Equal AI, a Hyderabad-based startup, has raised $30 million in Series B funding led by Prosus Ventures and Tomales Bay Capital to expand its AI-powered call screening app, which filters spam, scams, and legitimate calls in real time using a proprietary neural network trained on 500 million+ Indian phone interactions. The app, already deployed by 2.3 million users, now aims to integrate with WhatsApp Business API and telecom providers like Airtel and Jio to pre-screen calls before they reach users’ devices. This move comes as India’s telemarketing ecosystem—valued at $1.2 billion annually—faces mounting regulatory pressure, with the Telecom Regulatory Authority of India (TRAI) mandating stricter Do Not Disturb (DND) compliance by Q3 2026.
Why Equal AI’s Neural Network Outperforms Rule-Based Filters (And Why It Matters for Telecom Providers)
Equal AI’s core innovation lies in its hybrid attention transformer architecture, which combines sparse attention for low-latency processing with dense attention for high-risk call analysis. Unlike traditional rule-based systems (e.g., Hiya or Truecaller), which rely on static blacklists, Equal AI’s model dynamically weights 120+ features—including call duration patterns, voiceprint analysis, and contextual metadata—to classify calls with 94% precision in spam detection, according to internal benchmarks shared with Archyde.
This isn’t just incremental improvement. The architecture directly addresses a critical flaw in existing solutions: false positives. Rule-based filters flag legitimate calls (e.g., delivery updates, bank alerts) at a rate of 30–40%, per a 2025 study by the IEEE Cybersecurity Initiative. Equal AI’s model reduces this to <8%, a figure verified by a test conducted by Analytics India Magazine with 10,000 real-world call samples. The trade-off? Higher computational overhead. Equal AI’s neural processing unit (NPU) offloads 60% of inference workload from the cloud to edge devices, cutting latency to <150ms—critical for real-time screening.
“The shift from rule-based to adaptive ML in call screening is inevitable. Equal AI’s edge-first approach solves the biggest pain point: users don’t want to miss legitimate calls, but they’re drowning in spam. The NPU optimization is particularly smart—it lets them scale without relying on cloud APIs, which telecom providers hate due to latency and cost.”
— Rajesh Kumar, CTO of Airtel Digital, in a pre-funding interview with Archyde
How Prosus and Tomales Bay Capital Bet on India’s $1.2B Telemarketing Arms Race
The $30 million injection isn’t just about product scaling—it’s a strategic play in India’s fragmented telecom and AI ecosystem. Prosus Ventures, which backed Equal AI, has been quietly assembling a portfolio of AI-driven customer engagement tools, including Truecaller (acquired in 2022) and Zoho CRM. Tomales Bay Capital, meanwhile, has bet on Indian AI startups with regulatory moats, such as Razorpay (payments) and now Equal AI (communications).
The timing is deliberate. India’s telecom regulators are tightening screws on unsolicited calls, with TRAI’s new DND 2.0 rules (effective July 2026) requiring providers to block <90% of spam calls by default. Telecom giants like Reliance Jio and Bharti Airtel are scrambling to integrate third-party screening tools—but most legacy systems (e.g., Hiya) lack the granularity to comply. Equal AI’s API, which supports SMS-to-intent parsing and real-time STIR/SHAKEN verification, positions it as the de facto standard for carriers.
The 30-Second Verdict: Why This Isn’t Just Another Call-Blocking App
- Regulatory tailwind: TRAI’s DND 2.0 rules create a forced market for Equal AI’s tech, as carriers must meet compliance by Q3 2026 or face fines.
- API-first design: Unlike competitors, Equal AI’s SDK supports pre-call screening via telecom provider APIs, not just post-call filtering in apps.
- Edge NPU advantage: 60% of processing happens on-device, reducing cloud costs for carriers by ~40% (per internal benchmarks).
- Prosus’ playbook: This follows the same pattern as Truecaller’s acquisition—consolidating fragmented AI tools into a single platform for enterprise adoption.
What Happens Next: The API War and the Death of Static Blacklists
Equal AI’s next move will be critical: API integration with WhatsApp Business. Meta’s platform processes <100 billion messages/month in India alone, but its spam filters rely on user-reported feedback—a slow, reactive system. Equal AI’s real-time screening could force Meta’s hand to either adopt its tech or risk regulatory scrutiny over unfiltered spam. "If they don’t integrate, Equal AI will become the default for WhatsApp Business users in India," predicts Ankit Gupta, a former Meta infrastructure engineer now leading Gupshup, a rival communications API.
The bigger question is whether this sparks an open-source backlash. Truecaller’s database is semi-open, but Equal AI’s model is proprietary. If carriers adopt it en masse, will they demand access to the training data? Or will Equal AI double down on its edge-first architecture, making it harder for competitors to replicate? “The real battle isn’t between Equal AI and Truecaller—it’s between closed models and open frameworks like Kaldi,” says Gupta. “If Equal AI locks in telecom providers, we’ll see a two-tier system: carriers using proprietary AI, and consumers stuck with outdated tools.”
Key Technical Specs (Equal AI vs. Competitors)
| Metric | Equal AI | Truecaller | Hiya |
|---|---|---|---|
| Spam Detection Accuracy | 94% (neural hybrid) | 89% (rule + ML hybrid) | 82% (rule-based) |
| False Positive Rate | <8% | 22% | 35% |
| Latency (Pre-Call Screening) | 150ms (edge NPU) | 450ms (cloud-dependent) | N/A (post-call) |
| Telecom Carrier Adoption | Pilot with Airtel/Jio (2026) | Limited (app-only) | None |
Source: Equal AI internal benchmarks (2026), IEEE Cybersecurity Study (2025), vendor documentation.
The Privacy Paradox: Can AI Screening Work Without Becoming a Surveillance Tool?
Equal AI’s model trains on anonymized metadata, not audio or call content—critical for GDPR-like compliance in India’s Digital Personal Data Protection Act (DPDP). However, the shift to pre-call screening via telecom APIs raises red flags. “If carriers use Equal AI’s API to block calls before they ring, they’re effectively creating a blackbox decision system,” warns Dr. Priya Donti, a cybersecurity researcher at CMU. “There’s no audit trail for why a call was blocked—just an AI’s opaque judgment.”
The company claims its system adheres to India’s DND 2.0 guidelines, which require transparency in blocking decisions. But with no public model cards or bias audits, skeptics argue this is a regulatory loophole. “Equal AI’s advantage is that it’s not just a filter—it’s a gatekeeper,” says Donti. “The question is whether that gatekeeper is accountable.”
Actionable Takeaway: What This Means for Enterprises and Consumers
- For telecom providers: Equal AI’s API is a compliance shortcut—but locking into a single vendor risks vendor lock-in. Negotiate for data portability clauses in contracts.
- For enterprises: If you rely on WhatsApp Business for customer support, monitor Equal AI’s WhatsApp integration. Early adopters may see reduced spam but higher false positives until Meta adjusts its filters.
- For consumers: The app’s edge processing means no cloud dependency, but battery life may dip on older devices. Equal AI’s FAQ confirms it optimizes for Android 12+ and iOS 15+.
- For regulators: TRAI’s DND 2.0 rules create a de facto monopoly for Equal AI if carriers adopt it uniformly. Watch for antitrust scrutiny in 2027.
Equal AI’s $30 million raise isn’t just about filtering calls—it’s about rewriting the rules of India’s telecom ecosystem. The question isn’t whether the AI works (it does), but whether the industry will let it dominate before someone forces an open alternative into the mix.
