrasting with 'Healthy YAOs' who only see doctors annually. New payment models and business models are supporting this shift, driven by consumer demand for wellness-minded behaviors and enabled by AI-native capabilities.

• •Traditional healthcare serves 'Sick Mouse' (frequent sick interactions) and 'Healthy YAOs' (annual checkups only)
• •Healthy Mouse represents high-engagement healthy individuals using ongoing monitoring to prevent chronic disease
• •New payment models and AI-powered challenger clinics are emerging to serve this proactive wellness segment
• •Infrastructure players are building utility layers to support continuous health engagement
• •Both traditional insurers/providers and new AI-native startups are competing for this market

Healthcare monitoring is shifting from static, point-in-time measurements to continuous longitudinal signals. CGMs pioneered this transition for blood glucose, and the paradigm is expanding to blood pressure and other biomarkers. However, this creates an evidence gap - technological capabilities outpace our understanding of how to interpret continuous signals, risking false positives and 'incidentalomas' that cause unnecessary anxiety and costs.

• •Traditional vitals like blood pressure are measured statically and inaccurately; continuous monitoring provides more actionable longitudinal data
• •CGMs demonstrated the value of moving from sporadic to continuous measurement of blood glucose levels
• •Wearables (Apple Watch, Oura Ring, Whoop) have achieved mainstream adoption for steps, cardiovascular signals, and sleep tracking
• •Risk of 'incidentalomas' - finding deviations that aren't actionable but cause mental distress and unnecessary diagnostic costs
• •Critical opportunity: build infrastructure to run N-of-one real-world studies correlating continuous data to phenotypes and disease status
• •Evidence generation must catch up to technological capabilities through feedback loops interpreting findings

Consumer AI is evolving beyond productivity tools toward helping people feel connected and understood. The core emotional need is 'wanting to be seen' by others. AI can facilitate existing in-person relationships and create new interaction models, including AI-to-AI communication where 'your AI talks to my AI' to enable conversations that wouldn't otherwise happen.

• •2026 marks shift from AI helping you work to AI helping you stay connected with people you love
• •Core emotion being addressed: wanting to be seen and feel connected to others
• •Two use cases: people using AI for digital connection directly, and using AI to facilitate existing in-person relationships
• •Net new interaction model: AI-to-AI communication ('my guy talking to your guy') to prompt check-ins and open new conversations
• •Products must quickly understand users through digital footprint ingestion, photos, online/offline conversations rather than manual narration

Despite incumbents having platforms and networks, startups can win in consumer AI by creating net new user interaction models that don't natively live in existing platforms. AI enables different atomic units and creative outlets that are difficult for incumbents to replicate, creating opportunities for startup differentiation.

• •AI brings net new user interactions that may be difficult for incumbents to replicate
• •Startups can win by creating atomic units and creative outlets that look different from current platforms
• •Net new interaction models that don't natively fit existing platforms create startup opportunities
• •Success depends on addressing core emotions (being seen, feeling connected) rather than just features

As AI moves into human relationships and high-stakes decisions across voice agents, healthcare monitoring, and consumer connection, the key differentiator shifts from novelty to trust, reliability, and measurable outcome improvement. Success requires systems that can be tracked, made compliant, generate evidence, and genuinely improve real-world results.

• •Voice agents win through measurable compliance and reliability tracking in regulated environments
• •Healthcare AI requires evidence generation and interpretation layers to match continuous data collection
• •Consumer AI must prove it improves real relationship outcomes, not just provide novel features
• •The common thread: trust and reliability matter more than technological novelty as stakes increase