I use AI to study real reading behaviour and suggest calmer routes through complex
documents.
Our story
When I first started sketching Sethanoriva, I did not imagine a big platform. I
imagined a single reader at a desk, late in the evening, staring at a dense
document and wishing for a margin note that said “start here instead.” That
scene still guides how I make decisions. The product sits at the intersection of
reading analytics and workflow design. It does not just rank documents. It
studies the path you take through them, the pauses, the highlights, and the
sections that spark discussion. From those traces, it suggests smarter entry
points and sequences so you can spend time where it matters. I work with a
small, experienced group of designers and engineers who care about quiet
productivity. Our background spans knowledge management tools, natural language
processing, and collaboration software. We build slowly, test with real teams,
and adjust the system whenever the suggestions feel distracting or opaque. The
aim is transparency: you should always be able to see why the AI nudged you
toward a particular section or reading order.
Who we are
Today, Sethanoriva focuses on a simple promise: help people read less while understanding more. I see it working best for teams who live inside complex documents, like policy analysts, product managers, researchers, and operations leaders. They juggle long reports, scattered comments, and constant context switching. Instead of replacing their judgement, the system gives them a clearer map of what deserves attention first. I do not treat AI as a black box oracle. I treat it like a careful assistant that notices patterns humans ignore when tired or rushed. It watches which highlights you revisit, which sections you skim, and where conversations cluster. From there, it proposes reading sequences that feel like a well-edited playlist instead of a random queue. As Sethanoriva grows, I want to keep one constraint in place: every new feature must shorten the distance between an unread document and a confident decision. No vanity metrics, no noisy notifications, just quiet, observable improvements in how you and your team move through written work. Results may vary, and that is expected, because every organisation reads differently.
When I explain Sethanoriva to someone new, I often compare it with the tools they already use: document folders, email, chat, and generic note apps. Each solves a piece of the problem, but none of them watch how reading actually happens over time.
Traditional document management tools treat every file as a static object. They help you store and share, but they do not notice which pages people reread, which sections they skip, or where conversations stall. I built Sethanoriva to sit closer to the act of reading itself. It tracks patterns quietly in the background, then proposes better routes through familiar content, much like a navigation app that learns your commute instead of just listing all possible roads.
I also see a gap between chat-based summaries and real understanding. A quick summary can be helpful, but it often flattens nuance and hides the reasoning behind a document. With Sethanoriva, I wanted the AI to act more like a guide through the original text. It suggests a starting section, highlights likely pivot points, and points out where previous readers spent time, so you can still engage with the material directly rather than outsourcing judgement.
Over time, this approach builds a shared map of what matters inside your organisation’s documents. Instead of every new colleague asking, “Which part should I read?” they see suggested sequences shaped by past reading behaviour. The goal is not to promise perfect recommendations, because results may vary, but to reduce the friction between opening a long document and knowing where to begin.
Who is building Sethanoriva
Every role, from engineering to research, orbits around a simple question: does this change make complex reading feel lighter without hiding important detail? If the answer is unclear, we slow down and reassess before shipping.
Behind the product, there is a small group of people who care deeply about reading, writing, and the quiet infrastructure that supports both. I like to be clear about who is involved and how we work together.
I lead product direction and spend most of my time watching how teams actually use Sethanoriva. I
join onboarding sessions, review anonymised interaction logs, and turn those observations
into small, testable changes. This hands-on loop helps me avoid building features that look
impressive in a demo but do little for day-to-day reading workloads.
On the technical side, experienced engineers focus on natural language processing, recommendation systems, and systems reliability. They design the models that detect impactful sections and propose reading sequences, then stress-test them against messy, real-world documents. Rather than chasing abstract benchmarks, they ask whether a busy reader can reach a useful section sooner and with fewer distractions.
Design and research sit close to support. When someone struggles with a feature or feels
uncertain about a suggestion, that feedback goes directly into our design backlog. I prefer
this tight loop because it keeps us grounded in lived experience instead of internal
assumptions. Over time, this structure should make Sethanoriva feel less like a tool dropped into
your workflow and more like a companion that gradually learns how you prefer to read.
How we work
When I compare Sethanoriva to traditional reading tools, I see three main differences. First, it treats reading as a workflow, not a solitary act. Second, it uses behavioural signals like scroll depth, revisit patterns, and highlight clusters to suggest where you should focus. Third, it tries to surface the most impactful sections in a way that you can inspect, question, and refine over time.
Instead of pushing fixed templates, I follow a simple internal method I call the Reading Flow Loop. I observe how teams currently move through documents, model those paths inside the system, and then adjust the AI prompts and scoring until the suggestions match how people say they want to work over the next few years.
This approach keeps me honest. If a recommendation does not save clicks, reduce re-reading, or clarify ownership, it does not stay. I treat each deployment as an ongoing conversation with the team. Their feedback shapes which signals matter most, and the system evolves with their habits rather than forcing a rigid process on day one.
How I think about AI, reading, and meaningful focus
Privacy, security, and control
I take a conservative stance on data because reading patterns can reveal sensitive
priorities. By default, I aim to minimise exposure while still giving you enough insight to
improve how your team works with documents.
Because Sethanoriva works with reading behaviour, privacy and control matter as much as features. I design the system so teams in South Africa and beyond can understand what happens to their data and make informed choices.
Why this project exists
“I already have bookmarks and folders. Why would I need an AI to read with me?” I heard this so often that I started to treat it as a design brief. I wanted to see if I could turn scattered reading habits into a calm, guided workflow instead of yet another inbox of unfinished documents.
I built Sethanoriva because I kept drowning in reports, research papers, and long emails
that all felt equally urgent. Traditional tools only stored documents. I wanted
something that watched how I actually read and annotate, then quietly suggested what
to read next and where the real signal lived inside the noise.
