Archive position — measured, not model output
0 likes on Devpost
2,264 of the 7,856 archived projects have more likes, and 5,592 share exactly 0 — so this project's #4,621 place in the like-ranked listing is a tie-break inside that group, not a ranking.
Projects (log scale)
Likes on Devpost. ▲ marks this project's group.
Show the figures
| Likes | Projects | Share of archive |
|---|---|---|
| 0 | 5,592 | 71.2% |
| 1 | 1,758 | 22.4% |
| 2 | 285 | 3.6% |
| 3–4 | 132 | 1.7% |
| 5–9 | 75 | 1.0% |
| 10+ | 14 | 0.2% |
Executive Summary
What the company appears to be
In Touch is an ambient app that tracks contact rhythm across relationships and watches for sustained change in those patterns — detecting when a friendship is quietly fading or forming — before it becomes obvious. It surfaces one honest, specific sentence when a shift holds long enough to be real rather than noise.
What changed
The project evolved from a quiet, journal-like demo into an installable Android app with on-device data processing and privacy controls. The author built the entire system alone using GPT-5.6 as an engineering partner, with a strong emphasis on deterministic logic and user privacy.
Single most important open question
Is there evidence of real-world usage or adoption beyond the demo? The description states that it ships with six illustrative synthetic relationships but does not indicate whether users have imported real data or engaged with the app beyond initial testing.
What The Product Actually Is
The description states that In Touch is an ambient app designed to notice subtle shifts in friendship patterns. It tracks contact rhythm across relationships and watches for sustained change against each person's own history, never a universal average.
It imports real WhatsApp chat exports (entirely on-device) and scans Android calendars for meetup frequency using only on-demand permission requests — no message content or event details are stored or transmitted.
The app combines these two data sources to form richer patterns when histories overlap. For new contacts, it uses cold-start detection that compares against a reference built from the user's own established relationships.
It surfaces exactly one honest, specific sentence when a shift holds long enough to be real rather than noise — nothing before or after that moment.
The author built this with Codex running on GPT-5.6 as an engineering partner, but every decision was theirs: detection math, cold-start logic, privacy boundaries, visual system, and what the app would say to a user.
Positioning & Claim Evolution
The description states that most "stay in touch" apps are retrospective — they wait until a birthday has passed or a contact has gone quiet for months, then nudge with guilt-shaped notifications. In contrast, In Touch aims to notice earlier, while a friendship is still in the process of fading, not after.
The idea evolved from simply noticing endings quietly into also noticing beginnings quietly — symmetry between fading and forming relationships.
A key claim made is that no scores, rankings, or advice are included. The moment an app tells you what to do about your friendships, it stops being a quiet observer and becomes one more notification demanding something of you.
The author notes that the project started as something deliberately quiet, like a paper-and-ink field notebook feel, but was rethought toward a full-screen, installable experience with on-device persistence and offline operation.
Target Customer & ICP
The description does not state who the target customer is or define an ideal customer profile (ICP). It implies that the app is for individuals who want to maintain awareness of their personal relationships without being nagged by systems.
It focuses on people who value privacy, subtlety, and early detection of relationship changes — those who might be interested in tracking contact rhythm but not necessarily in a system that judges or advises them.
There is no mention of demographic data, usage scenarios beyond personal use, or any segmentation strategy.
Business Model & Pricing Evidence
The description does not provide evidence of a business model or pricing structure. It describes the app as an ambient tool for noticing shifts in relationships and emphasizes its privacy features and lack of scoring or advice.
No revenue streams, monetization strategies, subscriptions, or paid tiers are mentioned.
Technical & Delivery Signals
The author built the entire system alone using Codex running on GPT-5.6 as an engineering partner. Every decision was made by the author: detection math, cold-start logic, privacy boundaries, visual system, and what the app would say to a user.
The architecture is split into two halves:
- Deterministic core — four-month personal baseline per contact, normalized independently and combined with fixed, explainable weights. A flag only fires once a shift crosses threshold and holds for two consecutive months.
- Narrow LLM layer — only once something has genuinely fired does a model get involved, turning structured results into one well-phrased sentence, guarded against advice or scoring language.
WhatsApp parsing discards message content immediately after checking for it. Calendar scanning reads only what's needed to match an attendee, then keeps nothing but a monthly count. Neither the deployed web app nor the Android app calls an external API for anything a user does.
The app ships with six illustrative synthetic relationships and supports real WhatsApp chat exports and Android calendar scans — all on-device.
Traction & Maturity Signals
The description states that In Touch ships with six illustrative synthetic relationships to demonstrate the full range of what it notices. It also mentions support for real WhatsApp chat exports and Android calendar scans, both entirely on-device.
However, there is no evidence of actual user engagement or adoption beyond the demo. No mention of active users, retention metrics, or feedback from real-world usage.
The project was submitted to the OpenAI 2026 hackathon, indicating it's a prototype or proof-of-concept rather than a mature product.
Competitive Context
The description does not provide any information about competitors or competitive positioning. It only contrasts In Touch with typical "stay in touch" apps that are described as retrospective and guilt-inducing.
No mention of existing products in the space, market size, or differentiation strategy is provided.
Key Risks & Red Flags
- Lack of real-world usage: The app ships with synthetic data and supports real imports, but there's no evidence of actual users or adoption.
- Single-founder build: The entire project was built by one person (Riyan Sarkar), which raises questions about scalability and long-term maintenance.
- Platform constraints: Android's restrictions on call-log access forced a design change — this could be a recurring issue if more features are added.
- No monetization strategy: There is no indication of how the app will generate revenue or sustain itself beyond its current demo state.
Diligence Questions To Ask The Founders
- Have users imported real WhatsApp or calendar data, and if so, how many?
- What is the actual user engagement like with the app beyond initial testing?
- How does the cold-start detection logic scale to a larger number of contacts?
- Are there plans for live syncing of WhatsApp and calendar data, and what are the privacy implications?
- What are the technical challenges in expanding support to other platforms or messaging apps (e.g., Discord, Telegram)?
- Is there any plan to move beyond direct APK sideloading to Play Store distribution?
Investment/Partnership Verdict
Not evidenced.
The description does not provide sufficient evidence to assess whether In Touch is ready for investment or partnership. It remains a prototype built by one individual with no demonstrated traction, revenue, or customer base. The app shows strong attention to privacy and design but lacks commercial signals that would indicate viability as a business or product.
Any potential value lies in its conceptual innovation and execution of privacy-first principles — but without real-world usage or scalability evidence, it cannot be evaluated as a viable investment or partnership opportunity at this stage.
Source
Submitted to the OpenAI 2026 hackathon on Devpost. Project home on DevPost.
The analysis above was generated by a language model from the project's own one-line description. It is not independent research and contains no verified traction, revenue or customer data.