Founder of Cipheron Systems. Background, research, and the path between them.
Background
Ayesha Fareed holds an MSc in Food Microbiology from the University of Karachi (UoK) and a Digital Marketing Diploma from IBA Karachi, an unconventional route into AI. Early roles in SEO and digital marketing gave her a practical, results-driven grounding before she moved into building AI automation systems.
That scientific and marketing background gave her an outsider's view of industries like dental, real estate, and construction: sectors with real operational pain but few people building AI for them who understood how they actually worked day to day. Closing that gap became the starting point for Cipheron Systems, founded in February 2026, an AI automation agency based in Karachi, Pakistan, building production voice AI agents and workflow automations for dental, real estate, and construction clients.
That client-facing work led directly into research. While designing Noor and Shifa, two voice AI assistants built for a Masjid al-Haram use case, a lost-pilgrim wayfinding assistant and a medical emergency triage assistant, real deployment problems surfaced: how do you design AI that behaves respectfully in a sacred, high-stakes environment, and how do you make it reliable across the languages pilgrims actually speak? Rather than leave those as open questions, Ayesha followed them into two published papers, introducing the SACRED and later SACREDO frameworks for designing and governing conversational AI in culturally sensitive, high-stakes deployments.
Cipheron Systems remains the applied side of this work, and the research is where the harder questions from that work get investigated properly.
What Cipheron Systems builds
Cipheron Systems is an AI automation agency based in Karachi, Pakistan. It builds and deploys production voice AI agents and workflow automations for dental clinics, real estate and construction clients, using tools including VAPI, ElevenLabs, n8n, Make.com, GoHighLevel, Airtable and Twilio.
From client work to research
The applied work at Cipheron, building real voice agents for real clients, grew into two published research papers. Both came out of two voice AI assistants Ayesha built for a Masjid al-Haram use case: Noor, a lost-pilgrim wayfinding assistant, and Shifa, a medical emergency triage assistant.
Two papers grew out of Noor and Shifa, the two voice AI assistants described above.
Paper 01
Designing Culturally and Spiritually Sensitive AI Assistants for High-Density Religious Environments: A Prompt Engineering Framework Applied at Masjid al-Haram
Culturally sensitive AI design in sacred environments, grounded in the Noor and Shifa deployments.
Masjid al-Haram in Makkah, Saudi Arabia, receives millions of pilgrims annually from over 180 countries, presenting distinctive challenges in emergency response and pilgrim assistance within a setting of profound religious significance. This study applies prompt engineering principles to design, build, and test two conversational AI assistants for this environment: Noor, a lost pilgrim assistant, and Shifa, a medical emergency triage assistant. Both assistants were implemented using persona design, priority ordering, constraint setting, and few-shot logic and were iteratively tested on the VAPI voice AI platform using simulated pilgrim interactions. From this practice-based work, the SACRED Framework is proposed, a six-principle design guide for AI practitioners operating in culturally and religiously sensitive environments, covering Sensitivity, Accessibility, Clarity, Reliability, Escalation, and Debugging. Two case studies report the design decisions, bugs identified and resolved, and testing observations for each assistant. The findings indicate that prompt engineering, applied systematically and iteratively, can produce conversational AI assistants that are culturally respectful, behaviorally reliable, and appropriately scoped for high-stakes deployment in sacred spaces. Concrete limitations, including English-only language support and the absence of formal deployment oversight, are identified and treated as motivation for follow-up empirical work.
Multilingual Voice AI for Pilgrim Assistance: STT Reliability, Bias by Invisibility, and Ethical Deployment in Sacred Environments
Multilingual voice AI for pilgrim assistance: speech-to-text reliability, bias by invisibility, and ethical deployment.
The deployment of voice AI assistants in multilingual, safety-critical environments presents challenges that extend beyond prompt engineering. Building on prior work that proposed the SACRED Framework through two proof-of-concept voice assistants deployed at Masjid al-Haram (Fareed, 2026), this study addresses two previously identified gaps: multilingual speech recognition reliability and deployment oversight. Three commercial speech-to-text providers were evaluated within a production voice orchestration platform. Structured language routing achieved reliable transcription for English, Urdu, and Bahasa Indonesia, while Arabic remained unresolvable across four configurations and eight test calls. A structured field level evaluation across 41 calls in Urdu and Bahasa Indonesia found that identification numbers, the field most directly tied to a pilgrim's identity, were correctly relayed to staff in only 15 percent of Urdu calls and 52 percent of Bahasa Indonesia calls at final confirmation, with many of these errors traced specifically to the assistant's response generation rather than to speech recognition itself. A supplementary check using a second speaker produced results in a similar range, offering limited but direct evidence that this unreliability is not solely an artifact of one individual's accent or pronunciation. Cross-referencing language priorities against Hajj demographic data revealed that Bahasa Indonesia, representing the largest single national pilgrim quota of 221,000 pilgrims, was entirely omitted from the original design. This paper conceptualizes this as bias by invisibility, distinct from technical exclusion, and proposes empirical demographic verification as its remedy. Concrete ethical mechanisms, including a data retention policy, safety-critical escalation paths, and a spiritual distress acknowledgment mechanism, are proposed and tested within Shifa. A three-layer governance framework is introduced. Together these findings motivate four revisions to the SACRED Framework and the introduction of Oversight as a seventh principle, extending the framework to SACREDO.