The problem with most AI agents, Emergent would argue, isn't that they don't work. It's that nobody remembers to open them.
So the startup—backed by SoftBank Vision Fund 2 through its $70 million Series B—is trying something different. Rather than build yet another app competing for attention on crowded home screens, Emergent launched Wingman—an autonomous AI agent that operates inside WhatsApp, Telegram, and iMessage. The same places people already spend half their day, thumbs hovering over glass.
The concept carries a certain logic. Why force users to context-switch between a specialized agent dashboard and the communication tools they reflexively check? Instead, you text Wingman the way you'd ping a colleague: "Schedule coffee with Sarah next week" or "Draft a follow-up for that prospect from Tuesday." Behind the scenes, the agent rummages through your calendar, email, CRM system, and whatever else you've granted access to, executing tasks while you sleep or sit in meetings.
When something consequential comes up—a mass email about to go out, say, or a change to important data—Wingman surfaces a request for approval in the same chat thread. No separate platform to monitor. No new muscle memory to develop.
"The most powerful AI agent is the one people will actually use," the company said in late April, a line that doubles as mission statement and market diagnosis.
Whether Emergent can deliver on that thesis depends partly on factors beyond its control. Regulatory fights over messaging platform access, the uneven maturity of autonomous agent technology, user squeamishness about delegating sensitive workflows—all of it shapes the space Wingman is trying to occupy.
Living Where the Users Are
Mukund Jha, Emergent's CEO, told TechCrunch the messaging-first approach stems from observing where attention actually goes. People toggle into WhatsApp or iMessage dozens of times daily. Building an agent layer on top of that behavior, rather than asking for new habits, seemed like the path of least resistance.
Wingman doesn't function as a single bot, exactly. The product deploys what Emergent calls an "operating team" of specialized agents, each handling distinct workflows. Some triage email. Others manage scheduling, conduct prospect research, or draft social content. The system can trigger these agents automatically based on schedules or incoming events, and users can choose from pre-configured personas with names like Jarvis, Judy, and Venus—though tone and behavior settings are customizable.
Integration happens through OAuth sign-in rather than the more technical route of API keys. Connect Gmail, Google Calendar, Slack, your CRM, or any tool from Emergent's directory, and the agents begin working across those systems. The platform retains conversational context and claims to learn individual routines over time, though the company hasn't detailed precisely how that adaptation functions under the hood.
The pitch positions Wingman against a wave of AI agents requiring users to adopt new platforms or master specialized interfaces. For business users already drowning in productivity software, the appeal of consolidating agent interactions into existing messaging threads is evident enough. Whether it works in practice is another question.
The WhatsApp Complication
Here's where things get complicated. Meta banned general-purpose third-party AI chatbots from WhatsApp's Business API starting January 15, 2026—a policy announced the previous October. The restriction specifically targets "LLMs, generative AI platforms, or general-purpose AI assistants" when those capabilities serve as the primary function.
Microsoft pulled Copilot from WhatsApp as a result. Other AI assistants have navigated the ban with varying approaches: some curtailed WhatsApp support entirely, others found workarounds that may or may not withstand scrutiny.
Antitrust pressure complicated the picture. Brazil ordered Meta to suspend the ban in January, and WhatsApp subsequently exempted Brazilian users. Italy secured similar accommodations. In mid-April, the European Commission rejected Meta's proposed paid-access remedy, keeping regulatory pressure on the company to restore broader third-party access.
Emergent announced Wingman on April 15—three months after the ban took effect, into a regulatory landscape still churning. The company's materials state simply that Wingman "lives inside WhatsApp, Telegram, and iMessage," without clarifying whether WhatsApp functionality is geographically limited or depends on any specific technical implementation. (Emergent did not respond to questions about its WhatsApp access strategy.)
It's an uncomfortable ambiguity for a product whose central premise involves meeting users on their preferred platforms.
Ambition Meets Reality
The use cases Emergent highlights are deliberately practical rather than speculative: calendar management, email triage, sales outreach sequences, competitive research, content drafting and scheduling, hiring workflow coordination. The agents work asynchronously—churning through tasks overnight or during off-hours—and pause for user confirmation before taking high-stakes actions.
Jha acknowledged limitations in his April interview with TechCrunch, noting the system "struggles with consistency in really ambiguous situations, messy edge cases, unclear goals, or workflows where a lot of human judgment is needed." It's a candid admission that autonomous agents, for all the industry hype, still bump up against fuzzy boundaries.
The company frames its approach as "autonomy with accountability," deploying what it calls "trust boundaries" to separate routine actions from consequential ones. It's an attempt to thread a familiar needle: offering enough automation to feel valuable without demanding the kind of unchecked system access that makes users nervous.
Whether that balance holds depends on execution details that remain opaque. How reliably does Wingman distinguish routine from high-stakes? How often does it misread context or make errors that require cleanup? Emergent hasn't published performance metrics or error rates, and early user accounts remain scarce.
A Crowded Field
Wingman arrives into a market already populated with messaging-based AI agents. OpenClaw and commercial variants like Abacus Claw and PepperClaw offer persistent agents on messaging platforms. SolClaw handles on-chain crypto workflows via WhatsApp and Telegram. Poke, which secured funding in early April, runs AI agents over iMessage, SMS, and Telegram (WhatsApp "limited" due to Meta's policy, per the company's disclosures).
Emergent positions Wingman for business users seeking integration with existing tools, emphasizing user-friendly setup and pre-built integrations rather than technical sophistication. Where OpenClaw targets developers comfortable with open-source tools, Wingman aims at business users seeking plug-and-play automation. The learning curve matters when the product needs to slot seamlessly into existing routines.
The startup claims more than 8 million builders across 190-plus countries and 1.5 million monthly active users as of late April. Those figures, however, reflect Emergent's broader platform—a natural-language development tool called vibe coding that lets non-technical users create and monetize software. Wingman-specific adoption numbers haven't been disclosed, making it difficult to gauge traction for the agent product specifically.
The Money and the Metrics
Emergent closed a $70 million Series B in January at roughly $300 million post-money valuation, co-led by SoftBank Vision Fund 2 and Khosla Ventures. That round followed a $23 million Series A the previous September.
The company claimed $50 million in annual recurring revenue after seven months in operation, then said it hit a $100 million ARR run rate by mid-February. Those are staggering growth figures—if accurate and measured consistently. Moneycontrol noted in April that ARR calculation methods vary widely across startups, and early-stage companies sometimes conflate gross merchandise value, transaction volume, or annualized snapshots of short-term performance with true recurring revenue. Without independent verification or detailed methodology from Emergent, the usual caution applies to such startup-reported metrics.
What It Costs
Wingman is available now at app.emergent.sh/wingman. Users sign in with Google or email, connect their tools, and begin chatting with their agent team. The company offered a limited free trial at launch, transitioning to paid plans—though specific pricing hasn't been publicly disclosed. Some industry blogs have speculated about tiered consumer and professional plans, but those figures remain unconfirmed.
The absence of transparent pricing is standard practice for early-stage B2B software, though it complicates comparisons with competitors and leaves potential users guessing about cost.
The Bigger Bet
Wingman represents a strategic expansion for Emergent, which built early traction on vibe coding. The product applies similar principles to productivity automation: describe what you want done in natural language, let the agents handle the mechanics. It's software for people who don't think like software engineers, perhaps a larger addressable market than the company's initial developer focus.
Whether that vision scales hinges on variables Emergent doesn't control. Regulatory access to messaging platforms. The maturation of autonomous agent reliability. User willingness to trust AI with sensitive workflows. Competitive pressure from better-funded rivals.
For now, the company is wagering that meeting users in their messaging apps—rather than asking them to adopt yet another tool—lowers the activation energy enough to turn AI agents from novelty into habit. It's a reasonable hypothesis. Whether it's correct is the kind of question that gets answered in usage data, not launch announcements.
The regulatory uncertainty around WhatsApp access lingers as a wild card. So does the gap between what autonomous agents promise and what they reliably deliver. Emergent is betting that convenience and familiarity can bridge both. The market will render its verdict in the usual way: quietly, user by user, until the numbers tell a story nobody can argue with.
