SIDE BUILD · CONSUMER PRODUCT · 0→1
Speed-reading tools already existed. The opportunity I saw was narrower: Indian professionals whose jobs require them to process newsletters, reports, research papers and long-form content faster than their reading backlog grows.
I built SpeedLens independently from product framing through implementation, payments, Chrome Web Store launch and iteration. Claude Code accelerated development, but the product decisions — target segment, positioning, reading modes, pricing, payment rails and what not to build — remained the core of the work.
140+
Chrome Web Store installs
0
Paid acquisition
2 modes
Flash + Bionic
v1.1.0
Current release
₹799/year
Pro Yearly
140+ installs were acquired without paid marketing, providing early evidence of organic interest but not yet proof of product-market fit.
Most speed-reading products position themselves around a generic promise: “Read faster.” That creates a broad product with no clear user or urgency.
Chosen
Heavy readersThe problem is repeated and work-related.
Knowledge workersStronger willingness to pay for productivity than casual readers.
IndiaGlobal tools often use USD pricing and international-card payment flows.
Deprioritised
High price sensitivity and strong free-alternative behaviour.
Lower pain frequency.
Existing competition already serves the broad global use case.
Narrowing the segment made the next decisions easier.
The positioning shifted from a generic speed promise to a concrete job: clearing the reading backlog that heavy-reading professionals accumulate.
Initial framing
Read faster
Final framing
Clear your reading backlog
Words appear one at a time using RSVP so the eyes remain fixed. Optimised for high-volume reading where throughput is the priority.
Typical use cases
The first approximately 40% of each word is emphasised to create visual anchors. Designed for retention and comprehension in complex material.
Typical use cases
Three entry points
Select text on a webpage and launch SpeedLens directly.
Use Mozilla Readability to extract the primary article content.
Copy text from PDFs, email or documents and read it inside SpeedLens.
Different reading jobs require different modes. The product should adapt to the content rather than force one method everywhere.
The point was not technical minimalism for its own sake. The architecture reduced moving parts and was sufficient for the actual product requirements.
No React. No bundler. No webpack.
Extension
Content extraction
Backend
Payments
Website
Complete launch surface
Reliability progression
As the product matured, the operating standard changed from “does the feature work?” to “Can I prevent the same failure from returning?”
Prototype stage
Manual Chrome-extension testing
After real lifecycle failures emerged
Added automated regression coverage
Current
Playwright E2E tests load the real unpacked extension in Chromium and exercise critical flows
10 free reads per day
This restricted the core product habit too early.
Unlimited free reading during GTM
Optimise for usage, trust and learning before aggressively restricting the core experience.
Free core
Pro utility
Monetise depth of utility, not the core habit.
Instead of charging users simply for reading more, SpeedLens keeps the core reading loop unrestricted and monetises convenience and premium functionality around it.
Free
₹0
Unlimited reads during GTM
Pro Monthly
₹99/month
Pro Yearly
₹799/year
Lifetime
₹1,999 one-time
Global competitors often price in USD and rely on international payment methods. SpeedLens uses Razorpay, enabling payment methods that Indian professionals already use daily.
The pricing strategy and payment architecture came from the segment choice, not from copying global SaaS conventions.
India-first payment infrastructure creates a locally better conversion experience and increases friction for global competitors that have not optimised for domestic payment behaviour.
SpeedLens has been iterated across five releases. Each version addressed a specific gap that real usage revealed.
v1.0.1
Initial complete product
v1.0.2
Improved control and reliability
v1.0.3
Improved document fidelity
v1.0.4
Remove hard usage limits
v1.1.0
Define Free / Pro + improve reliability
Key learning
Bionic mode initially flattened documents into a wall of words.
The fix was not cosmetic. Research papers and reports depend on paragraphs, headings and table structure to retain meaning. The product was updated to preserve document structure rather than treating every input as a sequence of tokens.
Feature works technically ≠ feature solves the actual reading job.
Monetisation learning
Restrict usage to create upgrade pressure.
At this stage of the product, usage and learning mattered more than forcing conversion. The paywall moved from frequency to functionality — keeping the core reading loop unlimited and monetising higher-intent utility instead.
The paywall moved from frequency to functionality.
140+
Chrome Web Store installs
The product reached 140+ Chrome Web Store installs with zero paid acquisition — an early organic adoption signal, not proof of product-market fit.
SpeedLens was developed end to end using Claude Code across multiple sessions. Claude Code accelerated both the initial build and later product iteration — including monetisation changes, debugging, regression-test implementation, documentation and release updates. The product decisions remained mine.
Project context approach
A project-level CLAUDE.md maintained architecture, coding standards, feature state, testing expectations, payment configuration context and backlog conventions. This allowed development sessions to resume with project context intact.
The experiment was not
Can AI invent a product?
The experiment was
How much further can a PM with an engineering background extend their execution range by using agentic coding tools deliberately?
AI helped build the product
AI helped build the product.
Ashish still made the decisions
01
Choosing Indian heavy-reading professionals simplified positioning, pricing, payments and scope.
02
Shipping the extension, onboarding, payments, website and distribution loop created far more learning than continuing to ideate.
03
The first monetisation model restricted usage itself. I later removed that constraint and moved the paywall to higher-intent utility, keeping the core reading habit unlimited. Monetise depth of utility, not the core habit.
04
Agentic coding made it possible to operate across product, engineering, debugging and shipping independently — but only because product judgment determined what should be built.
SpeedLens taught me that AI can dramatically compress the distance between product judgment and working software — but it cannot make the judgment for you.
What's next
The next phase is less about adding more reading modes and more about reducing friction across control, modality and where the reading experience lives.
When Flash Mode is paused, show the full article with the current reading position visible.
Why it matters
Flash reading optimises for focus, but pausing should not force the user to lose context.
Product direction
Improve control without breaking the focused reading flow.
Add text-to-speech synced with the visual reading experience.
Planned as Pro-only.
Why it matters
Some reading jobs are better served by switching between visual and audio consumption.
Product direction
Extend SpeedLens from a purely visual reader into a multimodal consumption tool.
Apply Bionic formatting directly on the source webpage from the text-selection flow.
Why it matters
Today, users move into the SpeedLens reading experience. Inline Bionic would bring the reading mode closer to the original content instead.
Product direction
Reduce the distance between discovering content and consuming it.
Back to AI Lab