AI face-matching on a portable edge box. WhatsApp delivery during the event. Works with no Wi-Fi. Built for Tier-2 and Tier-3 India.
The problem
You go home from a 500-person event. Someone photographed you. You will never see that photo.
Every function in India generates thousands of photos. A portion land on one person's phone. The rest vanish. The tools that exist sell to professional photographers and require stable internet. They solve the wrong problem for the wrong customer in the wrong place.
The solution
"At your event, everyone's phone photos get pooled, sorted by face on a box we bring, and every guest gets their own photos on WhatsApp before they reach home, even with no Wi-Fi."
The name is Malayalam for elephant. Elephants remember everything. That is what aana does for every gathering, large or small, with or without a professional photographer.
Guests upload phone photos to a local Wi-Fi hotspot. The operator's edge box ingests them with no internet needed.
ArcFace embeddings run locally on the edge appliance. Every face is clustered and tagged in real time, entirely offline.
Guests register by selfie or QR. Their album arrives on WhatsApp before the event ends. Consent-first, DPDP-clean.
Market landscape
The photographer-SaaS market is a red ocean. The white space is the no-photographer small gathering, the bad-connectivity venue, and the hyperlocal operator relationship.
| Player | Positioning | Gap we exploit |
|---|---|---|
| Premagic | Event-marketing platform; face delivery at large events (GITEX, 2000+) | Enterprise only no small hostless gatherings |
| Kwikpic | Photographer SaaS; AI face recognition; VC-backed; ₹849–₹29,990/yr | Needs internet post-event sharing, not live |
| Memzo / Foto Owl / Photomall | Selfie/QR face delivery for weddings | Photographer-led no-photographer events ignored |
| Samaro | WhatsApp bot; targets Tier-2/3 and real-time explicitly | Direct rival must out-execute on edge + local trust |
| aana | Edge-first, offline-capable, operator franchise network | Our wedge hostless events + no Wi-Fi + local trust |
The beachhead is not weddings (incumbents own that) but community mid-size events: temple festivals, school annual days, college fests, association functions, milestone birthdays. High guest count, no professional photographer, strong demand, and no one serving it well.
Business model
Local operators buy or lease the edge kit, pay a monthly platform fee, and keep the host fee in their territory. The company's marginal cost per event collapses to WhatsApp delivery and cloud storage.
Pays ₹2,500/month platform fee. Buys or leases the edge kit (margin goes to aana). Earns the host fee in their territory. Becomes a local business owner, not a gig worker.
In the franchise model our marginal cost per event is Rs.75 to Rs.860 (cloud + WhatsApp + gateway). Everything else is the operator's cost structure, not ours.
The person hosting the birthday, festival, or function pays the operator a flat fee. Clean single transaction, aligned incentive.
HD downloads are free. Photobook (Rs.499 to Rs.1,499), private vault (Rs.99 to Rs.199), AI highlight reel. Company captures 30% of upsell revenue. Upside, not a pillar.
| Event size | Host price | Notes |
|---|---|---|
| Small (up to 75 guests) | Rs.999 to Rs.1,999 | Birthdays, anniversaries, small functions |
| Mid (75 to 250 guests) | Rs.2,499 to Rs.4,999 | Community, school, college, association events |
| Large (250 to 1000+ guests) | Rs.5,000 to Rs.15,000 | Weddings; competes on real-time + reliability |
| Operator platform fee | Rs.2,500/month | Unlimited events; covers cloud + delivery + platform |
Unit economics
| Cost line | 50 guests | 250 guests | 1,000 guests |
|---|---|---|---|
| Edge appliance amort. | Rs.90 | Rs.120 | cloud-burst |
| Venue connectivity (5G) | Rs.75 | Rs.100 | Rs.150 |
| Cloud-burst GPU | - | - | Rs.600 |
| WhatsApp delivery | Rs.25 | Rs.125 | Rs.500 |
| Cloud gallery (90 days) | Rs.20 | Rs.40 | Rs.120 |
| Payment gateway (2%) | Rs.30 | Rs.80 | Rs.240 |
| Operator time on-site | Rs.450 | Rs.750 | Rs.2,000 |
| Total cost (company-run) | Rs.690 | Rs.1,215 | Rs.3,610 |
| Indicative host price | Rs.1,499 | Rs.3,999 | Rs.12,000 |
| Contribution (company-run) | Rs.809 | Rs.2,784 | Rs.8,390 |
| Company marginal cost (franchise) | ~Rs.75 | ~Rs.245 | ~Rs.860 |
In the franchise model the operator absorbs kit, time, and connectivity. Company marginal cost collapses to WhatsApp + cloud + gateway. Platform fee and kit margin are near-pure contribution at scale.
Edge appliance
Real-time face detection (RetinaFace/SCRFD) and embedding (ArcFace via InsightFace, ONNXRuntime-GPU) benefits materially from a GPU. Two validated configurations cover the range.
Subsequent operator kits are not company capex; operators buy or lease them, and aana earns ~Rs.16k to Rs.20k margin per kit.
5-year financial model
The model is most sensitive to operator count and upsell take-rate. Pull the sliders to stress-test the base case.
| Line (Rs. lakh) | Year 1 | Year 2 | Year 3 | Year 4 | Year 5 |
|---|
Figures in Rs. lakh (1L = Rs.100,000). Operator ramp: 6% / 21% / 57% / 77% / 100% of Year 5 target at year-end. Event ramp factor applied as operators mature. Team cost scales with operator count and year. All figures illustrative; validate against live vendor quotes and market conditions.
Go-to-market
The crowdsourcing intent is preserved as a franchise network. Local people, local hustle, asset-light scaling — biometric matching stays on company-controlled infrastructure.
Use existing Immich setup (already does face recognition and shared albums). Take the home server and a 5G hotspot to 3 to 5 real events. Measure two numbers: guest download rate and host willingness to pay Rs.1,499. Zero build cost until those numbers are proven.
Build consent flow, WhatsApp delivery, operator app, host dashboard. Recruit first operators from people who saw the product work at a real event. Standardise the kit and playbook. Prove operator economics.
Channel partnerships with event decorators, caterers, and tent/sound houses who already sell to every function in town. Revenue share for bundling. Viral loop: real-time delivery creates the WhatsApp-share moment; every shared photo drives the next host enquiry at near-zero CAC.
Cloud-burst GPU for large events. Regional micro-PoPs as volume justifies. Each operator becomes a node in a distributed edge compute network — the foundation for the long-term infrastructure play described in the next section.
Network growth
Each operator proves the model locally, then refers adjacent towns. Local trust compounds faster than any ad spend.
Long-term vision
The event photo product is the wedge. The real long-term value is the operator network itself: bonded local operators in every Tier-2 and Tier-3 town, each running a standardised edge appliance, each earning a living from the network.
The core product. Operators run edge appliances at gatherings. aana earns platform fees, kit margin, and upsell share. Operators build a recurring local business. Every event is a proof point for the network's reliability and reach.
As operator count grows, the appliances in the field represent meaningful distributed compute capacity. Each node is already capable of local inference, storage, and delivery. The platform layer evolves to expose that capacity to adjacent use cases: local AI inference, edge CDN, IoT data aggregation.
A network of 300+ operator-run edge nodes across India's smaller cities and towns becomes a different kind of asset: distributed storage and compute built into the fabric of communities that hyperscalers do not reach. The operator earns from events today and from node revenue tomorrow. aana's moat is not the software feature; it is the trusted human network and the hardware footprint that took years to build.
DPDP compliance
DPDP Rules 2025 notified November 2025, phased rollout through May 2027. Facial data requires specific, informed, free, and unambiguous consent at collection. Penalties up to Rs.250 crore. Incumbents are vague on this — aana makes compliance a visible differentiator.
QR/selfie pre-registration is the consent moment. Plain-language notice, distinct parental-consent flow for guests under 18 (relevant at birthday parties).
Faces processed only for this event's delivery. No reuse across events. No cross-event profiling. Face templates deleted when the gallery expires, default 90 days.
Auto-purge on gallery expiry. Deletion-on-request built from day one. Auditable consent and deletion logs.
Face matching runs on company-controlled infrastructure only. Biometric data never distributed to uncontrolled operator nodes. Franchisees bound by Data Processing Agreements.
Images and templates encrypted in transit and at rest. Role-based access controls. Security review budgeted before Phase 2 launch.
As volume grows, Significant Data Fiduciary obligations may apply. Compliance review scheduled ahead of Phase 2 expansion.
Key risks
| Risk | Severity | Why it matters | Mitigation |
|---|---|---|---|
| Commoditised core feature | Medium | Incumbents already do face-matched delivery | Compete on edge/offline + hostless segment + local trust, not the feature itself |
| Low ARPU at small gatherings | Medium | Rs.999 to Rs.1,999 per event is thin at single-unit level | Franchise model shifts cost to operators; upsells and platform fee are higher-margin lines |
| DPDP liability | High | Biometric data, children present, Rs.250Cr penalty ceiling | Consent-at-capture, purpose limitation, auto-deletion, DPAs, controlled matching stack |
| Connectivity at venues | Medium | WhatsApp delivery requires uplink at some point | Edge box processes locally offline; delivery queues and fires when connectivity is available |
| Operator quality and churn | High | Network reputation depends entirely on operator execution | Standard kit + playbook, operator certification, delivery metrics monitored centrally |
| Samaro out-executes on Tier-2/3 | Medium | Direct competitor already targeting the same wedge | Edge offline capability and bonded operator network are harder to replicate than a WhatsApp bot |
90-day validation plan
You do not need to build anything new to test demand. The existing Immich setup already does face recognition and shared albums.
Take the current home server and a 5G hotspot to real gatherings. A family function in Aryanad is an ideal first one. Zero build cost.
Collect phone photos from guests, run Immich face grouping, hand each guest a retrieval link. Measure how many actually open and download.
Guest download rate (engagement signal) and whether the host would have paid Rs.1,499 (willingness-to-pay signal). Both must be clear yes before building.
Only if hosts say yes: consent flow first, then WhatsApp delivery, then operator playbook. Not the other way around.
aana — ആന
A network of local operators. An edge appliance at every gathering. Every guest gets their photos. Every town becomes a node. This is the long game.
Get in touch