# 2026-02-27 Session Log

## NEW CLIENT: HUD (hud.io)
- **What:** Runtime Code Sensor for AI coding agents. One line install, zero config. Category: "Runtime Intelligence"
- **Founders:** Roee Adler, May Walter, Shai Wininger (4 acquisitions, 3 IPOs)
- **Customers:** Monday, Lemonade, ZoomInfo, Drata, Cyera, Guesty, Axonius, Appsflyer, Tipalti, Windward, Au10tix, Guardz
- **CMO spoke to Assaf** — this deck goes to founders, free of charge, trying to win the business
- **Channel:** #hud (1476558140551069726)

## BRAND STRATEGY WORK DONE
- Full 8-section CE Due Process brand strategy plan (v3, Anton graded A)
- Built landing page at https://www.curiousendeavor.com/hud/
- All 8 sections populated with three-lens format (Their Info / Our Insight / Why Ours Is Better)

## KEY STRATEGIC DECISIONS
- **Category:** "Runtime Intelligence" — NOT observability. Category creation play, not disruption.
- **Crisis framing:** "The AI coding revolution isn't evolution — it's a production awareness crisis"
- **Brand thesis:** "Observability watched. Runtime Intelligence knows."
- **Visual recommendation:** "Purposeful absence" — radical simplicity direction (Anton picked Direction 3)
- **Human cost angle:** Assaf wants to add the emotional layer — 2am war rooms, missed dinners, cold coffee. The value Hud creates in engineers' LIVES, not just their workflows. This is the missing strategic layer.

## PAGE REFINEMENTS DONE
- CE red block → "curious endeavor" lowercase in white-on-red frame
- Killed redundant content (editorial + three-lens saying same thing in Context)
- Sticky callout sidebar in Context section
- Audience table killed (redundant with persona panels), entry points added as labels
- Orphan word fix in Landscape title
- Competitor logos added (favicons for 13 companies)
- Norms Hud Rejects → individual cards with norm + rejection
- Messaging table → 6 real copy cards (Ogilvy wrote, Anton critiqued: Slack alert A+, Demo A)
- Section 7 rebuilt: category-creating brands study (Stripe, Notion, Linear, Figma, Slack, Superhuman)
- Section 8 transformed into "Where We Go From Here" — 3-phase roadmap + CTA
- Nav spacing fixes (JS was overriding CSS `top` values)

## STILL IN PROGRESS
- **Brand image quality:** Slack, Linear, Figma, Superhuman, Datadog screenshots have cookie banners/quality issues. Multiple options pulled to /hud/visuals/options/ — Assaf needs to review and pick winners.
- **Section 7 needs:** Better images chosen, human cost visual layer, grounding "purposeful absence" with actual mockups
- **Full-page Anton review** not done yet — he only critiqued Section 7
- **dev.to article:** https://dev.to/aws-builders/hudio-doesnt-replace-observability-it-replaces-guessing-13m2 — relevant third-party content

## PROTOCOL FAILURES (LEARNINGS)
- Built HTML myself multiple times instead of delegating through Gerri → MUST delegate execution
- Anton was only used for end-of-section critique, not inline during work — fixed mid-session
- Didn't check Google Drive with CE token — asked Assaf to change sharing instead of solving
- Multiple image attempts wasted time — should pull multiple options upfront and choose together
- Font sizes got out of control (8 different sizes) — need to enforce hierarchy from start

## HANDOVER / NEXT STEPS
1. **Choose brand images together** — options at /hud/visuals/options/, need Assaf to pick per brand
2. **Full Anton review** of entire page top to bottom before sending to founders
3. **Human cost layer** — write into Context (Section 1) + Messaging (Section 6) + visual section
4. **Font size cleanup** — reduce from 8 to 5 levels across the whole page
5. **Option A label style** — apply "curious endeavor — insight" white-on-red Larken labels consistently
6. **Ground "purposeful absence"** — show what it looks like with sample mockups

---

## AGENT REBUILDS (continued from 2026-02-26)

### Completed Today
- **Gerri smoke test:** Passed — excellent multi-agent brief decomposition with correct dependencies
- **Tatiana assessed:** Voice ✅, Standards ⚠️ (no living baseline), Process ⚠️ (stale "Erica" ref), References ❌. Rebuild plan approved but NOT yet executed.

### Rubric (applied to all agents)
Voice | Standards | Process | References | Smoke test

### Still pending: Tatiana, Julia, Thibault rebuilds

## TASTE BOARD — TREND RADAR

### Julia Cron Set Up
- `julia-trend-radar` cron: daily 9am Lisbon, isolated session, posts to #improvement
- Scans Creative Boom, It's Nice That, Brand New, The Brand Identity for signals

### 6 New Trend Signals Identified (all approved by Assaf)
1. **Open-Source Branding** — DixonBaxi advocating agencies share methodology. CE angle: we already do this.
2. **Design as Collaboration Tool** — Phantasia: design as mechanism for collaboration, not just output.
3. **BBH Rebrand: Strategy-as-Type-System** — Philosophy ("zag") encoded into literal glyph design. High confidence.
4. **Food + Craft Branding Dominance** — Heavy pattern in Assaf's saves. Food/hospitality doing best brand work.
5. **Illustration Renaissance** — Custom illustration displacing stock/AI across brand identities. Medium-High.
6. **AI Companies Need Brand** — Intercom, Perplexity, Gemini by Porto Rocha. AI cos investing in real brand = business opportunity. Tagged "spark."

### NOT yet added to the radar code — pending architecture decision.

## TASTE BOARD ARCHITECTURE DECISION

### Debate+Judge Pattern Run
- **Thibault:** Proposed Vercel KV + Notion bidirectional sync (Option D). 15-20h est.
- **Jessica:** UX spec — image-dominant cards, trend inbox, cross-reference dots, radar viz.
- **Anton (skeptic):** Called both over-engineered for single-user tool. Proposed minimal path.

### Final Verdict: Anton's Minimal Path
- **Read:** Static JSON on Vercel, regenerated every 15 min by cron from Notion
- **Write:** One Vercel API route → Notion directly, optimistic UI
- **UX:** Jessica's image-dominant cards + trend inbox + cross-reference (kept)
- **Cut:** Vercel KV, Supabase, bidirectional sync, radar viz, mobile swipe
- **Effort:** 10-12 hours
- **Status:** Approved by Assaf, NOT yet started. Next step: create Notion Trend Radar DB, route build to Thibault.

---

## HUD Visual Research — Phase 07 Rebuild (10:30 - 15:00 UTC)

### What happened
Assaf wanted to buff up the visual research section (Phase 07: Visual Territories) of the HUD strategy page (v2). Major rebuild:

### Brand lineup (final)
1. **Slack** — 6 images from Assaf (3D brand assets, "Move work forward")
2. **Vercel** — 6 images from Assaf (Basement Studio case study: prism, triangle, Golden Gate, collateral)
3. **Linear** — 4 images from Assaf + 2 supplementary (brand is in-house by Karri Saarinen, no agency case study)
4. **Notion** — 6 images from Instrument case study CDN (storyblok URLs, "For Your Life's Work" campaign)
5. **Aesop** — 6 store photography from ArchDaily + Dezeen (Fillmore, Yorkville, Miami, Shinjuku, Piccadilly by Luca Guadagnino, Diagonal Barcelona)
6. **Oatly** — 6 campaign images from Forsman & Bodenfors case study (bus shelter, mural, packaging, subway, LA bench, oat field)

### Key decisions
- Replaced Stripe → Vercel (stronger visual case, closer to Hud's world)
- Replaced Figma → Aesop (non-tech, architecture as brand)
- Replaced Superhuman → Oatly (non-tech, punk typography as rebellion)
- Each panel has: accordion "Why they had to look different" → 3×2 image grid (16:9) → ONE thing + Said NO to
- "What Hud replaces" separated from category leaders into standalone section below pattern callout
- Added blind test (4 screenshots: Sentry/Hud/Dynatrace/Datadog) + Color DNA palette comparison
- Hud and Sentry palettes are nearly identical (#19181D vs #1F1633 backgrounds, purple accents)

### Image sourcing lesson
- Screenshots of websites look like accidents, not brand research
- Real brand assets come from: agency case studies (Instrument, Basement Studio, Forsman & Bodenfors), architecture sites (ArchDaily, Dezeen), CDN URLs from case study pages
- Assaf's manual screenshots are always better than automated crops
- Aesop Cloudflare blocks headless browsers; Korean site (/kr/) bypassed but architecture portfolios have better images

### Wireframe
- Built landing page wireframe at /hud/wireframe/
- Key moves: light mode (rejects dark observability category), warm accent #E85D3A, `npm install hud` as hero, before/after section, anti-features (Oatly move), pricing on page, no dashboard screenshots
- Live: ce-website-peach.vercel.app/hud/wireframe/

### Files
- All images in: public/hud/visuals/ (crop files: slack-crop-*, vercel-crop-*, linear-crop-*, notion-crop-*, oatly-crop-*, aesop-crop-*)
- Visual research section: public/hud/v2/index.html (Phase 07)
- Wireframe: public/hud/wireframe/index.html

## END OF DAY STATUS (18:25 UTC)

### Taste Board
- Prototype live, 34 entries all extracted
- 192 candidates from Julia scans awaiting Assaf review
- Full build blocked on Notion API token
- Architecture decided (Anton's minimal path), ready to build once unblocked

### HUD
- Brand page live, 8 sections complete
- Waiting on Assaf: brand image picks, human cost layer, full Anton review

### Agents
- All 8 SOULs rebuilt and tested

### Blockers (all on Assaf)
1. Notion API integration token for taste board
2. Review 192 taste board candidates
3. HUD brand image selections + human cost direction