Scripture for the moment you’re in.

Research

I started with the market,
before committing to the product.

I wanted to understand where Bible apps already helped people, what disappointed them and whether there was room for another approach.

I built an ASO agent to collect the evidence. A separate tracker kept the source, storefront and date attached to each observation. I set the questions and reviewed what the material could actually support.

34apps compared

540reviews analysed

Bible Bff

A field for how you feel.

Personal-state input was already part of the market.

What does that
search give back?

The store sample showed the field, but not its results. I kept that distinction in the comparison.

The opening was in the experience around search.

Some apps returned a passage, others a conversation or a prayer. I compared the whole journey: what a person could say, what came back and where they went next.

Selected competitors from the store research.
Product	How you begin	What comes back	Where it leads
WordScoutVerse finding	Phrase, paraphrase, topic or reference	Passage list; a shown example recognises a paraphrase	Reader & search historyA visible path into the book.
HavenDaily Bible Verse	A situation in your own words	Verse matches with displayed confidence percentages	Reflect, Saved, JourneysFull reader not shown.
Bible BffState input & chat	A personal state, or a chat question	Chat with references; state-search results not shown	An open questionThe field did not establish the response.
Echo GracePersonal reflection	A longer personal entry	A verbatim verse, followed by a generated prayer	Entry history & activityThe passage leads into another practice.
From BibleCards & widgets	Mood and topic categories	Verse cards and widgets with references	Saved versesFree-text state input not shown.
Bible Verses AboutTopic index	An alphabetical topic list, filtered by typing	A collection of passages for the topic	Translation switchSearch works on topic names.
Bible ChatBook Vitals	Text, voice or a suggested question	Chat, explanations and Bible cards	Reader & daily journeyReading, reflection, tasks and streaks.
YouVersionBible reader	General search and reading navigation	Reading content; state-search response not studied here	Reading & plansA broad reference for the reader.

“Not shown” marks a gap in the sample, not an absent feature.

Language set expectations.

Bible Chat had Ukrainian store copy, while Ukrainian was absent from its listed interface languages. Reviews described that mismatch.

Access affected perceived value.

Bible Bff and From Bible reviews raised payment barriers. I separated those concerns from judgments about passage quality.

Unknowns remained visible.

Echo Grace’s pricing and current-version feedback were not established. The comparison retained those gaps.

Product direction

Your book,
every day.

I defined Dar as an everyday Bible reader. Search was another way into the same book: describe a situation, find a passage and continue in its chapter.

I wrote that direction into a PRFAQ, a future press release and the questions behind it. It gave me a way to decide what belonged in the first release.

Pick up where
you left off.

Open the book at the place you stopped. Read a chapter and return to passages you saved.

Bring your
own words.

Describe a situation to find a passage for the moment.

Keep reading
in context.

Open the surrounding chapter and read for as long as you want.

Preset emotions could wait.

Typing a personal situation was enough to test the idea. I left the second entry path out of the first release.

Prefer passages you haven’t seen.

I replaced “never repeats” with a specific rule: prefer unseen or least-seen passages, then cycle once everything has been shown.

Search experiments

A similar passage
could still be the wrong answer.

I was testing a hypothesis, not yet committing to a launch. I tried several approaches myself and discarded versions whose results failed the situation. The question was what a passage offered the person, beyond a match in topic.

Match the meaning of the passage.

Could semantic similarity find a useful response?

InputYour words
Query vector
Encode the query
Represent the person’s words as a vector.
Prepared indexPassage vectors
Query vectorCompare
Compare passages
Find nearby vectors in the passage index.
Selected referencesBook · chapter · verse
Look up in Bible corpusOriginal Scripture
Read the matches
Resolve the references to original Scripture.

What the test exposed

The topic matched. The support did not.

In my tests, anxiety about travel could bring back more fear, with too little reassurance. Similarity alone did not express the quality I was looking for.

Mechanisms reconstructed from my development reports. The approaches overlapped in time. Candidate letters illustrate a change of order, not measured scores.

Validation

The early iOS app gave me
a reason to continue.

I tested this early app with people before deciding whether to launch Dar. They could describe a situation, read the passages and rate the results. The interface was deliberately simple.

Users were impressed by the relevance of the passages and recognised moments when they would want this search. The conversations also raised other needs.

I kept the first release narrow.

There was more I could build. I chose to keep the scope small so I could release and learn more often.

Recovered Bible Search prototype with an English query, Matthew 11:28–30 and passage rating controls.

Design through my AI OS

The bird had to stand out
against the glass.

I used my AI OS to generate studies and develop the design system and screens in Figma. I judged the icon in stages: the symbol’s meaning, the bird’s pose and the space the glass left around it.

Symbol

The dove connects
faith with comfort.

Dar needed to suggest Scripture and support without an explanation beside the icon. Familiarity helped the dove, but also made it harder to distinguish from other religious apps.

Symbol comparisons recreated for this case.

Cross emblem recreated for the case — The cross makes the Christian connection immediate. Its simple shape survives at icon size, but gives Dar little of its own identity on a shelf full of religious apps.

Recreated study: a classical lyre with curved arms and fine strings — The curved frame is distinctive; the fine strings risk merging at icon size. Its music association would set the wrong expectation for Dar.

Book with a ribbon bookmark, recreated for the case — The book and bookmark make reading and returning to a passage clear. That fits Dar’s purpose, though the same shape could belong to almost any reading app.

Recreated study: a frontal dove with raised wings — The dove already suggests faith and comfort. That makes the meaning easier to establish, but leaves the pose and glass to distinguish Dar from other religious icons.

Pose

The turn made the bird
approach the reader.

The first poses felt flat or formal. Overlapping wings gave the bird depth, while the angle changed how it met the reader.

Recreated study: a dove gliding with wings spread horizontally — The wide silhouette is easy to recognise as a bird. The flat, frontal view loses the sense of approach I wanted for Dar.

Recreated study: a symmetrical dove landing with wings raised — Symmetry makes the shape stable and recognisable. It also feels ceremonial, which was too formal for an app meant to offer personal support.

The fuller feather contour gives the bird life. Its side-facing body sends it across the icon, so it still misses the feeling of approaching the reader.

Recreated study: a dove turning towards the viewer with asymmetrical wings — I chose this turn early on: the overlapping wings made the bird feel alive and directed towards the reader. I accepted less wing area to gain that depth.

Window

The glass had to leave
room for the bird.

Once the emblem was settled, I focused on the background. I compared window shapes in the same palette to judge the space each left for the dove.

Vault stained-glass window study without a dove — The pointed arch gives the icon height. The heavy surround takes too much of the square, leaving a narrow opening for the bird.

Arcade stained-glass window study without a dove — The arches and gold give it a clear church-window character. I kept it in reserve. The vertical bars could compete with the wings.

Tall arch stained-glass window study without a dove — The broad blue panels leave plenty of space for a bird. There are too few divisions to give me the stained-glass detail I wanted.

Rose stained-glass window study without a dove — The radial panes draw attention to the centre, where the bird sits. I kept the rose, though the top and bottom of this pattern still needed to match.

A clear bird against darker glass.

Later I chose a frontal pearlescent bird. Against the dark blue glass, the pale feathers stay distinct, while the gold sits around the edge.

Final app icon

Interface design

Search had to lead back into the book.

ReadTypography & chapter transitionsSearchField, results & selected passagesReturnHistory, preferences & states

Search, passage results and return paths belong in one flow. I put them beside the reader in Figma to check the transitions and reuse the same components.

Art direction

The reading needed
a rhythm beyond typography.

Engravings gave a long page of text moments of visual character. I explored whether reusable motifs could produce enough variety while keeping each scene coherent.

I built a blind comparison. The assembled scenes lost the unity I wanted; complete compositions held together. That changed the production approach.

Assembled motifsRecognisable pieces.
A less coherent scene.

Original whole-composition engraving for Psalm 46: a rocky river landscape — Whole composition · selectedThe linework belongs
to one landscape.

Psalm 46: original artwork, with the modular version rebuilt using the study’s saved preview layout.

Reuse the complete plate.
Preserve the character of the scene.

I dropped the visible module approach. Whole plates could be assigned to readings; any motion would need to respect the linework and layers of that particular composition.

In the native app

Search for a passage,
then keep reading.

I built these flows in SwiftUI so a useful search result could become part of someone’s regular reading. The captures use sample results, history and tags.

Native search screen with a personal query, previous searches and the keyboard. — **Describe the moment.**Previous queries stay close to the field.

Native light results screen with a sample query and the beginning of Psalm 91. — **Choose a passage.**Read enough of the result to decide where to go.

Psalm 1 with highlighted text and personal tags: Courage, Gratitude and Peace. — **Keep what matters.**Highlight a passage and give it a personal tag.

Interaction

Reading should
stay steady.

The reader sets the pace.

I kept continuous vertical scrolling. In the accepted engraving study, the image stops when the reader stops and reverses when they scroll back. The text keeps moving normally.

Glass belongs on the controls.

The search field uses native Liquid Glass. Across the full header, glass changed the paper’s tone and created a separate band. I brought the header back onto the same paper surface as the text.

BeforeThe keyboard disappears and returns.

AfterThe field and keyboard stay in place.

In use

A gesture should not
move its target.

Testing the app exposed movement I had missed in the designs. Opening a book could move its row away from the tap. Dragging past searches could make the keyboard disappear and return.

I changed the book list to open in place and kept the search field active during a history drag. The simulator recordings show the keyboard correction: the field stays where the person is using it.

Actual iOS simulator recordings. The corrected gestures still needed another check on the phone.

Scrolling still needed a phone check.

Reading felt uneven on the phone. I reduced the work happening during scrolling and checked the changes in the simulator. I still needed to test how they felt on the phone.

What I carry forward

Search had to help someone
before I built the rest.

I did not know at the start whether I wanted to launch Dar. Research gave me a direction. Testing the early iOS app with people gave me a reason to keep going.

My AI OS helped carry the work from one decision to the next. I still had to judge what the evidence meant, direct the design and decide how much product to build.

I started with the market, before committing to the product.

A field for how you feel.

What does that search give back?

The opening was in the experience around search.

Language set expectations.

Access affected perceived value.

Unknowns remained visible.

Your book, every day.

Pick up where you left off.

Bring your own words.

Keep reading in context.

Preset emotions could wait.

Prefer passages you haven’t seen.

A similar passage could still be the wrong answer.

Match the meaning of the passage.

Encode the query

Compare passages

Read the matches

The topic matched. The support did not.

Give the candidates a second reading.

Retrieve candidates

Rerank the set

Read the reordered set

Extra processing kept a familiar mismatch.

Match everyday situations first.

Encode the query

Match a situation

Follow the passage links

The full returned passage still mattered.

Ask a model for references.

Select references

Parse the response

Resolve the exact text

A parser defect looked like a model failure.

The early iOS app gave me a reason to continue.