The trick is simple: short slash-codes that tell an AI image tool exactly which camera, lens, angle, or film style to reinterpret your photo with.

Below are all 50 codes, grouped by vibe, plus how to use them and a few tips for better results.

How to use the codes

1. Open your AI image tool and start a new chat.

2. Upload your photo. A clear, high-resolution image works best.

3. Type a code, for example /polaroidimage.

4. Send the message. The AI reads the code as a visual transformation and applies the matching camera, perspective, or photographic treatment.

5. Try another photo and reuse the same code.

Example: Upload your photo, type /dronecam, and send. You’ll get back an elevated, overhead drone-style version with your subject still recognizable.

Pro tips

Keep the person exactly the same. Add this line after your code:

“Keep the person’s identity, face, body, clothing and pose unchanged; only change the camera/perspective treatment.”

Control the strength. Add a short instruction like “make it subtle” or “make the camera effect obvious.”

Remember the rule behind the codes. They change how the photo feels captured, not just the color filter. Camera position, lens behavior, film medium, framing, distortion, and capture artifacts all shape the result.

The 50 codes

📸 Instant & Retro Snapshot Looks

• /polaroidimage: Instant-film print with a white border, soft flash, analog grain, and slightly faded, nostalgic colors.

• /vintagecamera: A 1970s–80s consumer-camera look with warm film tones, subtle grain, and imperfect exposure.

• /photostrip: Several sequential frames of the same moment stacked vertically like a developed photo strip.

• /oldb&wphoto: Aged black-and-white film with silver-gelatin texture, scratches, faded contrast, and old photo paper feel.

• /photoboothprint: A tightly framed portrait with direct flash, a spontaneous expression, and authentic booth-print texture.

• /disposablecamera: Harsh built-in flash, imperfect focus, heavy grain, warm skin tones, and an accidental-looking composition.

• /instantflash: A close-range snapshot with bright flash falloff, soft analog colors, mild overexposure, and instant-film imperfections.

🎞️ Film Stocks, Formats & Darkroom

• /35mmfilm: Natural analog grain, realistic film halation, subtle color shifts, and classic photographic depth.

• /mediumformat: Highly detailed, with soft tonal transitions, shallow depth of field, and refined rendering.

• /35mmcontactsheet: Several small film frames arranged like a photographer’s analog proof sheet.

• /darkroomprint: Traditional photographic paper with deep blacks, chemical-print texture, and imperfect development.

• /negativefilm: Inverted colors and tones with visible film-strip characteristics and a translucent negative texture.

• /slidefilm: Saturated colors, crisp detail, and rich contrast, like projected transparency film.

• /halfframecamera: Two portrait-oriented photos presented together in one frame, mimicking a half-frame film camera.

• /panoramicfilm: An unusually wide horizontal composition with analog grain and cinematic framing.

📼 Old-School Video Frames

• /camcorder: Early digital-video aesthetic with low dynamic range, slight motion softness, and consumer camcorder character.

• /vhsframe: A VHS freeze-frame with analog noise, scanlines, color bleeding, and tracking imperfections.

• /minidv: Early-2000s digital video with interlaced-looking detail, harsh highlights, and consumer-camera colors.

• /8mmfilm: Strong grain, flicker, rounded frame character, warm colors, and a handmade home-movie feel.

• /super8: Nostalgic cinematic grain, soft focus, warm exposure, light leaks, and amateur-film atmosphere.

🔭 Lens Effects

• /fisheye: Extreme ultra-wide perspective with curved edges, enlarged foreground elements, and dramatic distortion.

• /ultrawide: Expansive environmental framing with strong foreground-to-background depth and exaggerated perspective.

• /telephoto: Compressed background, isolated subject, flattened depth, and creamy distant scenery.

• /macroshot: An extremely close view that emphasizes fine texture while the background melts out of focus.

• /tiltshift: A miniature-world effect with a narrow sharp plane and heavily blurred surroundings.

• /softfocuslens: Glowing highlights, diffused edges, dreamy skin rendering, and reduced micro-contrast.

📐 Camera Angles & Positions

• /dronecam: An elevated overhead perspective looking down, with the environment dominating the composition.

• /birdseye: A nearly vertical top-down view that turns the scene into a graphic composition.

• /wormsview: A camera near ground level looking up, making the subject and architecture appear towering.

• /groundlevel: The camera almost touching the ground, with foreground texture dominating the frame.

• /overtheshoulder: The camera sits behind the subject, showing part of their shoulder or head and what they’re looking toward.

🪞 Mirrors, Glass & Windows

• /convexmirrorselfie: Shot through a curved mirror with optical distortion and an expanded peripheral view.

• /sideviewmirror: The subject appears reflected in a car’s side mirror while the road or surroundings stay visible.

• /rearviewmirror: The scene is captured through a car’s rear-view mirror, framed by the vehicle interior.

• /rearwindowcam: The camera sits inside a vehicle looking out through the rear windshield, with the car interior framing the scene.

• /reflectionshot: The subject photographed through reflective glass, blending reflections with the real scene.

• /windowreflection: A window-side portrait with natural reflections, layered foreground glass, and soft distortion.

• /prismreflection: Refracted duplicate fragments, rainbow flares, and geometric reflections around a recognizable subject.

• /crystalball: The scene appears refracted inside a glass sphere in the foreground while the environment stays visible around it.

• /doorpeephole: A circular vignette with strong barrel distortion, as if seen through a door peephole.

🌗 Light & Shadow

• /shadowcam: A composition built around the subject’s shadow, creating a deliberate relationship between person and cast shadow.

• /silhouettecam: Strong backlighting turns the subject into a dark silhouette against a bright sky or glowing background.

• /flashcam: Frontal on-camera flash, a sharply lit subject, darker surroundings, and a spontaneous snapshot feel.

🕵️ Surveillance, Night & Thermal

• /cctv: A fixed high-corner view with wide-angle distortion, timestamp-style overlay, and compressed image quality.

• /securitymonitor: The image appears recorded off a monitor, complete with screen glare, compression artifacts, and slight moiré.

• /nightvision: Green-tinted low-light imagery with infrared-style illumination and electronic noise.

• /infraredphoto: A surreal tonal response where foliage turns unusually bright and skies go dramatically dark.

• /thermalcam: A heat-map visualization that separates the subject from the environment through temperature patterns.

🎥 First-Person POV Cams

• /bodycam: A chest-mounted, first-person viewpoint with wide-angle distortion and natural environmental framing.

• /helmetcam: An action-camera view from head height, with a very wide field of view and dynamic first-person composition.

Final thoughts

The fun part is mixing and matching. Run the same portrait through /polaroidimage, /cctv, /dronecam, and /vhsframe, and you’ll get four completely different “photographers” from a single image.

Start with your favorite photo, pick a code, and see where it takes you. Then try a different photo with the same code and compare.

Save this post so you can come back to the full list anytime, and share it with a friend who loves experimenting with photos.

On 6 October 2026 Google released Nano Banana 2.1, the newest version of the image model inside the Gemini app. It’s free to try in the Gemini app (your plan’s daily image limit applies). Google says it’s better at keeping a face consistent, at quality, and at writing text inside an image. That makes it perfect for festive photos: give it one clear selfie, describe the Diwali scene, and you get a photo that looks like you.

What you need

• The Gemini app (iPhone or Android), updated, signed in with your Google account.

• One clear selfie: face the camera, good light, nothing covering your face, just you in the photo.

• For a couple or family photo: one clear photo of each person, added to the same message. Only use photos of people who are happy for you to do this.

How to do it

1. Open the Gemini app.

2. Tap + and add your selfie from your photos.

3. Paste one of the prompts below and send.

4. Don’t like it? Don’t start again. Reply with one change, for example: “Keep my face exactly the same as the original photo, only change the background.”

5. Download the one you like.

10 Diwali prompts (copy and paste)

1. Classic Diwali night: “Using my photo, create a realistic portrait of me on Diwali night, wearing a cream silk kurta, standing on a balcony lit with clay diyas and warm string lights, soft golden glow, background softly blurred. Keep my face exactly the same as in the photo.”

2. Rangoli: “Using my photo, create a realistic photo of me sitting next to a colourful flower-and-colour rangoli on the floor, with diyas around it, wearing traditional Indian clothes, shot from a slightly high angle, warm evening light. Keep my face exactly the same as in the photo.”

3. Saree or lehenga: “Using my photo, create a realistic photo of me in a red and gold silk saree, holding a lit diya in both hands, warm festive lights behind me, elegant and natural. Keep my face exactly the same as in the photo.”

4. Sparkler shot: “Using my photo, create a cinematic night photo of me smiling and holding a phuljhadi (sparkler), golden light trails around me, a festive lit-up street softly blurred behind. Keep my face exactly the same as in the photo.”

5. Couple (add both photos): “Using these two photos, create a realistic Diwali portrait of the two of us together in traditional outfits, standing in a doorway decorated with marigold garlands and diyas, warm golden light. Keep both faces exactly the same as in the photos.”

6. Family (add each person’s photo): “Using these photos, create a realistic family Diwali photo of all of us sitting together at home, diyas and flowers around us, everyone in festive clothes, warm and happy. Keep every face exactly the same as in the photos.”

7. Greeting card with your name: “Make a Diwali greeting card using my photo, maroon and gold design with a border of diyas and marigolds, and the text ‘Happy Diwali from [your name]’ in elegant gold lettering. Keep my face exactly the same as in the photo.”

8. Professional post (LinkedIn / office group): “Make a clean, professional Diwali greeting image with my photo, minimal design with soft gold accents and a few diyas, and the text ‘Wishing you a bright and prosperous Diwali’ in a modern font. Keep my face exactly the same as in the photo.”

9. 90s family album: “Turn my photo into a 1990s film photo of Diwali at home: slightly grainy, warm faded colours, diyas on the windowsill, a festive living room behind me. Keep my face exactly the same as in the photo.”

10. Watercolour painting: “Turn my photo into a watercolour painting of me on Diwali night, glowing diyas and soft gold and orange washes of colour. Keep my face recognisable.”

Tip: add “4:5 portrait” for an Instagram post or “9:16 vertical” for a story or WhatsApp status at the end of any prompt.

Before you post

• Free has a daily limit. Google hadn’t published new limits for 2.1 at launch, so your plan’s usual daily image allowance applies. Spend it on the prompts you like most.

• AI images are marked. Gemini adds an invisible SynthID watermark to every image. In some countries you can switch off the small visible mark in Settings › Media Watermark; the invisible one always stays.

• Ask first. Only upload photos of people who have said yes.

• Faces can still drift. If it doesn’t look like you, reply asking it to keep your face exactly the same, and try again.

Sources

• Unite.ai: Nano Banana 2.1 debuts at half the image cost of its predecessor (October 2026). Release date, free in the Gemini app with daily limits, improvements.

• Tech Insider: Nano Banana 2.1 launches, cuts image price 50% (7 Oct 2026). Based on Google DeepMind’s model card and pricing page.

• PCWorld: Google now lets you nix visible Gemini image watermarks (August 2026). Settings › Media Watermark; SynthID stays.

Found this useful? Follow me on Instagram for a new AI tool every day.

You give it something written — a product concept, a pitch, a news story, a launch announcement — and tell it in plain English what you want to know.

It builds a digital world populated with thousands of AI agents. Each one has its own personality, its own memory that persists across the simulation, and its own behavioural logic. They talk to each other. They argue. They change each other’s minds.

Then it tells you what the crowd concluded, and why.

The use cases people have actually run with it are stranger than the pitch suggests. Public opinion forecasting on real controversies. Financial and political prediction. And in the demo that got it attention, predicting the lost ending of Dream of the Red Chamber — the Chinese classic whose final chapters were never finished.

How it works under the hood

Four stages, and knowing them changes how you write your input.

1. Graph building. It extracts the key entities from your material and builds a knowledge graph using GraphRAG. Everything downstream depends on this, which means vague input produces a thin graph and a useless simulation.

2. Environment setup. It pulls the relationships between entities and generates the agent personas from them. The personas come out of your material, not from a generic template — so the crowd you get reflects the world you described.

3. Simulation. Agents run across two parallel social platforms simultaneously, with dynamic temporal memory so they remember what they said and who said what to them. This is where the arguing happens.

4. Report generation. A dedicated report agent analyses the whole interaction log and produces the prediction. It has its own toolset, so you can interrogate the result afterwards rather than just reading a static output.

The practical takeaway: stage one determines everything. Give it a detailed concept document, not a one-line idea.

Which version to actually run

This is the part the GitHub page won’t make obvious. There are three routes and they have very different costs.

Route 1 — The original repo

github.com/666ghj/MiroFish

The canonical project, 75,000+ stars. Chinese-language interface. Uses Zep Cloud for graph memory and DashScope or OpenAI for the model calls.

Cost: you pay per API call, and a simulation with thousands of agents interacting over multiple rounds is a lot of calls. Run a small simulation first and watch the bill before you scale it up.

Use this if: you want the original, you’re comfortable with a Chinese UI, and you’d rather pay for API than manage infrastructure.

Route 2 — The offline English fork

github.com/nikmcfly/MiroFish-Offline

Fully local, fully English. Over 1,000 interface strings translated. Replaces Zep Cloud with Neo4j Community Edition and swaps the cloud model for Ollama running locally.

What you need:

• Docker and Docker Compose (the easy path), or Python 3.11+, Node 18+, Neo4j 5.15+ and Ollama manually

• Two models pulled: qwen2.5:32b for the LLM, nomic-embed-text for embeddings

• 16GB RAM minimum, 32GB recommended

• 10GB GPU VRAM minimum, 24GB recommended

• 20GB storage minimum, 50GB recommended

Setup: clone it, copy .env.example to .env, run docker compose up -d, pull the two models.

If your machine is lighter, you can drop to qwen2.5:14b or 7b. CPU-only works but is slow enough that you’ll want to start a simulation and go do something else.

Cost: zero per run, after the hardware. This is the version to use if you’re going to run this repeatedly.

Route 3 — CLI forks

Several exist, including amadad/mirofish-cli. Lighter, scriptable, no web interface. Good if you want to wire a simulation into an existing workflow rather than click through a UI.

My honest take on what it’s for

The creator’s own framing is the right one: this is a rehearsal space, not a crystal ball.

A simulated crowd is a crowd of language models reproducing patterns from their training data. It will tell you how a plausible-sounding population might react. It will not tell you how your actual customers will react, and anyone selling it as a replacement for talking to real people is selling you something.

Where it genuinely earns its place:

Finding the objection you didn’t think of. Thousands of agents attacking your pitch from thousands of angles will surface the one framing problem you’re blind to. That’s worth the setup cost on its own.

Stress-testing before an expensive launch. If the thing you’re about to ship costs real money to get wrong, a dry run is cheap insurance.

Pressure-testing messaging variants. Run the same product with three different positioning statements and see which crowd reaction looks healthiest. This is the use case I’d actually build a workflow around.

Where it will mislead you: anything where the real answer depends on information the models don’t have. Local market conditions, your specific customer base, pricing sensitivity in your category. The simulation will produce a confident answer anyway. That’s the trap.

If you’re going to try it

Start with the offline fork if you have the hardware — the per-run cost of the API version adds up faster than you’d expect with multi-round simulations.

Write a proper input document. One paragraph produces a thin knowledge graph and a simulation that tells you nothing.

And run something you already know the answer to first. Take a product launch that already happened, feed it in, and see whether the simulated crowd predicted what actually occurred. That’s the only way to calibrate how much to trust it on something you don’t know yet.

^{The four stages of a MiroFish run. Stage one determines the quality of everything after it.}

Repo: github.com/666ghj/MiroFish · Offline English fork: github.com/nikmcfly/MiroFish-Offline

Researchers led by Stanford published this in Science. They tested 11 major AI models — ChatGPT, Claude, Gemini among them — by feeding them real situations people had described in online forums, then comparing the AI’s response to what actual humans said about the same situation.

Across all 11 models, the AI endorsed the user’s actions 49% more often than humans did. Including in cases involving deception, illegal behaviour, and other genuine harm.

Two findings matter more than the headline number:

It’s not a bug in one model. Every system tested showed it. It comes from how these models are trained — reinforcement learning from human feedback rewards responses people rate highly, and people rate agreement highly.

You can’t feel it happening. Participants rated the sycophantic responses as more trustworthy and said they’d use that chatbot again. The same people were then less likely to admit fault in a conflict and more convinced they were right. Even participants who went in sceptical of chatbots were affected.

That last part is why a fix is worth bothering with. You will not notice this on your own.

The base instruction

Open ChatGPT → Settings → Personalization → Custom Instructions. Paste this:

Be direct and honest with me, not agreeable. Challenge my assumptions when they’re weak. If I’m wrong, tell me I’m wrong and explain exactly why. Rate my ideas honestly out of 10, and if you’re not sure, say so instead of guessing.

Save it. It applies to every new chat from that point. Existing chats keep their old behaviour, so start a fresh one to test.

One thing worth knowing: this is Custom Instructions, not memory. They’re separate features. Custom Instructions get applied to new conversations as standing context; memory stores facts about you. Plenty of posts conflate the two.

Three variants that work better

The generic instruction is fine. These are sharper because they’re scoped to what you actually use the model for.

If you use it for work and decisions:

Be direct, not agreeable. When I describe a plan, your first job is to find the strongest objection to it, not to improve it. Tell me what would have to be true for this to fail. If I’m wrong, say so plainly and explain why. If you’re uncertain, say “I’m not sure” instead of producing a confident answer.

If you use it for writing:

Don’t praise my drafts. Start every critique with the weakest part and why it’s weak. Don’t soften criticism with compliments first. If a piece doesn’t work, say it doesn’t work. Rate honestly out of 10 and tell me what a 9 would require.

If you use it for code:

Don’t tell me code is good. Tell me what breaks it. Identify the failure mode, the edge case I haven’t handled, and the thing that will cause problems at scale before you comment on anything that works. If my approach is wrong, say so before writing the implementation.

Use one. Stacking all three makes the instruction long enough that the model starts ignoring parts of it.

How to actually verify it worked

This is the part most posts skip, and it’s the only part that matters.

Don’t ask ChatGPT to audit itself. The obvious test — “what have you been agreeing with me about that I’m actually wrong on?” — doesn’t work. It can’t review your past conversations and grade its own behaviour. It will generate something that sounds like an audit, and it will feel shocking, and it will be made up. That’s sycophancy producing a performance of anti-sycophancy.

Do this instead. Take an idea you already believe is strong — a business plan, a piece of writing, an approach to a problem. Something you have a stake in.

Ask for an honest rating out of 10, in a fresh chat, before you save the instruction. Note the number and the tone.

Save the instruction. Open a new chat. Ask the same question about the same idea.

If the rating drops or the response leads with a problem instead of a compliment, it’s working. If you get the same warm 8/10 both times, the instruction isn’t landing and you need a sharper variant.

The one-word trick from the study itself

Buried in the research: the Stanford team found that simply making a model begin its response with “wait a minute” made it noticeably more critical and less agreeable.

You can use this per-message without changing any settings. Start your prompt with “Wait a minute — before you answer, tell me what’s wrong with this.” It’s crude and it works.

When you should turn this off

Honesty settings have a cost, and nobody mentions it.

Early-stage brainstorming. If you’re generating options and the model attacks each one as it arrives, you’ll stop generating. Turn it off, get the ideas out, turn it back on to evaluate them.

Emotional conversations. If you’re using it to think through something difficult, a model instructed to find the weakest point in everything you say is the wrong tool. That’s not what you need from it.

When you’ve already decided. If the decision is made and you’re executing, a critique loop is friction, not insight.

The point isn’t to make the model hostile. It’s to make sure that when you ask whether something is good, the answer means something.

What this doesn’t fix

A custom instruction changes the output. It doesn’t change the training.

The Stanford researchers were blunt about why this persists: sycophancy is preferred by users and it drives engagement, so there’s little commercial incentive to reduce it. Their study found people were 13% more likely to prefer the flattering model over the honest one.

So the instruction is a patch on your side of the interface. The underlying pull is still there, and it will drift back in long conversations as the model picks up your tone and position. Re-anchor it occasionally — drop “be honest with me here, not agreeable” into a long thread and watch whether the answers change.

If they do, that tells you something.

Source: “Social sycophancy in large language models,” Stanford University, published in Science. 11 models tested, 2,400+ participants.

It’s real and it’s good at coding. 77.9% on DeepSWE v1.1, the best published score, ahead of Claude Opus 5.5 at 74.2% and GPT-6 Astra at 74.1%.

It doesn’t win everything. Astra beats it by 10.5 points on FrontierSWE v2. Opus 5.5 beats it by 9 points on Terminal-Bench 4.0.

The cheap price is temporary. $2/$10 per million tokens is introductory, rising to $4/$20 — at which point it costs exactly the same as Claude Opus 5.5.

You almost certainly can’t use it. Access is limited to vetted cyber defenders through Google’s Fairwind Program. No public date announced.

Google’s own staff are split on it. Bloomberg reported internal doubts about real-world coding performance the same day it launched.

1. The scorecard

Every number below is vendor-reported, from Google’s own comparison table at launch. No third party had independently reproduced the full table at the time of writing. Argon leads 12 of the 18 benchmarks Google published.

DeepSWE v1.1 — long-horizon software engineering
Argon 77.9% · Opus 5.5 74.2% · Astra 74.1% · Fable 5.1 67.4%

FrontierSWE v2 — harder SWE tasks
Astra 65.5% · Argon 55.0%

Terminal-Bench 4.0 — real command-line work
Opus 5.5 66.4% · Argon 57.4%

CWE-bench v1 — vulnerability remediation
Argon 68.0% · Astra 68.0% (tied) · Opus 5.5 67.0%

Gray Swan IPI — prompt-injection attack success, lower is better
Argon 0.7% · Opus 5.5 1.0% · Fable 5.1 1.0% · Astra 8.5%

Vals Index — finance, legal, tax, coding
Argon 68.9%

AutomationBench — end-to-end business tasks
Argon 51.3%

Artificial Analysis Intelligence Index — high reasoning
Opus 5.5 57.6 · Argon 53 · Astra 53

How to read the headline number

The 3.7-point DeepSWE lead over Opus 5.5 is real but modest. For scale, that’s roughly the gap between Claude Opus 5 and Opus 5.5 — one vendor, one generation apart.

Google’s own Gemini 3.8 Flash already scores around 74% on the same test, so Argon sits about four points above Google’s previous model. That’s an improvement, not a leap from behind.

And 77.9% is Google’s own run. Clean separation between these models needs a same-conditions rerun by someone other than Google, which hadn’t happened when I wrote this.

The number nobody is talking about

On Gray Swan’s indirect prompt-injection benchmark, Argon posts a 0.7% attack success rate against GPT-6 Astra’s 8.5%.

If you’re building agents that read untrusted web pages, documents or emails, this is arguably the most consequential figure in the entire table — more than any coding score. For context, Google’s previous Gemini 3.6 Flash scored 49% on the same test back in July.

2. What it actually costs

Argon launched at $2 per million input tokens and $10 per million output. Google has said this rises to $4/$20 after an introductory period it hasn’t dated.

Current rates per million tokens, input / output:

• Gemini 4 Argon, introductory — $2 / $10

• Gemini 4 Argon, post-introductory — $4 / $20

• Claude Opus 5.5 — $4 / $20 (cached reads $0.20)

• Claude Fable 5.1 — $10 / $50 (cached reads $0.25)

• GPT-6 Astra, up to 272K input — $10 / $50 (cached reads $1.00)

• GPT-6 Astra, above 272K input — $20 / $75

The “80% cheaper” claim, checked

It holds, but only against two models and only for now.

At the introductory rate, Argon is 80% below GPT-6 Astra and Claude Fable 5.1, which both list $10/$50. Against Claude Opus 5.5 at $4/$20, it’s 50% cheaper today — and exactly the same price once the introductory period ends.

A worked example

A job sending 1M input tokens and generating 200K output tokens, at list rates:

• Argon, introductory — $4.00

• Argon, post-introductory — $8.00

• Claude Opus 5.5 — $8.00

• Claude Fable 5.1 — $20.00

• GPT-6 Astra — $20.00

Per-token price is not per-task cost

This is the trap in every pricing comparison you’ll read this month.

Early independent testing found Argon uses more than twice as many tokens per task as GPT-6 Astra. A model that charges a fifth as much per token but writes 2.5x as many tokens is not five times cheaper in practice.

Artificial Analysis measured roughly $1.99 per index task for Argon against $3.26 for Astra and $7.63 for Fable 5.1. A real advantage — but far narrower than the headline rate card suggests. And once Argon moves to $4/$20 while still writing 2x the tokens, the arithmetic may invert against Opus 5.5 entirely.

Benchmark your own workload before switching anything. Nobody’s published cost-per-task number is your cost-per-task number.

3. The million-token thing

Most coverage got this wrong, so it’s worth being precise.

Google raised the maximum output from 64,000 tokens to 1,000,000 — how much the model can write in a single response. That’s a 15x increase and it’s genuinely unusual. GPT-6 Astra and the Claude models cap output at 128,000 tokens per request.

It is not the context window, which is how much you can send in. Context windows at this tier already sit around 1M tokens across all four models, and Google didn’t publish a new context figure for Argon. Anyone telling you Argon has “a new 1M context window” is repeating a misreading.

What it unlocks in practice:

• Generating an entire codebase, migration or test suite in one pass, instead of stitching together chunked responses

• Long-form research output with no continuation loop and no state-management headaches

• Multi-stage agentic work where the full plan plus execution lives in one response

The honest caveat: a 1M-token response costs $10 at the introductory output rate and $20 after. It’s a capability, not something you reach for casually.

4. Access reality check

At launch on 30 September 2026, Argon went to exactly one group: vetted cybersecurity defenders through Google’s Fairwind Program, alongside the company’s participation in the US government’s voluntary pre-release model access process.

Not paying API customers. Not Google AI Ultra subscribers. Not Pro, Plus or free Gemini users.

Google has stated the order of the queue — paid API customers and AI Ultra subscribers come next — but has published no date. If you opened the Gemini app looking for it and didn’t find it, nothing is wrong with your account.

Argon also replaces the Gemini 3.5 Pro release Google announced at its May I/O conference and then scrapped, a cancellation Bloomberg reported as costly in both time and money.

Google’s own staff are not unanimous

On launch day, Bloomberg reported that some Google employees with direct access found Argon performs worse on real coding work than its benchmarks suggest, with front-end design singled out. Two sources described the scores as affected by “benchmaxxing” — optimising for the test rather than the task.

Google disputed this directly, saying it would be inaccurate to claim Gemini 4 underperforms in coding, and one employee cited a “large consensus” internally that the model is at the frontier. Other staff told Bloomberg the opposite of the critics: that Gemini has caught up.

The picture is genuinely split, not settled. Alphabet shares pared gains from over 2% to 0.5% on the day the report landed.

Treat both the benchmark table and the internal criticism as claims awaiting independent testing.

5. What to do right now

Don’t wait for Argon

There’s no announced availability date and no way to join the queue. Anything you’re shipping in the next quarter should be built on a model you can actually call today. Treat Argon as something to re-evaluate when it opens up, not something to plan around.

If you’re choosing a coding model today

Claude Opus 5.5 at $4/$20 is the value pick at this tier. It leads Terminal-Bench 4.0 and the Artificial Analysis Intelligence Index, and its $0.20 cached-read rate matters enormously for agents re-sending the same context.

GPT-6 Astra leads FrontierSWE v2 by a wide margin and writes notably fewer tokens per task, which narrows its real cost gap despite the higher rate card. Watch the above-272K input repricing — it doubles input and raises output to $75.

Claude Fable 5.1 sits at the same $10/$50 as Astra and trails both on DeepSWE. Reserve it for work where its particular strengths earn the premium.

If prompt injection is in your threat model

This is the one area where Argon’s lead is large enough to be worth waiting for. A 0.7% attack success rate against Astra’s 8.5% is an order-of-magnitude difference, and it ships with chain-of-thought and action monitors that can halt execution mid-task.

If you’re building agents that browse the open web or process inbound email, put Argon on your evaluation list for the day it opens.

Three things to watch

Independent benchmark reruns. Google’s table is vendor-reported. The first same-conditions third-party comparison will tell you far more than anything published so far.

The introductory period ending. When Argon moves to $4/$20, it’s priced identically to Opus 5.5 while using roughly twice the tokens per task. That changes the calculus completely.

General availability. Paid API and AI Ultra are next in the queue. Until there’s a date, there’s no plan to make.

The one-line verdict

Argon is a real frontier model with a genuine lead in agentic coding and a commanding one in prompt-injection resistance — sold at a price that’s temporary, measured on benchmarks nobody outside Google has reproduced, and available to almost nobody.

Interesting. Not yet actionable.

Sources: Google DeepMind launch announcement (Koray Kavukcuoglu, 30 Sep 2026), TechCrunch, Decrypt, DevX, Bloomberg (Julia Love and Davey Alba), 9to5Google, The Next Web, Artificial Analysis, NeuralTrust, The Stack, and Anthropic’s and OpenAI’s published API pricing pages.

🌍 Introduction

Online education platforms like Coursera or edX offer thousands of courses, but learners often face the same problem: too many choices and not enough guidance.

This case study explores how a Personalized Learning Management System (LMS) can help students discover the right courses, stay motivated, and actually complete them.

🎯 The Goal

Design a recommendation system that:

1. Increases engagement by 25% (more weekly activity)

2. Improves completion rates by 30% (fewer drop-offs)

3. Enhances satisfaction by 20% (better feedback scores)

🧩 The Problem

1. Course Discovery Gap – Learners waste time browsing instead of learning.

2. Low Completion Rates – Many drop out mid-course without motivation.

3. Weak Feedback Loop – Platforms rarely adapt to user behavior in real-time.

💡 The Solution

A personalized recommendation system with four core modules:

1. User Profile Module – Tracks interests, goals, and learning history.

2. Recommendation Engine – Uses collaborative + content-based filtering to suggest courses.

3. Progress Tracking System – Visual progress bars, milestones, and streaks.

4. Feedback Loop – Collects ratings and behavior data to refine recommendations.

👥 User Personas

• Learners – Students/professionals seeking career growth.

• Content Creators – Instructors who need insights into learner behavior.

• Admins – Platform managers tracking system performance.

🛠 Key Features

• Personalized Course Recommendations – Based on past activity + goals.

• AI-Driven Learning Paths – Suggested skill journeys.

• Progress Visualization – Keeps learners motivated.

• Course Popularity Metrics – Helps learners trust their choices.

• Real-Time Alerts – Notifications on trending or relevant courses.

🔮 Future Enhancements

• Conversational AI – Chatbots for real-time course recommendations.

• Gamification – Rewards and streaks for completions.

• Peer Recommendations – Social proof via networks.

• Dynamic Goals – Learners set and adapt their learning paths.

✅ Conclusion

A Personalized LMS can bridge the gap between too many options and too little guidance. By combining AI-driven recommendations, progress tracking, and strong feedback loops, platforms can boost engagement, completion, and satisfaction.

The big lesson: personalization is the difference between learners browsing courses — and actually completing them.

View Complete Document

🌍 Introduction

Rapido has emerged as India’s largest bike taxi player, commanding a 65% market share in this segment. With affordability and reach into Tier 2 and Tier 3 cities, it has disrupted the ride-hailing landscape.

But for many potential users, bike taxis are still not the default choice. In this case study, I explore how Rapido can increase adoption by addressing trust, safety, and usability challenges.

🎯 The Problem

Despite market dominance, many users hesitate to choose Rapido because of:

• Safety Concerns – Unhygienic helmets, lack of visible bike condition.

• Driver Professionalism – Inconsistent rider behavior and lack of training.

• Pricing Perception – Competitive for short trips, but comparable to autos/cabs for longer rides.

• Gender Gap – Female riders make up just 31% of users.

👥 User Insights

From interviews and surveys, here’s what stood out:

• Why users prefer Rapido: Faster in traffic, affordable for short trips, convenient for solo rides.

• Why they avoid Rapido: Concerns about safety, hygiene, professionalism, and price surges.

• Female riders: Often uncomfortable with gear bikes and hesitant due to safety concerns.

💡 Strategic Recommendations

🔹 Low-Effort, High-Impact Ideas

1. Driver Info Transparency – Show full-screen driver image + license verification in-app.

2. Bike Maintenance Info – Display last service date & condition when booking.

3. Gearless Bike Filter for Women – Let female riders opt for more comfortable rides.

🔹 High-Effort, Massive-Impact Ideas

1. Female Captain Program – Onboard women captains for women riders to boost trust & empowerment.

2. Surge-Free Short Rides – Keep 5–10 km rides surge-free to strengthen adoption.

3. Driver Dashboard + Gamification – Incentivize captains for punctuality, safety, and customer satisfaction.

📊 Prioritization with RICE

Using the RICE framework, the highest scoring initiatives included:

• Driver Info Transparency (highest reach & confidence, low effort)

• Pricing Mechanism Changes (significant retention boost)

• Female Captain Program (massive impact on safety perception, though higher effort)

✅ Conclusion

Rapido has already transformed urban mobility, but the next stage of growth depends on trust and inclusivity. By improving transparency, empowering women riders and captains, and ensuring fair pricing, Rapido can shift from being a backup option to a daily commute essential.

The lesson here: winning adoption in mobility isn’t just about affordability — it’s about trust, safety, and reliability.

View Complete Document

🌍 Introduction

Caber, a new entrant in India’s cab aggregation market, has been relying on discount-heavy strategies to capture users. While effective for acquisition, this approach does little for sustained revenue growth.

To solve this, I proposed a monthly premium subscription plan — designed to increase Average Revenue Per User (ARPU) while improving customer retention.

🎯 The Business Problem

• Current State: One-off discounts attract users but fail to drive loyalty.

• User Pain Points: Riders face unpredictable surge pricing and lack consistent benefits.

• Business Need: Increase ARPU and build a predictable revenue stream.

⭐ The Hypothesis

A monthly subscription plan with exclusive benefits can:

• Encourage frequent riders to stay loyal

• Smoothen price sensitivity through capped surge pricing

• Create a predictable revenue stream for Caber

🧩 The Solution: Caber Premium Plan

For ₹299/month, subscribers get:

• 15% Discount on all rides

• Priority Ride Allocation during peak hours

• Capped Surge Pricing (max 1.5x)

• Partner Offers (restaurants, cafes, entertainment)

• Free Ride Support (1 instance/month)

👥 Target Segments

1. Frequent Riders – 8–10 rides/month (commuters, business travelers).

2. Price-Sensitive Users – students, young professionals.

3. Loyalty-Driven Riders – users who prefer consistency and perks.

🛠 Experiment Plan

1. Alpha Testing – Internal testing of discounts, surge logic, and payment flows.

2. Beta Rollout (2 Cities) – Pilot in Mumbai & Delhi, with A/B testing to track ARPU impact.

3. Feedback Loop – In-app surveys + NPS tracking.

4. Iteration – Adjust pricing and perks based on adoption.

5. Nationwide Rollout – Scale with marketing campaigns.

📊 Success Metrics

• ARPU Growth → +15–20% within 3 months

• Subscriber Uptake → 8–10% of active riders in 6 months

• Ride Frequency → 30% higher among premium users

• Churn Reduction → 10–15% lower for subscribers

• Customer Satisfaction → Higher NPS vs. non-subscribers

✅ Conclusion

Caber Premium is more than a discount plan — it’s a loyalty engine. By giving frequent riders predictable value and perks, Caber can reduce churn, increase ARPU, and strengthen its position against larger competitors.

The big lesson: in mobility, loyalty comes not from discounts, but from reliability and added value.

View Complete Document

🌍 Introduction

The fitness app market is crowded, yet most apps struggle with the same challenges:

• Generic workout plans that don’t fit individual needs

• Low motivation that causes drop-offs

• Overwhelming interfaces that frustrate users

In this case study, I break down the vision for Godbod, a fitness app designed to be part of a user’s daily routine, guiding them toward their goals with AI-driven personalization and engaging UX.

🎯 The Goal

Make Godbod the go-to fitness companion by:

• Delivering personalized workout plans

• Driving consistent daily engagement

• Building retention through gamification and social features

Key success metrics include:

• High DAU (Daily Active Users) with 20+ minutes spent per session

• 70% monthly retention rate

• 60% goal completion within 3 months

• User satisfaction ≥ 4.5/5

🧩 The Problem

1. Lack of Personalization – Most apps give generic plans that fail user expectations.

2. Motivation Drop-Off – Without real-time progress tracking, users lose interest.

3. Cluttered Interfaces – Complex UI makes it hard to log workouts or track milestones.

💡 The Opportunity

1. AI-driven Personalization – Tailor workouts based on goals, performance, and history.

2. Gamification – Add rewards, streaks, leaderboards to sustain motivation.

3. Simplified UI/UX – Keep it clean, intuitive, and habit-friendly.

👥 Target Audience

• Age: 18–45

• Fitness Levels: Beginners to advanced athletes

• Lifestyle: Busy professionals, students, or anyone seeking flexible workout routines

• Goals: Weight loss, muscle gain, flexibility, or general well-being

🛠 Key Features

1. Personalized Workout Plans – AI-powered routines adapted to the user.

2. Progress Tracking – Visual milestones, calories burned, streaks.

3. Social Engagement – Challenges, leaderboards, and sharing progress.

4. Real-Time Feedback – Form suggestions and workout tips.

5. Gamification – Achievement badges and rewards to keep users hooked.

🔮 Future Roadmap

• Wearable Integration – Sync with smartwatches for live performance data.

• Nutrition Plans – Personalized meal recommendations.

• Live Classes – Real-time coaching with community engagement.

✅ Conclusion

Godbod is built around a simple belief: fitness should be personal, engaging, and frictionless. By combining AI-driven personalization, gamification, and a clean interface, it has the potential to not only attract users but make fitness a daily habit.

The big lesson: In fitness, consistency is king — and great product design can make consistency effortless.

View Complete Document

🌍 Introduction

Blinkit, one of India’s fastest-growing quick-commerce players, thrives on speed and convenience. But in a category where switching costs are low and competition is fierce, customer churn is a looming challenge.

This case study looks at how Blinkit can tackle churn using North Star Metrics and OKRs, focusing on retention as the foundation for long-term growth.

🎯 Business Goal

Reduce Churn.

Churn directly impacts profitability. If Blinkit retains customers, it reduces reliance on costly acquisition campaigns and builds a predictable revenue stream.

⭐ North Star Metric

Customer Retention Rate.

Retention signals whether users see ongoing value in Blinkit’s service. Higher retention = lower acquisition costs + stable growth.

🛠 Strategy & OKRs

Objective 1: Boost Customer Engagement

• KR1: Achieve NPS ≥ 50 this quarter → stronger word-of-mouth referrals.

• KR2: Increase average orders per customer by 20% → signals satisfaction and habit.

• KR3: Implement a feedback loop with 75% survey response rate → keeps Blinkit tuned into customer needs.
  

Objective 2: Improve Onboarding Experience

• KR1: Cut onboarding time < 5 minutes → smooth first impression.

• KR2: Reduce onboarding-related support tickets by 30% → frictionless entry.

• KR3: Increase feature usage by 25% in first month → ensures customers explore value early.

🔑 Insights

1. Retention is cheaper than acquisition – engaged customers bring higher lifetime value.

2. Onboarding defines habit formation – the first few minutes determine long-term stickiness.

3. Feedback loops fuel loyalty – when customers feel heard, they stay.

🚀 Recommendations

1. Invest in early wins – surprise first-time users with rewards or free delivery.

2. Strengthen personalization – push AI-driven product recommendations to increase basket size.

3. Build community trust – leverage NPS promoters as advocates on social and referral channels.

4. Gamify repeat use – streaks, badges, or loyalty milestones to make ordering habitual.

✅ Conclusion

For Blinkit, churn is the silent killer. By adopting customer retention rate as the North Star Metric, and aligning OKRs to both engagement and onboarding, Blinkit can turn casual shoppers into loyal customers.

The big lesson: in quick-commerce, speed wins orders — but retention wins the market.

View Complete Document

Introduction

Electric two-wheelers in India are at a turning point. Ather, one of the earliest and most recognized EV players, has earned a reputation for design, technology, and performance. But despite this, EV adoption is far from universal.

In this case study, I explore user perceptions of Ather and competing brands, based on real interviews. The goal: to uncover what drives adoption, what holds people back, and how Ather can accelerate growth.

🎯 Research Goals

We set out to answer three key questions:

1. How do users perceive EVs in terms of design, performance, and technology?

2. What motivates users to buy or switch to an EV?

3. What are the pain points and barriers for Ather scooter buyers and potential buyers?

👥 User Interviews: What We Heard

🚀 The Fans (Ather Owners)

• Trusted Ather’s brand and technology.

• Impressed with speed, design, and advanced features.

• Found charging convenience manageable with home setups and growing infrastructure.

• Positive experiences with portable chargers and service support.

🛑 The Fence-Sitters (Considering EVs)

• Recognize EVs as innovative, but fear:

  • Lack of charging infrastructure

  • Safety concerns (influenced by OLA fire incidents)

  • High upfront costs (≈1.4x petrol scooters)

  • Limited variants compared to petrol options.

⛽ The Petrol Loyalists

• Choose brands like Honda Activa, Suzuki, TVS Jupiter for budget, reliability, and convenience.

• Concerned about range anxiety and battery issues.

• View petrol vehicles as more practical and trustworthy right now.

🔑 Insights

1. Infrastructure Gap – Charging stations remain the biggest barrier.

2. Awareness Gap – Many potential buyers outside metros don’t even know Ather exists.

3. Perception Gap – Safety fears and lack of EV knowledge create hesitation.

4. Product Gap – Limited models and higher costs make petrol scooters seem like safer bets.

🚀 Strategic Recommendations

1. Educate Users – Run EV awareness campaigns and community workshops to reduce misconceptions.

2. Boost Visibility – Expand physical store presence and demo test rides in high-footfall areas.

3. Highlight Safety – Lean into Ather’s zero-incident record compared to competitors.

4. Expand Product Range – Launch variants that cater to different price points and riding needs.

5. Leverage Social Proof – Promote positive customer reviews and testimonials widely.

✅ Conclusion

Ather has already captured the imagination of early adopters. But to truly scale, it must close the infrastructure, awareness, and perception gaps. By focusing on education, visibility, and affordability, Ather can lead the EV transition and win over the skeptics.

The lesson here: EV adoption isn’t just about performance — it’s about trust.

View Complete Document

Introduction

Swish, a Bengaluru-based startup, is entering the ultra-competitive 10-minute food delivery space. Its target audience? Young urban professionals battling traffic and craving convenience.
While the promise is exciting, my own onboarding experience revealed some critical cracks that could slow down adoption. In this case study, I break down Swish’s growth loops, the bottlenecks in its user experience, and my recommendations to strengthen acquisition and retention.

Step 1: The User Journey – Where It Breaks

During onboarding, I faced three major blockers that directly impact conversion and retention:

1. OTP Authentication Failures

   • Impact: Prevents new users from signing up.

   • Fix: Improve reliability and speed of OTP verification.

2. Address Search Challenges

   • Impact: Hard to locate and save delivery addresses.

   • Fix: Enhance search functionality with smarter geolocation.

3. Duplicate Address Error

   • Impact: App blocked progress even when the address wasn’t already saved.

   • Fix: Debug and eliminate duplicate detection errors.

If a new user can’t even get past onboarding smoothly, no growth loop will sustain itself.

Step 2: Identifying Growth Loops

I mapped out Swish’s growth engine into three core loops:

1. Referral Loop – Users invite friends for rewards.

2. Retention Loop – Consistent speed and quality make users come back.

3. Feedback & Improvement Loop – Customer feedback shapes product refinements.

Among these, the Referral Loop stands out as the fastest way to acquire users at scale.

Step 3: Proposed Improvements

Referral Loop Enhancements

1. Fix Onboarding Issues First – A smooth entry is non-negotiable.

2. One-Click Referral Sharing – Simple WhatsApp/social share for virality.

3. Dynamic Rewards System – Tiered incentives (e.g., free delivery credits, vouchers).

4. Gamified Referrals – Leaderboards, milestones, and badges to keep users engaged.

Step 4: Prioritization with RICE

Using the RICE framework, I scored initiatives based on Reach, Impact, Confidence, and Effort.

• Highest Priority: Simplified referral sharing + onboarding bug fixes.

• Medium Priority: Dynamic rewards.

• Lower Priority: Gamification (fun, but long-term ROI).

Step 5: Implementation Roadmap

• Immediate (Month 1) – Fix OTP/address bugs + launch one-click referral sharing.

• Short-Term (Quarter 1) – Deploy dynamic rewards to build momentum.

• Long-Term (Next Year) – Expand gamification, refine referral system, and strengthen delivery backend.

Conclusion

Swish is in a promising market, but execution is everything. By fixing onboarding friction and doubling down on a referral-driven growth loop, Swish can accelerate adoption while building a loyal user base.
The lesson here: growth loops only work if the core experience is seamless. Start with reliability, then amplify virality.

View Complete Document

Setting the Context

Urban Company has become the go-to platform for home services in India. From grooming to home repairs, its value proposition is convenience.

But recently, something alarming happened: its App Store rating in India dropped from 4.7 to 4.1 stars in just six months — specifically on iPhone devices.

As part of a product management assignment, I set out to analyze why this happened and uncover actionable recommendations.

My Process

I approached the problem through a Root Cause Analysis (RCA):

1. Analyzing User Reviews

   • Studied 100 one-star reviews.

   • Grouped issues into buckets: service quality, support, delays, interface.

2. External Factors

   • Overpromising ad campaigns.

   • Poor partner service quality.

3. Internal Factors

   • Buggy interface issues on iPhone.

   • Customer support relying on automated responses.

   • Unclear communication about delays.

4. User Journey

   • Booking and app navigation were fine.

   • The breakdown came after booking, during service delivery.

Key Insights

The numbers made it clear:

• 48% complaints: poor service quality.

• 27% complaints: delays & rescheduling without consent.

• 22% complaints: unresolved issues via customer support.

• 3% complaints: buggy interface.

The drop wasn’t about the app flow. It was about trust, reliability, and execution.

Recommendations

To reverse the trend, here’s what I’d propose:

1. Stricter Partner Quality Audits – Continuous training and certification for service professionals.

2. Revamped Customer Support – From robotic replies to personalized, accountable service.

3. Proactive Communication – Notify users early about delays or reschedules.

4. Expectation Management – Align marketing promises with actual service quality.

Takeaway

This exercise reinforced a critical truth:

In service marketplaces, the real product is the experience.

No matter how well the app is designed, customers will rate it poorly if the on-ground service doesn’t live up to the promise.

As product managers, it’s our responsibility to think beyond the app — ensuring consistency between what’s marketed, what’s promised, and what’s delivered.

⚡ That’s the story of my Urban Company assignment. In future posts, I’ll dive deeper into retention strategies, product case studies, and growth experiments.

View Complete Document

Why This Project?

One of the most interesting challenges I worked on during my Product Management Launchpad capstone was centered around MPL, India’s mobile gaming platform.

The problem statement was simple but bold:
👉 Reduce MPL’s 30-day churn by 25% over the next two quarters.

Most new users were dropping off after just 2–3 sessions, never forming a habit of returning. For me, this was the perfect opportunity to apply product thinking, user research, and behavioral insights to a real-world challenge.

My Approach

I broke down my process into three key steps:

1. Secondary Research

   • Studied MPL’s current app journey.

   • Compared it with competitors like Dream11, My11Circle, and WinZO.

   • Looked at proven mechanics from apps like Duolingo (streaks) and Candy Crush (loss recovery).

2. Primary Research

   • Ran a survey with 41 users.

   • Conducted 3 in-depth interviews to dig into motivations and frustrations.

3. Behavioral Insights

   • Users love daily streaks but MPL didn’t have a strong system.

   • Losses without consolation frustrated casual players.

   • Many installed the app only during IPL season and uninstalled right after.

   • Social play was important but hidden.

What I Discovered

The biggest blockers to retention weren’t technical bugs — they were experience gaps:

• Onboarding was too complex (fantasy cricket rules overwhelm new players).

• Losing meant no feedback or encouragement — players simply left.

• No daily progress system kept users coming back.

• Social gameplay wasn’t integrated enough to create stickiness.

The Solutions I Proposed

To address these, I designed a set of solutions focused on habit formation and social engagement:

• Guided Onboarding: Interactive tutorials for new fantasy players.

• Loss Recovery: “Try Again” tokens + tips after close losses.

• Streak Rewards: A daily streak system with escalating incentives.

• Friend Leagues: Easy WhatsApp invites and private leaderboards.

• Real-Time Match Dashboard: Live cricket updates inside the app.

Here’s a look at the redesigned onboarding and streak features I created in Figma:

View Prototype

The Roadmap

Using the RICE framework, I prioritized features in two phases:

• Phase 1: Enhanced onboarding, streak rewards, live match data.

• Phase 2: Friend leagues, loss recovery mechanics, and gamified challenges.

The north star metric was 30-day retention, with success measured by onboarding completion, session frequency, and time spent in-app.

For a more detailed walkthrough of my process and findings, here’s my full presentation deck :

View Slides

What I Learned

This project reminded me that:

• Retention is emotional — frustration, delight, and habit cues matter more than features.

• Competition inspires innovation — looking at Dream11 or Duolingo sparked fresh ideas.

• Product management is about trade-offs — you can’t do everything, so prioritization is key.

View Full Video Presentation

Wrapping Up

Working on MPL’s churn problem as my capstone was both challenging and rewarding. It pushed me to apply frameworks, but also to think about the human side of products — why people stay, why they leave, and what makes them come back.

I’ll be sharing more of these case studies here on Substack, along with insights from my own work in product management, design, and growth.

I’ve decided to use this space as a journal of sorts — a mix of reflections, case studies, and lessons from my work in product management, design, and digital growth. If you’re someone who enjoys unpacking how products are built and scaled, I think you’ll feel at home here.

My Backstory

I started out in Computer Science Engineering, fascinated by technology’s ability to solve problems. That curiosity quickly grew into entrepreneurship — I founded a cybersecurity company, conducted training sessions, and even worked alongside government departments on digital forensics projects.

A few years later, I launched Viral Inbound, a digital agency that went on to deliver 200+ websites and marketing campaigns. Running an agency gave me a crash course in leadership, project delivery, and the importance of understanding clients deeply.

My journey then took me into SaaS with Nexweave, where I worked on creating interactive video experiences that personalized how businesses communicated with their customers.

Now, I’m a Digital Product Manager at Selligion Technologies, where I oversee product strategy and manage e-commerce growth across Shopify and Amazon.

Why I Chose Product Management

For me, product management sits at the intersection of business, design, and technology. It’s not just about building features — it’s about understanding what people need, what frustrates them, and how we can create solutions that genuinely make a difference.

What I love most is that it’s both creative and analytical — one day I’m working on user flows, the next I’m analyzing retention metrics, and the day after, I’m aligning stakeholders around a roadmap.

What I’ll Be Sharing Here

This newsletter is where I’ll share:

• Case studies — like my MPL churn project or Urban Company root cause analysis.

• Lessons learned — from delivering 200+ projects and collaborating with startups.

• Ideas and frameworks — that connect growth, product strategy, and design.

Why I’m Writing

Writing helps me slow down and process what I’ve learned. More importantly, it’s a way to connect with others who are equally curious about building better products.

If you’re interested in product management, digital growth, and design thinking, consider subscribing — I’d love to share this journey with you 🚀