From messy AI outputs to clean, editable diagrams. I built this to solve the friction of "re-drawing" charts generated by LLMs.
Shipped usage-based pricing and tracked retention + MRR weekly.
I turned Bhava from a promising hack into a trust-first workspace. The AI shows its work, asks for context, and outputs clean, editable diagrams.
Achieved by pairing prompt presets, progress states, and failure logging with weekly eval loops.
Removed free mode for usage-based billing. We tracked retention and MRR weekly to ensure pricing matched utility.
From hacked-together prototypes to a polished product.
The first build was a blank prompt. Users froze. We added suggested prompts, live thinking states, and "show your work" logging.
Mapped credits per plan to cap abuse. This informed usage-based billing that filtered low-intent signups.
Built an onboarding path that asks for system type, generates a diagram with context. Lifted activation to 60%.
"Free Forever" tier attracted spam. Switching to "Paid Only" improved quality.
"Every week, I watched designers redraw the same diagram—like déjà vu with arrows."
By day, I'm deep in B2B dashboards. By night, I watch my team waste hours redrawing the same system diagram in Figma, Draw.io, Excalidraw, and Miro.
So I started building Bhava—an AI tool that generates diagrams instantly.
What Bhava looked like before vs after
We're 4 weeks post-launch. I handle design, product, and evals.
The activation problem: In early tests, only 38% of users created their first diagram. The other 62% bounced without trying.
"AI doesn't need to be perfect—it just needs to show it's trying."
Our bet: Build on top of Draw.io (largest user base) and make AI feel reliable, not random.
Every design decision mapped back to a trust framework for AI research
Can the AI actually do the task?
Does it feel like it's helping me?
Is it honest about what it can and can't do?
Does it work consistently?
Before redesigning anything, I spent 2 weeks analyzing user behavior—watching session recordings, tracking prompts, and interviewing people who churned.
Deep dives with designers and PMs who churned within 48 hours
Analyzed recordings to identify drop-off patterns and friction points
Live observation of first-time users attempting diagram creation
Lands on homepage via Product Hunt or Twitter
Curious, skeptical
Creates account, sees empty editor
Confused, uncertain
Stares at text box with no guidance
Paralyzed, frustrated
Hits generate, sees spinner for 7 seconds
Anxious, doubting
Gets poor result or gives up
Disappointed, churns
Users landed on an empty editor with no guidance, no examples. They froze.
"Intelligent" vs "Basic" results varied wildly. Trust eroded fast.
3–8 seconds of spinner. No updates. Pure anxiety.
Only 15% exported their first diagram. The happy path was invisible.
I don't know what to type, so I just close the tab.
Why does intelligent mode give me different results?
It's just spinning... is it even working?
Where do I export this diagram?
"The activation gap is usually a clarity gap—not a capability gap."
Understanding how Bhava stacks up against existing diagramming solutions
AI-first approach cuts diagram creation time from 30 minutes to 30 seconds
Shows reasoning, progress states, and allows editing—not a black box
Export to editable Draw.io, PNG, SVG—works with existing tools
Presets and examples eliminate blank canvas paralysis
Each redesign tackled a specific trust or activation gap. Here's what worked.
First we fixed clarity, then trust, then monetisation.
Getting users to understand what to do and how to start
Problem: Vague CTAs meant visitors signed up without understanding what to type.
Solution: Elevated a giant prompt box with example chips and a mini walkthrough so users preview the experience before creating an account.
Problem: New users froze on an empty chat and churned without generating anything.
Solution: Added diagram-type cards, contextual hints, and a three-step progress indicator that nudges people into action.
"Clarity unlocks activation. But trust keeps users coming back."
Building reliability into the product experience
Problem: The legacy "Basic" mode produced low-quality diagrams that tanked perceived reliability.
Solution: Sunset the free mode, offered one premium try, and introduced usage-gated access to keep output quality consistent.
"Monetisation isn't just about pricing—it's about signalling reliability."
Aligning value with sustainable monetisation
Problem: Unlimited $10/month plans were unprofitable and encouraged abuse.
Solution: Swapped to a $10 base plan with transparent credit packs and real-time usage tracking.
Problem: Pricing changes created confusion-users couldn't tell where credits went.
Solution: Built an always-available tutorial and a usage dashboard detailing credits, modes, and expiry.
Problem: Diagram quality varied by type and we lacked clarity on failure patterns.
Solution: Logged ~100 failed diagrams, clustered errors, and routed high-volume types through specialized sub-agents.
A snapshot of where things stand after the first month of shipping.
Guided onboarding was the key driver for new signups.
Interactive prompt demos helped visitors understand the value faster.
Prompt caching reduced median generation time significantly.
Per-diagram cost optimization via Claude API caching.
MRR from 30 initial customers on usage-based pricing.
Users returning to create another diagram within the first week.
"I learned that AI trust is built in microseconds—even a 7-second latency feels fine only when users see progress."
One broken diagram erodes more trust than five perfect ones build. Quality consistency matters more than feature breadth in AI products.
Charging more increased trust because it signaled we stood behind the quality. Usage-based pricing tells users the output is valuable enough to meter.
Transparency about what's happening builds confidence. Generation time feels different when users see specific steps like "Analyzing structure..."
Jump from 38% to 60% activation was about making the product clearer. When users see examples, they immediately understand the possibility space.
Four weeks is too early to claim PMF.
Need day-30 and day-60 retention metrics.
Export rate stuck at 15%—my next focus.
Eval rubric is still subjective (v1).
Metrics estimated; refining instrumentation.
Pricing tests ongoing; willingness forming.