Coral research → proposed changes
Operator review · 2026-07-09 · source: hermes coral-juergen-expansion-2026-07-09/REPORT.md (3a768fa) · meeting: Juergen Hoebarth, Fri 10 Jul 08:00Z
Nothing below is applied. Each item ships only on your OK. No CRM writes anywhere in this doc.
A — Live guest-portal pack edits (coral-content.json → deploy before 08:00Z)
Surface Juergen sees. File: services/meet-boas-dev/assets/guest/coral-content.json — one PR, CI deploy.
A1 · Omnibus scope claim — wrong direction high
partnership.situation[1]
currentOmnibus I (in force 2026-03-18) locks CSRD scope: a wave of mid-sized DACH companies now needs compliance tooling they do not have.
proposedOmnibus I (Directive (EU) 2026/470, in force 18 March 2026) locks CSRD scope at >1,000 employees and >€450M turnover: fewer, larger DACH manufacturers in scope — and ESRS 2.0 (adopted 3 July 2026) cuts mandatory datapoints ~60%, lowering the cost of saying yes to tooling.
Why: Omnibus RAISED thresholds — mid-caps left scope, they didn't enter it. Juergen's domain; the current line is backwards. Also lands the 6-day-old ESRS 2.0 hook. [hermes §1, EU Commission primary]
A2 · Same stale framing in the story timeline high
story[1].body
currentDirective 2026/47 locks CSRD scope. Mid-sized companies across DACH enter compliance obligations — the greenfield window opens, and it will not stay open.
proposedDirective (EU) 2026/470 locks CSRD scope. The in-scope class — DACH manufacturers over 1,000 employees — must choose tooling now; whoever their advisors recommend first becomes the default.
Why: same wrong-direction claim + truncated citation. Keeps the urgency beat, grounds it correctly.
A3 · Truncated directive citation in April meeting summary low
meetings.m-2026-04-24.summary
current… riding the Omnibus I wave (Directive 2026/47, in force 2026-03-18) that locks regulatory scope and pushes mid-sized companies into CSRD compliance.
proposed… riding the Omnibus I wave (Directive (EU) 2026/470, in force 2026-03-18) that locks regulatory scope for DACH companies.
Why: cite fix + drops the mid-size framing from our recap. (It is a record of the April call — minimal touch, not a rewrite.)
A4 · Partner roster stated as verified fact medium
partnership.situation[0]
currentCoral.li is an ESG/CSRD compliance platform with a world-class partner roster — KPMG (signed), WSP (6-month pilot complete), Bureau Veritas, EDAI, Contango/ADQ — and proven partner-led distribution in MENA and beyond.
proposedCoral.li is an ESG/CSRD compliance platform with partner-led distribution proven in MENA — the roster Coral described in April: KPMG, WSP (pilot), Bureau Veritas, EDAI, Contango/ADQ.
Why: none of the 7 names verifiable publicly (absent from Coral's own llms.txt). Source was Coral's own April statements — keep the names, attribute them instead of asserting. Content-rules §2.3: attributed, not flat fact.
A5 · HubSpot + ClickUp integration stated as shipped medium
partnership.resolution[1]
currentPowered by the engine you are looking at: pre-meeting intelligence, post-meeting analysis, CRM knowledge graph, and agent-driven outreach — plugged into Coral's HubSpot + ClickUp piping.
proposedPowered by the engine you are looking at: pre-meeting intelligence, post-meeting analysis, CRM knowledge graph, and agent-driven outreach — ready to plug into the HubSpot + ClickUp piping Daniele offered in April.
Why: same class as the machine_note purge in PR #254 — unshipped capability as fact. Offer stays, attributed.
A6 · NEW beat — Coral's live EU/Germany deployment addition · your call
partnership.complication — append as complication[2] (or fold into [1])
proposed additionCoral already runs a live EU/Germany production deployment with full in-country data residency — the DACH plumbing exists. What is missing is the go-to-market: people, language, presence. That gap is this partnership.
Why: hermes verified (fact HIGH) — contradicts April "no EU expansion" read. Strongest new argument FOR the partnership; makes the pitch time-sensitive. Optional because it's a new claim about THEIR infra — you may prefer to voice it live instead of printing it.
B — Pre-meeting brief fold-ins (prep_md / your talking points)
Say
- ESRS 2.0 adopted 3 July (~60% fewer datapoints, EP/Council scrutiny pending, applies FY-2027, early-adopt FY-2026). Fresh hook — his messaging problem.
- The Envoria signal: 11 of our 20 roster orgs come from competitor Envoria's public certified-partner directory — firms already running the CSRD-channel motion for a rival. Warmest enumerable target class.
- Their EU/DE deployment + data residency — "you built the plumbing; we're the GTM."
- Machine-produced roster in one day — 20 orgs, every row ≥2 sources, top 10: WIRAS Verbund, Sustainable AG, HSP Gruppe, Systain, dhpg, valantic, ESTURION, WTS Advisory, Kleeberg, :response. Plus DStV (36,500-member association) as a co-marketing play.
Don't say / don't lean on
- German CSRD-Umsetzungsgesetz being in force — NOT confirmed passed (only "expected summer 2026"). Re-check a dated source right before the call if needed.
- Coral funding status — unknown either way; don't assume a raise.
- The 7 named partners as live DACH channels — unverified; if raised, mirror their April framing only.
C — Implementations (post-meeting; NOT meeting-blocking)
C1 · Roster intake — K/A/N gate before any CRM write
20 orgs land in a staging pass under rule #20 §6 Class-N: identity core ≥2 sources (hermes already delivered per-row URLs), namesake sweep, per-row provenance note, status=raw. Excluded-not-guessed stay out (msg global, SI PRO, part.de, LeverX). Nothing touches crm_* until gated.
C2 · OSS machine adopt sequence (rule #18 clean — hermes deep-verified top 3)
| # | Leg | Tool | License | Note |
|---|
| 1 | Dedup + org-fit scoring | Splink (DuckDB backend) | MIT ✅ | hermes RAN it — 0.5s, no Spark. Highest ROI, first. |
| 2 | Registry baseline | GLEIF LEI API (n8n HTTP node) | free ✅ | Live-tested BUT SMB-sparse (DE 252k LEIs vs ~3M cos) — baseline only. |
| 3 | SMB registry gap | bundesAPI/handelsregister scraper | ⚠️ none declared | License clarification before productionizing — treat all-rights-reserved until resolved. No official DE registry API exists. |
| 4 | OSINT enrich | SpiderFoot | MIT ✅ | Cost = upstream data-source keys; size first. Output = P3-at-best → gate. |
| 5 | Safety layer | Supabase staging + per-row provenance + operator gate | — | Built alongside, not after (rule #20 §3). |
Chain: n8n → Firecrawl (wired) → Splink → GLEIF/Handelsregister → SpiderFoot → staging(status=raw) → K/A/N gate → crm_*. Rejected with reason: Zingg (AGPL), dedupe/recordlinkage (stale), Maltego CE (not OSS), OpenCorporates (commercial), crawl4ai (redundant).
Gate summary: A1–A5 = one pack PR + deploy on your OK (A6 optional). B = folded into prep_md on your OK. C = filed as board items, executed after the meeting. Invitation lane unchanged — still waiting on apollo 093 design return + your wording OK before any send.
metis · sources: hermes REPORT.md §1/§2/§3 + sources/A,B,C · live pack read 2026-07-09 · tiers per row noted in source files · no CRM writes performed or staged