Now in Early Access · Bitcoin-Native · 16+ Yrs Telecom DNA

Every Call Becomes Your Smart Notes

CallJots is the AI-powered VoIP platform born from over sixteen years of carrier-grade telecommunications experience — from SS7 signalling and SIP/RTP infrastructure to real-time AI transcription. Every call is automatically converted into searchable, timestamped smart notes — paid for with Bitcoin and settled on the Stellar XLM network.

Bitcoin Payments
XLM Settlements
AI Transcription
🔒
E2E Encrypted
📞
SIP/RTP VoIP
🏢
Joomo Enterprises
CALLJOTS · MON 16th · WEEK VIEW
2:08 PM AI NOTED
🔊 Team Meeting — Project Kickoff
"…sprint deadline moved to the 22nd. Sarah to deliver mockups by Thursday EOD…"
📋 AI NOTES — BART SUMMARY
• Sprint deadline: 22nd
• Sarah: mockups → Thu EOD
• Budget approval pending CFO
• Next call: Wed 10 AM
₿ BTC Pay
⬍ XLM Settled
🤖 AI Notes
{{ s.icon }}
{{ s.value | number:0 }}{{ s.suffix }}
{{ s.label }}
Our Heritage

16 Years of Telecom DNA

From SS7 signalling networks to AI-powered VoIP — every protocol we mastered, every carrier we integrated, every byte of CDR data we processed shaped the platform you're using today. This is the Joomo Enterprises story.

2008

SMSDAM — The Bulk SMS Pioneer

Joomo Enterprises entered the Telecom arena with SMSDAM, a carrier-grade Bulk Messaging Platform engineered from the ground up on the SMPP v3.4 protocol. At a time when enterprise messaging was dominated by expensive, closed-vendor stacks, SMSDAM offered operators and enterprises a robust, open-protocol alternative capable of handling millions of messages daily across MT (Mobile-Terminated) and MO (Mobile-Originated) message flows.

The platform integrated directly with operator SMSCs (Short Message Service Centres) over dedicated SMPP sessions, maintaining persistent TCP connections with full bind_transmitter / bind_receiver / bind_transceiver negotiation. Delivery confirmation relied on end-to-end DLR (Delivery Receipt) tracking routed back through the originating SMSC, giving enterprise clients real-time visibility into message delivery states — from ENROUTE and DELIVRD through to UNDELIV error codes.

Routing intelligence was built on live HLR (Home Location Register) queries over SS7/MAP signalling, enabling SMSDAM to perform number portability lookups, validate MSISDN ranges, and route traffic across interconnected operators without revenue leakage. The billing engine natively consumed operator CDR (Call/Message Detail Record) feeds and supported pre-paid credit controls — a capability that would later evolve into the full IN billing subsystem.

SMPP v3.4 MT/MO SMS SS7/MAP HLR Lookup DLR Tracking SMSC Integration
2010 – 2013

Convergent Billing & IN Platform

Growing operator demand pushed Joomo to expand beyond messaging into Convergent Billing. We built an Intelligent Network (IN) mediation layer compliant with CAMEL Phase 2 & 3 (CAP), connecting to operator SCP (Service Control Points) via INAP/CAP over SS7 to intercept prepaid call events, apply real-time charging, and enforce credit limits without breaking the bearer path.

The CDR pipeline was redesigned for carrier-scale throughput: raw ASN.1-encoded CDRs from MSCs and SMSCs were ingested, decoded, correlated across originating and terminating legs, and fed into a mediation engine that produced normalised records for downstream billing systems. We integrated with RADIUS for broadband session accounting and Diameter (RFC 6733) for EPC/LTE Online Charging — giving our platform a foot in both legacy 2G/3G and emerging 4G/LTE domains.

Inter-operator settlement used the TAP3 (Transferred Account Procedure v3) roaming data exchange format mandated by the GSMA, enabling automated reconciliation of roaming charges between partner networks — experience that would later inform our cross-border crypto settlement design.

CAMEL/CAP IN / SCP CDR Mediation RADIUS Diameter TAP3/GSMA ASN.1
2014 – 2017

VoIP Infrastructure & Carrier Interconnect

The shift from circuit-switched PSTN to packet-switched VoIP (Voice over IP) was the defining technology transition of this era, and Joomo was in it from day one. We built out a carrier-grade VoIP core using Asterisk PBX and FreeSWITCH as session border functions, handling call signalling via SIP (RFC 3261) with full SDP (Session Description Protocol) offer/answer negotiation for codec selection — supporting G.711 (PCMU/PCMA), G.729, G.722 wideband, and Opus for high-definition voice.

Media transport ran over RTP (RFC 3550) with SRTP (RFC 3711) for encrypted media streams, using DTLS-SRTP (RFC 5763) for key exchange — the same mechanism now powering WebRTC endpoints in CallJots. NAT traversal was solved through a full STUN / TURN / ICE stack, allowing clients behind enterprise firewalls and carrier-grade NATs (CGN) to establish direct media paths. Call quality was monitored using RTCP metrics — jitter, packet loss, and MOS (Mean Opinion Score) scoring — feeding QoS dashboards used by interconnected carriers.

Carrier interconnects were provisioned over SIP trunks with Tier-1 operators, with call routing governed by LCR (Least Cost Routing) logic that factored in per-destination termination rates, codec compatibility, and real-time quality scores from RTCP feedback.

SIP/SDP RTP/SRTP DTLS-SRTP STUN/TURN/ICE Asterisk FreeSWITCH MOS/QoS LCR
2018 – 2021

AI Research & ASR Pipeline Development

With a decade of call data and deep VoIP experience, the next logical step was intelligence. Joomo's R&D arm began investing in Automatic Speech Recognition (ASR) research, benchmarking acoustic models from Kaldi and wav2vec 2.0 through to OpenAI's Whisper and Google's Conformer architecture — selecting the optimal combination for telephony-quality audio (8 kHz narrowband to 48 kHz wideband).

The NLP layer explored summarisation and information extraction using BART (Bidirectional and Auto-Regressive Transformers), T5 (Text-to-Text Transfer Transformer), and classification with BERT / RoBERTa fine-tuned on business conversation datasets. Speaker diarisation — separating who said what — was tackled with GMM-HMM speaker models and later upgraded to neural speaker embeddings.

Operationally, this period saw the adoption of Docker and Kubernetes for containerised microservice deployment, enabling the AI inference pipeline to scale independently of the core VoIP signalling — a cloud-native architecture essential for the latencies demanded by real-time transcription.

Whisper ASR Conformer BART/T5 BERT/RoBERTa wav2vec 2.0 Docker/K8s Speaker Diarisation
2022 – 2024

Crypto-Native Payments & Global Settlement

Traditional telecom billing had always been constrained by correspondent banking relationships, inter-operator settlement delays, and high FX costs on cross-border traffic. We set out to dismantle those barriers entirely. After deep research into Layer-2 Bitcoin infrastructure, we adopted the Lightning Network as the primary payment rail — enabling sub-second, micropayment-capable, trust-minimised settlement for per-minute call billing with transaction fees measured in satoshis.

For network-layer settlement between platform nodes and partner operators, we chose the Stellar (XLM) network: a purpose-built financial infrastructure with 3–5 second finality, native multi-asset support, and a decentralised exchange (SDEX) for on-chain currency conversion. Integration with the Stellar Anchor framework enables fiat on/off-ramp corridors in 180+ countries — a settlement model that mirrors the TAP3 roaming reconciliation we built a decade earlier, but without the 30-day clearing cycle.

Bitcoin Lightning Network Stellar XLM SDEX Micropayments Cross-Border Settlement
2025  ✨

CallJots — Everything Converges

CallJots is the culmination of 16 years of Telecom and IT engineering at Joomo Enterprises. It is not a startup's first product — it is a seasoned Telecom operator's answer to a universal problem: enterprises lose institutional knowledge every time a phone call ends without a reliable record.

We fused our carrier-grade VoIP core (SIP/RTP/SRTP, STUN/TURN/ICE, WebRTC) with a battle-tested AI transcription pipeline (Whisper + Conformer ASR, BART/T5 summarisation, GMM speaker diarisation) and wrapped it in a crypto-native billing layer (Bitcoin/Lightning + Stellar XLM). The result: every call becomes a timestamped, searchable, AI-summarised knowledge asset — paid for in seconds, settled globally, owned entirely by you.

WebRTC Real-Time ASR AI Summaries Bitcoin Pay Stellar Settlement Zero Data-Loss
🏢
Joomo Enterprises — the ICT venture behind CallJots. Building carrier-grade communications since 2008. From SMSDAM bulk messaging to CallJots AI-VoIP, we have always been at the intersection of Telecom protocol depth and real-world operator experience.
How It Works

From Call to Notes in Seconds

Three effortless steps transform every conversation into a timestamped, searchable knowledge base.

1
📞
Make or Receive a VoIP Call
Use CallJots as your primary VoIP number powered by a carrier-grade SIP/RTP core with SRTP + DTLS-SRTP encryption on every media stream. Whether you're dialling out over SIP trunk, receiving inbound via DID, or running a WebRTC browser session, the call is routed through our ICE/STUN/TURN NAT traversal stack for crystal-clear audio even behind enterprise firewalls. Opus wideband codec ensures HD voice quality; RTCP feedback monitors jitter and packet loss in real time, automatically switching codec if the network degrades.
2
📞
AI Transcribes & Summarises
The moment audio begins flowing over RTP, our Whisper + Conformer ASR engine begins transcribing — handling accents, telephony-quality audio (8 kHz narrowband to 48 kHz wideband), and mixed-language utterances. BART and T5 transformers run abstractive summarisation in parallel, distilling hour-long calls into crisp action items and decisions. GMM speaker diarisation tags every transcript segment with the speaker identity, so the final notes are always attributed — no more guessing who made which commitment. The entire pipeline runs in under 2× real-time latency.
3
📞
Browse & Search Your Notes
Every call is indexed by date, time, duration, participants, and AI-extracted keywords — stored in a structured CDR-style record built on our decade of call data engineering. Jump to the exact 2:08 PM meeting from last Monday in under a second. Full-text search across all transcripts. Export notes as PDF or share a read-only link. Set retention policies. Integrate via REST API. Your call history becomes a searchable institutional knowledge base — the answer to "what did we agree on?" is always one click away.
Technology Stack

Carrier-Grade Protocols.
Enterprise-Grade Reliability.

CallJots is built on the same protocol layers that run the world's telephone networks — from SS7 signalling in the core to WebRTC at the browser edge. No shortcuts. No proprietary lock-in. Just battle-proven Telecom standards, re-engineered for the cloud.

Application Layer
WebRTC SIP/SDP REST API HTTPS/TLS 1.3 OAuth 2.0 JWT

All user-facing interactions — browser calls, mobile clients, third-party API integrations — operate over WebRTC for media and SIP for call control, secured end-to-end with TLS 1.3. Our REST API uses OAuth 2.0 bearer tokens and signed JWTs, ensuring every API call is both authenticated and auditable.

Media & Signalling Layer
RTP/SRTP DTLS-SRTP RTCP/SRTCP ICE/STUN/TURN Opus G.711/G.722/G.729

Voice media travels over SRTP (RFC 3711) with DTLS-SRTP key exchange, making every media stream encrypted by default — no opt-in required. ICE/STUN/TURN handles NAT traversal for clients behind enterprise firewalls and carrier-grade NAT. RTCP feedback loops provide real-time jitter, loss, and MOS quality metrics used by our QoS engine to dynamically select the best codec — Opus for wideband, G.711 for legacy interop.

Telecom Core Layer
SS7/MAP SMPP v3.4 CAMEL/CAP Diameter SIP Trunking HLR/HSS

Deep carrier integration is our heritage. SS7/MAP signalling for HLR queries and number portability, SMPP v3.4 for legacy SMSC interconnects, CAMEL/CAP for IN service triggers, and Diameter for LTE/EPC online charging — all the protocols that keep the world's mobile networks running are part of our operational DNA at Joomo Enterprises. CallJots inherits this carrier-grade reliability in every call.

AI & Intelligence Layer
Whisper ASR Conformer wav2vec 2.0 BART T5 BERT/RoBERTa GMM-HMM

Our AI pipeline is purpose-built for telephony audio. Whisper and Conformer handle acoustic modelling across 8–48 kHz sample rates. BART and T5 produce abstractive summaries of call content, extracting decisions and action items. BERT/RoBERTa powers intent classification and keyword spotting. GMM-based speaker diarisation segments transcripts by speaker identity — so every note knows exactly who said what.

Payments & Settlement Layer
Bitcoin Lightning Network Stellar XLM SDEX Horizon API

Telecom billing has always been complex. We simplified it. Bitcoin/Lightning enables sub-second, satoshi-denominated micropayments per call minute — no credit card, no bank required. Stellar XLM handles inter-node settlement with 3–5 second finality across 180+ countries, using the SDEX for on-chain FX conversion. It's the TAP3 roaming settlement model — rebuilt for the blockchain era.

Infrastructure Layer
Docker Kubernetes MySQL 8 Redis Apache/PHP 8 CDR Pipeline

The platform runs on a containerised microservice architecture deployed via Kubernetes, with independent scaling for the VoIP core, AI inference pipeline, and billing engine. MySQL 8 stores call metadata and CDR records with ACID guarantees. Redis caches session state and ASR result queues. The CDR mediation pipeline — a direct descendant of our 2010-era billing work — provides the audit trail that enterprise compliance teams require.

The AI Engine

State-of-the-Art Algorithms
Powering Your Notes

Selected from our comprehensive algorithm research, these are the most productive models for Voice-to-Transcript-to-Notes pipelines.

All Algorithms
Speech Recognition
NLP & Summarisation
Sequence Models
Probabilistic
🔊 Whisper (OpenAI)
Speech-to-Text
CallJots Use: Primary transcription engine. Converts VoIP audio to raw text transcript with 98%+ accuracy across 99 languages.
Ref: Radford et al. "Robust Speech Recognition via Large-Scale Weak Supervision", OpenAI (2022)
🤖 Conformer
ASR Backbone
CallJots Use: Real-time audio processing backbone. Combines convolution + attention for superior speech recognition on streaming audio.
Ref: Gulati et al. "Conformer: Convolution-augmented Transformer for Speech Recognition", Interspeech (2020)
🤖 wav2vec 2.0
Self-Supervised ASR
CallJots Use: Fine-tuned for noisy VoIP conditions. Self-supervised pre-training enables robust speech features from raw audio waveforms.
Ref: Baevski et al. "wav2vec 2.0: A Framework for Self-Supervised Learning of Speech Representations", NeurIPS (2020)
📋 BART
Summarisation
CallJots Use: Core notes generation. BART's denoising seq2seq architecture extracts action items and key decisions from transcripts.
Ref: Lewis et al. "BART: Denoising Sequence-to-Sequence Pre-training", ACL (2020)
📝 T5 (Text-to-Text)
Structured Notes
CallJots Use: Converts transcripts to structured bullet notes. T5's unified text-to-text paradigm handles summarisation, Q&A extraction simultaneously.
Ref: Raffel et al. "Exploring the Limits of Transfer Learning with T5", JMLR (2020)
📜 BERT / RoBERTa
NLU
CallJots Use: Intent & entity extraction from notes. Identifies names, dates, tasks, and deadlines mentioned in calls with context-aware accuracy.
Ref: Devlin et al. "BERT: Pre-training of Deep Bidirectional Transformers", NAACL (2019)
📈 BiLSTM + CRF
Sequence Labeling
CallJots Use: Named entity recognition in transcripts. Identifies speakers, organisations, products, and dates for structured note generation.
Ref: Huang et al. "Bidirectional LSTM-CRF Models for Sequence Tagging", arXiv (2015)
💡 Attention Mechanism
Neural Attention
CallJots Use: Focuses on key moments in long calls. Bahdanau attention ensures the summariser attends to critical decision points in meetings.
Ref: Bahdanau et al. "Neural Machine Translation by Jointly Learning to Align and Translate", ICLR (2015)
🔊 Transformer
Core Architecture
CallJots Use: Underpins all AI models in the pipeline. Self-attention enables parallel processing of entire call transcripts for lightning-fast notes.
Ref: Vaswani et al. "Attention Is All You Need", NeurIPS (2017)
🔇 GMM Speaker ID
Diarisation
CallJots Use: Speaker identification & diarisation. GMMs model unique vocal fingerprints so notes are attributed correctly to each speaker.
Ref: Dempster et al. "Maximum Likelihood from Incomplete Data via the EM Algorithm", JRSS (1977)
📊 HMM (Hidden Markov)
Temporal Modelling
CallJots Use: Acoustic modelling for phoneme sequences. HMMs encode temporal dynamics of speech, improving accuracy on accented or fast speech.
Ref: Rabiner "A Tutorial on Hidden Markov Models and Selected Applications in Speech Recognition", IEEE (1989)
⏰ CTC (Connectionist Temporal)
End-to-End ASR
CallJots Use: End-to-end speech recognition without forced alignment. CTC enables streaming transcription with minimal latency for live call notes.
Ref: Graves et al. "Connectionist Temporal Classification: Labelling Unsegmented Sequence Data with RNNs", ICML (2006)
Features

Everything You Need,
Nothing You Can't Control

📅
Time-Based Call Browser
Logged in Monday morning. Clicked the tab for one week ago. Scrolled to the 2:08 PM team meeting call. The notes were all I needed. That's CallJots — every call, perfectly archived, instantly found.
Week View Day Timeline Search Notes Export PDF
🔊
Real-Time Transcription
Whisper + Conformer ASR models trained on telephony audio transcribe while you talk — zero post-call processing wait. Notes appear as words are spoken, handling accents and mixed languages with accuracy benchmarked against wav2vec 2.0 baselines on real call-centre audio. Supports 8 kHz PSTN-quality to 48 kHz WebRTC wideband streams.
Whisper ASR Conformer Real-Time Multilingual
🔇
AI Smart Summaries
BART and T5 transformer models perform abstractive summarisation — not just keyword extraction, but genuine comprehension of call context. Decisions, tasks, deadlines, and commitments surface automatically as structured bullet points. BERT/RoBERTa classifies intent and flags action items for follow-up, turning every call into a self-organising to-do list.
BART/T5 BERT/RoBERTa Action Items Abstractive NLP
📅
Speaker Diarisation
GMM-HMM speaker models segment and label every transcript line by speaker identity — the same statistical modelling approach used in carrier-grade call analytics. Each note segment carries a speaker tag, so there's no ambiguity about who said what. Multi-party conference calls with up to 8 participants are correctly attributed, giving you an auditable record of every commitment made on every call.
GMM-HMM Multi-Speaker Attribution Audit Trail
📅
End-to-End Encryption
Security is not a feature — it's a protocol requirement. Every media stream is encrypted via SRTP + DTLS-SRTP (RFC 3711 / RFC 5763). Call notes and transcripts are encrypted at rest using AES-256. All API traffic travels over TLS 1.3. Key exchange uses ephemeral ECDH, ensuring forward secrecy — a compromise today cannot decrypt yesterday's calls. Your conversations stay yours, always.
SRTP/DTLS AES-256 TLS 1.3 Forward Secrecy
📅
Global VoIP + Crypto Pay
Call anywhere in the world via SIP trunks with Tier-1 carrier interconnects, using LCR (Least Cost Routing) for optimal per-destination pricing. Pay per-minute with Bitcoin/Lightning — satoshi-denominated micropayments with sub-second settlement, no credit card or bank account required. Inter-platform settlement runs on Stellar XLM with 3–5 second finality across 180+ countries and near-zero transaction fees. The future of Telecom billing, today.
LCR Routing Bitcoin/Lightning Stellar XLM 180+ Countries
Real User Story

Monday Morning. Five Seconds. Done.

"I jumped into the office Monday morning and began working on the new project. Logged into my CallJots and clicked on the tab for one week ago on the 16th. Then scrolled to the 2:08 PM team meeting call. The notes were all I needed…"

📅 Week of 16th — located instantly
🕐 2:08 PM — exact timestamp
🤖 AI notes — ready, no replay needed
<5 seconds — from login to notes
Crypto-Native Payments

Built on Bitcoin.
Settled on Stellar.

CallJots is the first VoIP platform built natively on Bitcoin for user payments and the Stellar XLM network for instant, borderless platform settlement. We didn't add crypto as an afterthought — we redesigned the billing stack from scratch using the same engineering rigour we applied to our CAMEL/CAP IN billing and TAP3 roaming settlement systems at Joomo Enterprises. The result: per-minute, per-call, or subscription billing settled in seconds — not 30-day clearing cycles. No banks. No intermediaries. No borders. Payments as global as your calls.

Bitcoin
Primary payment rail & store of value
Stellar XLM
Network remittances & instant settlement
<5s
XLM settlement finality
₿ sats
Per-minute Lightning billing
180+
Countries, no FX friction
24/7
No banking hours. Ever.
$0.00001
Avg Stellar transaction fee
16+
Years of billing engineering
Pricing

Simple, Transparent Plans

Pay with Bitcoin. Settle in XLM. Cancel anytime.

Starter
$0
Free forever
30 mins/month transcription
7-day call history
Basic AI notes
Speaker diarisation
Export to PDF/CSV
BTC payment integration
Start Free
Team
$49/mo
Up to 10 users
Everything in Pro
Unlimited call history
Shared team workspace
Admin dashboard
API access
Priority support
Get Team Plan
Early Access

Join the CallJots Waitlist

Be among the first to turn every call into perfect notes. Sign up now and get 3 months of Pro for free at launch.