Build in Public

Building Pactify: Updates & Learnings

I'm building Pactify in public—sharing what I'm learning, what's confusing, and what's not working. Not marketing, just honest progress updates.

Try What We've Built So Far

Install Extension
2026-02-24

v1.1.6 & v1.1.7 - Security First, Then Getting Out of Your Way

SecurityUX ImprovementBuild in Public

Hey everyone,

Two releases in quick succession this time. Here's what changed and why.

A Note on How Fast We Move

Pactify ships updates almost every week. That's intentional—we'd rather fix something in three days than wait three weeks. But it also means if you're running an older version, you're missing fixes and improvements that are already live.

Chrome should auto-update extensions in the background, but it doesn't always happen immediately. If anything feels off, check your extensions page and make sure you're on the latest version. It genuinely makes a difference.

v1.1.6: Strengthening the Foundation

After the security incident we shared in v1.1.5, I wasn't satisfied with just stabilizing things. I went deeper.

The core of this release was building real-human verification into our system. The attack we experienced wasn't just about traffic—it was bots distorting signals, triggering false usage, and making it harder for real users to be seen. So we added verification checkpoints across the critical flow to distinguish genuine users from automated actors. When a bot can't get through, it stops doing damage.

The practical effect: less noise in our systems, fairer resource allocation for actual users, and better signal for us to improve the product based on real behavior rather than bot artifacts.

I also expanded the free trial to 30 conversations. If you've been on the fence, that's now a proper amount of time to see whether Pactify actually fits your workflow before committing.

One more thing—and I'll be direct about it: our Chrome Web Store visibility still hasn't fully recovered from the attack. The best thing you can do, if Pactify has been useful to you, is leave an honest review. Not for us—for the next person who needs this tool and is trying to figure out if it's worth trusting. Reviews from real users are the most effective counter to the kind of manipulation we went through. If you haven't yet, it would mean a lot.

v1.1.7: Reducing Friction

One thing I've noticed in user feedback: small interruptions break flow more than people account for. A dialog box that pops up mid-session—even a helpful one—can cost you three seconds of attention and derail a thought.

We removed the trial timeout dialog that used to appear while you were in the middle of a conversation. It was well-intentioned, but it was interrupting you at the wrong moment. The information it carried is still accessible, just no longer shoved in your face unprompted.

We also added a quiet version check. When a new version of Pactify is available, the extension will let you know—no pressure, just a nudge. Given our release cadence, staying current is worth it.

The Short Version

Security is tighter. Interruptions are fewer. And if you're not on v1.1.7 yet, it's worth the update.

2026-02-19

v1.1.5 - A Lesson We Needed to Learn

InfrastructureSecurityBuild in PublicTransparency

Hey everyone,

I want to be honest with you about what's been happening behind the scenes these past two weeks.

Pactify Was Under Attack

We were hit by a coordinated negative SEO campaign. If you've tried searching for Pactify on the Chrome Web Store recently and couldn't find us—that's why. The attack was deliberate and precisely targeted, and the impact was significant.

We spent a large portion of these two weeks investigating, diagnosing, and upgrading our security defenses. The situation has mostly stabilized, but our visibility in Chrome Web Store search is still recovering. Daily traffic dropped sharply, and I won't pretend that didn't hurt.

Why This Hit Harder Than Expected

Here's the honest truth: Pactify is not a mass-market product. We serve a specific group of people—those who use AI tools intensively and want to connect them with Notion. That's a small, focused audience. We're not running aggressive freemium growth or viral referral loops. We grow slowly, by being genuinely useful.

Which means when someone targets that foundation, it matters.

What stings more is that I should have caught this earlier. I had neglected some of the infrastructure fundamentals—monitoring, anomaly detection, platform health checks. By the time I sat down and did a serious analysis, a meaningful amount of damage had already been done. A well-designed attack was already weeks old before I recognized it for what it was.

What Comes Next

We're doing an internal review of our roadmap. This situation forced me to confront something I'd been deferring: we need a more stable foundation before we keep adding features on top of it.

That means three things going forward:

First, service stability and infrastructure will get the attention they deserve. Not glamorous work, but necessary.

Second, we're diversifying how we reach people. Relying too heavily on one discovery channel made us fragile. SEO, content, community—these need to work together.

Third, new feature development will slow down—temporarily. I know that's frustrating to hear if you've been waiting on something. But shipping features into an unstable foundation isn't responsible to you or to the product.

A Note to Everyone Still Here

This was a lesson. I'm not going to pretend otherwise or dress it up as a strategic pivot. We took damage, we made mistakes, and we're fixing them.

But I also know what Pactify is, what it's for, and who it's for. That hasn't changed. I'm asking for your patience while we make the underlying service worthy of the trust you've placed in it.

We'll come back stronger. Thank you for staying.

2026-02-13

v1.1.4 - The Details That Matter

Bug FixUser FeedbackSync Improvements

Hey everyone,

This week was all about fixing the small things that add up to big frustrations.

The Gemini Title Bug

Some of you noticed conversations syncing to Notion as "Untitled" when they came from Gemini. I finally tracked it down—Gemini changed their DOM structure, and my title scraper couldn't find the conversation name anymore. Rewrote the logic to handle their new layout. Your conversation titles should now show up correctly in Notion.

Understanding Why You Leave

I added a simple feedback form that shows up when you uninstall the extension. Just two questions: why you uninstalled, and what would bring you back. I'm not tracking you or selling data—I just need to know what's not working. If you ever decide to leave, your honest feedback would help more than you think.

The Title & URL Sync You Asked For

This one's been requested a few times: you rename conversations in ChatGPT to stay organized. In Claude, you move conversations between Projects. But Notion never knew about these changes—so your database got out of sync with your actual chats.

Now the sync is smarter. When you update a conversation, it checks if the title or URL changed (like when moved to a different project). If they did, it updates your Notion database automatically. One less thing you have to manually fix.

That's it for this week. Small improvements, but the kind that matter when you're using this daily.

2026-02-07

v1.1.2 - A Milestone: Going All-In on Notion Integration

Notion IntegrationPublic LaunchUI RedesignProduct Milestone
  • Minimalist UI Overhaul: Spent the last two weeks redesigning everything—both the Chrome extension and the website. Generous negative space, crystal-clear hierarchy, visual consistency across all touchpoints. It sounds simple, but completely overhauling UI in a short time while maintaining functionality? Not easy. But it matters. Your conversations deserve a clean, focused interface.
  • Frictionless Notion Sync Goes Public: Card-free trial is now live. After months of beta testing with your feedback (thank you), Notion sync is officially out of beta and production-ready. No waitlist, no beta invites—registered users can start syncing immediately with 30 free conversations.
  • What This Means: For existing Starter members and beta testers, nothing changes—you keep your unlimited access. For new free users, you get 30 conversations to try the full auto-sync experience. This has been the most requested feature, and I'm relieved it's finally stable enough to open publicly.
  • What's Next: This is a milestone, but also a starting point. We're pivoting fully to Notion Integration—not just one-way sync, but true bidirectional AI ↔ Notion workflows. I have ideas (contextual injection, knowledge graph connections, adaptive syncing), but I need to validate them with real usage first. Expect more experiments in the coming weeks.
2026-01-31

v1.1.1 - This Week's Bug Fixes & Small Wins

Bug FixPerformanceWeekly Iteration
  • Spent a day figuring out why images weren't exporting. Turns out each AI platform wraps image blobs differently in their DOM. Wrote custom adaptors for Gemini and ChatGPT — your DALL-E creations should finally survive the export.
  • Claude's interface update broke our button positioning (it started overlapping their 'Share' menu). The joy of building on moving platforms. Adjusted our CSS injection to play nice with their new layout.
  • The Sidepanel started feeling heavy on my 500+ message conversations. Profiling showed too many unnecessary re-renders. Switched to virtualized lists — scrolling should be buttery smooth now.
  • Felt the navigation header was stealing too much vertical space from the actual content. Compacted the design while making click targets larger. Still iterating on how to maximize reading area in such a narrow window.
2026-01-27

Pactify v1.1.0: Finally, a Real "Second Brain Portal"

SidepanelGlobal SearchNotion IntegrationBuild in PublicProduct Update

I built the v1.0 Popup following standard tutorials. Big mistake. In daily use, I found myself hating it—it covered my content and felt like an interruption, not a helper.

The deeper issue I faced: My "Second Brain" (Notion) and my "AI Brain" (ChatGPT/Claude) were completely siloed. I was a human copy-paste bridge, wasting hours moving context back and forth.

So I made a scary decision: stop all marketing, pause new features, and rewrite everything for the Chrome SidePanel API. v1.1.0 isn't just an update; it's the workflow I actually wanted.

What's New

  • Architecture Shift: Scrapped the React Popup implementation entirely. Moved to Chrome SidePanel API + Manifest V3. It was a headache to handle the different content security policies, but the persistent workspace is worth it.
  • The "Lost Conversation" Fix: I kept forgetting which AI I used for specific tasks. Built a unified local index that aggregates ChatGPT, Claude, and Gemini history into one search bar.
  • Context Injection: Solving the "Alt-Tab" fatigue. You can now select Notion pages and inject them into your current AI chat context without leaving the tab (Experimental).
  • Waitlist Reality: Notion's API limits are stricter than I realized. Had to prioritize stability for paid users over massive free growth.

What I Learned

  • Context switching is the enemy. If a tool makes you switch windows, it costs more cognitive load than it saves.
  • Rewriting is terrifying. We went "dark" for 3 weeks to fix the architecture. Usage dropped, but the foundation is finally solid.
  • Sustainability vs. Growth. Notion's API quotas forced a hard choice: restrict deep integration to paid users or crash the service. I chose stability over vanity user counts.
2026-01-07

Notion Auto Sync Beta: The Bridge Between AI and Your Second Brain

Notion IntegrationAuto SyncBuild in PublicPKMProduct Update

I exported an AI conversation to Google Docs last month. Perfect formatting, clean structure.

Then I closed the tab and forgot it existed. Three weeks later, I repeated the same conversation.

The export worked. The workflow didn't. That's when I realized: I'd been solving the wrong problem.

AI conversations need to live where your thinking happens—not in export folders. That's why I built Notion Auto Sync.

What's New

  • Notion Auto Sync (Beta): Automatically sync ChatGPT, Claude, and Gemini conversations to your Notion workspace
  • Full Conversation Content: Properly formatted with tables, code blocks, and LaTeX formulas preserved
  • Smart Metadata: Auto-includes conversation title, date, AI platform, and direct link back to original
  • Multi-Platform Support: Works seamlessly across ChatGPT, Claude, and Gemini
  • Auto-Generated Tags: Coming soon—AI-powered conversation summarization and tagging
  • Chrome Sidepanel: Quick access to Pactify features directly in your browser (in development)

What I Learned

  • Export problem ≠ workflow problem. The real gap is "my conversations are disconnected from my knowledge system"
  • Different people organize differently: by project, by date, by topic. One-size-fits-all integration doesn't work
  • My workflow isn't your workflow. Need real user scenarios before building Phase 2 features
  • The bridge isn't just about moving text—it's about actual workflow integration between AI and your "thinking place"
  • Building Phase 1 (one-way sync) proves the concept. But the bigger vision is two-way interaction: Notion → AI and AI → Notion
  • This bet might be wrong. Maybe people don't want deep integrations. That's why I'm building in public—to find out
2025-12-18

One-Click Notion Export: Connect Your AI Conversations to Your Second Brain

Notion IntegrationAI WorkflowProductivityProduct Update

Remember spending 5-10 minutes copy-pasting ChatGPT conversations into Notion? Losing formatting, manually adding metadata, fixing broken tables?

After hearing this frustration from dozens of researchers and consultants, I built what should have existed from day one: one-click Notion export.

Now your AI conversations land directly in your Notion workspace—formatted, organized, and ready to use.

No more workflow friction. No more context switching. Just click, export, done.

What's New

  • One-Click Notion Export: Export ChatGPT, Claude, and Gemini conversations directly to Notion with a single click
  • Multi-Platform Support: Works seamlessly across all three major AI platforms with consistent quality
  • Format Preservation: Tables, code blocks, LaTeX formulas, and formatting maintained perfectly
  • Smart Metadata: Auto-includes conversation date, platform, model info, and token count
  • Instant Speed: Complete export in under 5 seconds—no waiting, no processing delays
  • Flexible Organization: Choose target database, add custom tags, customize export location

What I Learned

  • Export friction isn't a minor inconvenience—it's a workflow killer that makes people abandon valuable AI conversations
  • The real problem isn't "saving AI chats" but "integrating them into existing workflows where actual work happens"
  • High-frequency users (5+ conversations/day) lose 30-60 min/week on manual export—time that should go to actual thinking
  • Different users need different organization: some by project, some by date, some by topic. One-size-fits-all doesn't work
  • Quality matters more than speed: a 5-second export that preserves formatting beats a 1-second export that loses context
2025-12-09

Beyond Export: Building Workflow Connections

Workflow ConnectionsProduct UpdateBuild in PublicUser Research

After talking with dozens of researchers, I realized something: just exporting AI conversations isn't enough.

You use Zotero, Obsidian, Overleaf, Google Docs. Your workflow is fragmented by design.

What people actually need is better connections between the tools they already use.

This led us to completely rethink what Pactify should be—and what we built next.

What's New

  • Connect to Google Docs: AI conversations land directly in your Google Drive where your drafts live
  • Connect to Obsidian (Markdown): Drop conversations into your vault, link to existing notes, build knowledge graphs
  • LaTeX Copy on All Platforms: ChatGPT, Gemini, Claude—all standardized with one-click copy
  • Light/Dark Mode Toggle: Switch seamlessly without reloading, perfect for late-night work
  • Freemium Upgrade: 30 conversions/month (6x), 1MB per file (10x), all templates available

What I Learned

  • Format problem ≠ workflow problem. The real question is "where does this go in my system?"
  • You shouldn't need to hit limits on day 3 just to test if something fits your workflow
  • I don't have all the answers—metadata preservation, bi-directional links, Zotero integration? Need your feedback
  • Quality feedback beats vanity metrics. Would rather grow slowly with people who actually need this
  • Building in public means admitting uncertainty and asking for help, not showing off what's "done"
2025-11-25

LaTeX Support, Multi-Platform, and Quality Improvements

LaTeXMulti-PlatformQualityTesting

After two weeks of 14-hour days, I shipped what users asked for.

LaTeX copying, Claude and Gemini support, Markdown exports, better formatting—real problems from real researchers.

But here's what keeps me up: did I build what people actually need, or what I think they need?

Am I solving core problems, or just enabling workarounds?

What's New

  • LaTeX Copy Button: Click "Copy LaTeX" on ChatGPT formulas, paste directly into your editor
  • Claude & Gemini Support: Custom handling for each platform's quirks to maintain export quality
  • Markdown Export: Academic-standard formatting with proper LaTeX delimiters and citation formats
  • LaTeX Formula Fixes: ChatGPT's non-standard delimiters ([...] vs $$) now render correctly in Word
  • Styling Improvements: Tables, headers, lists spacing and alignment match academic standards
  • Testing Obsession: 20 academic scenarios × 3 platforms × 100+ test cycles = 97%+ accuracy

What I Learned

  • Every change breaks something else—fixing LaTeX broke table spacing, improving Claude changed ChatGPT citations
  • Multi-platform support might be enabling workarounds rather than solving core problems
  • "Ready to use" means different things to different people—need more clarity on what quality bar matters
  • Infrastructure is invisible when it works. You only notice when it breaks.
  • The goal isn't just "export"—it's "ready to use without manual cleanup"
2025-11-17

Why I Built This (And Why I'm Not Sure What to Build Next)

Build in PublicProblem DiscoveryUser Research

After 3 years building AI products that didn't work, I finally admitted something:

I don't know what people actually need.

So I'm building Pactify in public—starting with my own problem (2-4 hours/week wasted organizing AI conversations).

But I'm openly asking: is this even a real problem worth solving?

What's New

  • Built minimum viable export: ChatGPT conversations → professional Word documents
  • Academic document standards: LaTeX formulas, tables, proper formatting
  • Enterprise-grade security: Temporary storage, auto-deletion, no third-party exposure
  • Chrome Extension: One-click export directly from ChatGPT interface
  • Started with my own pain point: AI saves 30 minutes, but organizing takes 90 minutes

What I Learned

  • Closed-door development never works—spent 3 years building things nobody used
  • Everyone's AI use cases are completely different: research, creative, legal, technical
  • The gap isn't AI capability—it's the bridge from "AI generated" to "actually usable"
  • Most export tools just save as plain text. When there are formulas, tables, code—everything breaks
  • Need to hear real user scenarios before building more features, not assume I know what they want

Help Me Build Better

I'm not trying to sell you something—I'm trying to understand if this direction makes sense. Your real-world scenarios and feedback matter more than anything else.