Log

Append-only chronological record of operations on the wiki. Each entry begins with ## [YYYY-MM-DD] <op> | <title> so it can be parsed with grep "^## \[". Operations: ingest (new transcript processed), query (notable Q&A filed), lint (health check), meta (structural change to the wiki itself).

[2026-04-23] ingest | Amplemarket Overview

Added comprehensive overview of Amplemarket as sales automation platform; Elad Flishko presented product capabilities (signal-based sequences, account management automation, multi-channel execution) and strategic positioning for Monday.com’s low-touch account segment. Identified DEC-021 (no voice agents due to legal restrictions), AI-032 (internal SDR alignment needed), and OQ-028/OQ-029/OQ-030 regarding integration strategy and legal frameworks.

[2026-04-23] ingest | AI for AM

Brainstorm session with Jonathan Buchen (Account Manager, Neema fintech) on AI applications for go-to-market roles. Key insights: AMs use ChatGPT daily for transcript summarization and account health analysis (Tableau dashboards); QBR prep is scattered and could benefit from AI-powered deal room; AI voice agents are valuable for scale segments but risk perception damage in enterprise; private in-call AI chat (Win.ai model) is more viable than voice agent on-call. Identified use cases, blocking questions on data input burden and threshold analysis, and concept definitions around gap-closer role and segment-level feedback loops.

[2026-04-23] ingest | AI for GTM with Shirin (AAICO)

GTM research conversation (52 min, English) with shirin-aalami (Head of Sales ROW + AI Solutions Architect, aaico). Key findings: third external validation of agent-in-call — Shirin already queries Gemini mid-call, prefers text over voice. wonderfull competitive intel: ~$500K ACV but churn after 2 weeks — validates “immediate aha moment” thesis. Dropbox learning: outsourced BDRs = low quality; AI should work invisibly to bridge info and action. PLG benchmark: Lovable. Created source [[ai-for-gtm-shirin-2026-04-23]], entity pages [[shirin-aalami]], [[aaico]], [[lovable]]. Updated [[agent-in-call]], [[competitive-landscape]]. Added AI-028 (Saar to connect with Mike, Shirin’s ex-Dropbox colleague).

[2026-04-20] ingest | AI for GTM (Cerebras)

GTM research conversation (33 min, English) with anthony-landsberg (AE, cerebras). Niv led the interview. Key findings: primary GTM tool is ample-market (CRM+LinkedIn signals, sequences, dialer); Claude Code for research/copy. AI multiplied outreach 4-5x but conversion rate flat — volume is solved, hit-rate is not. Biggest pain: getting through to people (spam, call screening). Has not used voice agents. Cerebras filed S-1 (pre-IPO). Created source [[ai-for-gtm-cerebras-2026-04-20]], entity pages [[anthony-landsberg]], [[cerebras]], [[ample-market]]. Updated [[agent-in-call]] (contrasting data point), [[niv]].

[2026-04-20] ingest | New Startup Workspaces: Risks & Alignment

Core strategic session (37 min, Hebrew) with Vitali, Nizan, Saar + Barak (IT/Security), Gil Lan (Legal), finance/ops team. Topic: operational separation model for Harmony.ai and Charm.ai under Monday’s legal umbrella. Key decision: semi-autonomous operations (independent domain, Slack, G-Suite, AWS, GitHub, Okta, bank account) but single Monday legal entity. Both teams target SOC 2 compliance by Sept 2026; will hire external SOC 2/compliance advisor from day 1. Unresolved: whether technical separation suffices for regulatory compliance, device-level isolation requirements, Okta/AWS management model. Created source [[new-startup-workspaces-2026-04-20]], added decisions DEC-006/DEC-007/DEC-008 (workspace separation & compliance model), action items AI-018–AI-022 (advisor, legal audit, fast-track process, control mapping, tech-stack docs), open questions OQ-020–OQ-024 (device separation, regulatory liability, identity mgmt, DLP/EDR flexibility, fast-track criteria). Updated domains/infrastructure.md implications.

[2026-04-20] meta | Remove podcast-prep-2026-04-19 from memory (not Harmony-related)

Per user request, deleted transcripts/raw/podcast_prep_2026-04-19_08-00.json, the source page, the notetaker-brain concept page (only sourced from podcast-prep), and all related references. Removed from processed.md, added to ignored.md. Kept designer-as-developer (still anchored in ai-first-design-setup-2026-04-20), yuval-ezov (referenced from the AI-first-design session as front-end rules author), and elad (attended the AI-first-design session).

[2026-04-20] ingest | AI first design setup

Internal session (33 min, Hebrew) with saar-arbel, ben, ayala + elad (lead of monday’s vibe design system). Goal: set up AI-first design + dev infrastructure for harmony.ai before the team starts building. Created source [[ai-first-design-setup-2026-04-20]], entity pages [[elad]] and [[yuval-ezov]], concept pages [[vibe]] and [[claude-code]]. Updated [[designer-as-developer]] and domains/design.md. Added action items AI-024–AI-027 and open questions OQ-016 / OQ-017. Direction emerging: standardize on Claude Code over Cursor; build on Vibe atoms (not from scratch); install established Claude Code skills.

[2026-04-20] meta | Add lint-wiki tooling

Added scripts/lint-wiki.sh (Bash wrapper) + scripts/lint-wiki.py (mechanical checks). Catches INDEX_MISSING / INDEX_STALE / ORPHAN / BROKEN_LINK / FRONTMATTER issues. Updated CLAUDE.md Lint operation to two passes (mechanical via script, semantic via LLM). Added /lint-wiki slash command.

[2026-04-20] lint | First mechanical pass on the wiki

Initial run surfaced 25 issues: 2 index_stale (false positive — ../ paths; fixed in script), 23 broken_link (real misses — entities/concepts referenced in [[wikilinks]] but no page existed). Created 15 stub pages: companies (google-ads, facebook-ads, hockeystack, clari, dreamdata, 6sense, retell, bland, vapi, sativ, lupo) and concepts (amanda, ab-testing, language-strategy, kremer). Updated index.md with new pages, regrouping companies by relationship-to-Harmony. Re-ran lint: 0 issues.

[2026-04-19] meta | Migrate to Karpathy LLM Wiki pattern

Restructured memory/ into entities/concepts/synthesis/sources layout. Moved structured ledgers under ledgers/. Replaced thematic map.md with comprehensive index.md. Replaced daily journal/ with this log.md; legacy daily files preserved under archive/journal-legacy/. Updated CLAUDE.md schema with wikilink conventions, page templates, and ingest/query/lint operations. See .claude/workflow/plan.md.

[2026-04-19] meta | Remove context_layer_weekly_2026-04-15 from memory

Per user request, deleted context_layer_weekly_2026-04-15_08-00.json and all extracted artifacts: DEC-020, DEC-021, AI-020, AI-021, AI-022, OQ-015, the Context Layer Pipeline section in domains/infrastructure.md. Removed from processed.md, added to ignored.md to prevent re-fetch. ID gaps left intentionally — no renumbering.

[2026-04-19] meta | Add user_interviews.md (later retired in wiki migration)

Created memory/user_interviews.md as a top-level memory file for external user/prospect/design-partner interviews. Migrated Atera CS/AM interview content from domains/product.md. Updated map.md and CLAUDE.md. (Note: this file was retired in the subsequent LLM Wiki migration; Atera content now lives in entities/companies/atera.md + entities/people/rachel-rembrandt.md + sources/atera-2026-04-19.md.)

[2026-04-19] ingest | Atera | monday — the future of AMs & CSM

Customer research meeting (48 min, Hebrew) with Rachel Rembrandt (Director of Customer Success, Atera, ~$100M ARR). Source: atera_monday_the_future_of_ams_csm_2026-04-19_07-00.json. Created entity pages [[rachel-rembrandt]] and [[atera]]. Source page: [[atera-2026-04-19]]. Action items added: AI-023 (connect with Atera SDR/BDR lead). Validated [[crm-replacement-thesis]] and [[agent-in-call]] concept (first external customer validation).

[2026-04-15] ingest | Vision (Nadav Greenberg session)

Strategy crystallization meeting with Nadav Greenberg. 5 decisions added (DEC-015 through DEC-019), 4 action items (AI-016 through AI-019), 5 open questions (OQ-011 through OQ-014). Key outcomes: voice is a feature not the core, brain layer works for human workers too, API-first architecture, lightweight marketing connectors as second vector, support bring-your-own agents.

[2026-04-14] ingest | Offsite (full-day Harmony team session)

Full-day team offsite. Decisions added: DEC-004 through DEC-008 (Global Maximum vision, self-learning SDR, July 2026 production target, Touch GTM, revenue-platform positioning). Action items: AI-008 through AI-014. Open questions: OQ-004 through OQ-009.

[2026-04-14] ingest | Competitor strategy (offsite follow-up)

Sharpened wedge strategy. Decisions: DEC-012 (three-layer packaging), DEC-013 (self-learning revenue voice agent positioning), DEC-014 (mid-market B2B ICP).

[2026-04-20] meta | Remove product-impact-measurements-2026-04-13 from memory (not Harmony-related)

Per user request, deleted the source page, the product-impact-measurement-framework and ab-testing concept pages (only sourced from this meeting), the raw transcript, and the ledger entries DEC-009 / AI-015 / OQ-010. Removed cross-references from domains/product.md, entities/companies/monday.md, entities/people/tamar.md, index.md, and the related-decisions list in roadmap.md. Kept the existing transcripts/ignored.md row (already marked “Not relevant to Harmony”). ID gaps left intentionally — no renumbering.

[2026-04-13] ingest | Harmony — Analyzing campaigns (with Tamar)

Campaign analysis discussion with Tamar (marketing). Decision DEC-003 (monthly cadence, all campaigns, include cost + ARR data). Action items AI-003 through AI-007. Open questions OQ-002, OQ-003.

[2026-04-12] ingest | Harmony — BizApps sync (with Guy Kori)

Data access kickoff with Guy Kori (BizApps). Decisions DEC-001, DEC-002 (Salesforce/Gong API access via Guy; prefer raw API over BigBrain). Action items AI-001, AI-002. Open question OQ-001.

[2026-04-20] ingest | AI first design setup + Atera CS partnership

AI first design setup (2026-04-20)
Team designing Harmony with AI-first approach: leveraging Claude Code visual design tool + Figma tokens on Wave component library. Strategy emphasizes product velocity over heavy design systems, using agents for rapid iteration.

Atera | monday - future of AMs & CSM (2026-04-19)
Strategic CS knowledge sharing: Atera VP (Rachel Rembrandt) shared customer success operationalization model with Harmony team. Key learnings: layered CSM metrics (commercial, operational, engagement), AI-driven account mapping + one-click QBR using Claude Code, critical product-feedback-loop gap. Long-tail customer model (100M ARR) requires segmented approach.

New entities: rachel-rembrandt, atera
New concepts: cs-metrics-layered, customer-mapping-ai, one-click-qbr, product-feedback-loop-gap, design-system-lightweight, claude-code-design-workflow, product-velocity-design, agent-driven-design
New sources: 2 pages