AdapttoAI — Internal

Hey Raffaello,
here's how we work.

Four sections: GitHub setup, how the agentic system works, what's on the agenda right now, and reference docs.

1
Working together on GitHub

The entire B2B consulting practice lives in one repo. All client files, proposals, rules, and notes are versioned here. No Notion, no Google Drive for working files — everything is in the repo.

1

Clone the repo

If you don't have it yet:

git clone https://github.com/gbelpiede/AdapttoAI.git
cd AdapttoAI
2

Always pull before working

Things change fast. Before any session, pull to get the latest state.

git pull
3

Never commit directly to main

Always create a branch first. Claude Code enforces this too.

git checkout -b raffaello/feature-name
# do your work
git add . && git commit -m "describe what you did"
git push origin raffaello/feature-name
# then open a PR on GitHub
4

Where things live

b2b/clients/

  • One folder per client
  • call-notes.md — source of truth
  • proposal.html — internal + external
  • module-*.html — one-pagers

b2b/docs/

  • pricing-philosophy.md — pricing model
  • module-registry.md — what we build
  • icp.md — ideal client profile

b2b/tasks/

  • todo.md — active pipeline tasks
  • lessons.md — what we've learned

b2b/.claude/rules/

  • Proposal rules + deploy recipe
  • Pricing rules
  • Pipeline coverage
  • Meeting prep standards
2
How this works

Open Claude Code from inside b2b/ and it already knows everything — client names, confirmed pains, pricing model, pipeline state. You don't explain context; you just give it a situation. Everything you do goes through a workflow, not a single command. You trigger it, Claude chains through the steps, asks you questions at key gates, and updates the right files automatically.

cd ~/path-to/AdapttoAI/b2b
claude
Every time you open Claude Code

Start with /status — always

Reads all client files, todo.md, and pipeline.json and gives you a brief you can read in 10 seconds. Stale clients are flagged automatically. Here's what it actually looks like:

PROJECT: AdapttoAI B2B STATUS: 3 proposals sent, 1 call Tuesday Apr 7 PENDING: 5 tasks Align with Raffaello before Lamosa call Follow up Eurostar (Ferdi) — 33 days no reply Review new pricing framework before next proposal STALE: Eurostar call-notes — 33 days. Client going cold. NEXT: Lamosa call Tuesday. Goal: verbal yes.
From there, you pick what to work on and the relevant workflow takes over.
Workflow — New company, first call coming up

/meeting-prep [person] [company]

Give it a name, company, and LinkedIn URL. Claude reads our internal docs first — ICP criteria, modules, pricing — then researches the person and company externally. It flags ICP mismatches before you're in the call, maps their likely pains to our modules (Most Likely / Secondary / Longest Shot), writes 5 tailored discovery questions, and builds a decision tree for how to handle the call.

We ran this for Antu before the Lamosa call. The output was a 10-section brief — ICP assessment, Antu's background and what drives her, company snapshot, confirmed pains, agenda, decision tree, tech stack, red flags, what to bring, what to do right after. → See the Lamosa meeting prep

/meeting-prep [person] [company]
meeting-prep.html saved + opened in browser
Workflow — You just had a call

/client-intake → pricing → proposal → one-pagers

This is the most important chain. Paste in your notes — transcript, bullets, anything — and the system runs from there. You don't decide the next step; Claude asks you at each gate.

1
/client-intake — Claude structures your notes: updates call-notes.md, extracts confirmed pains, updates pipeline.json with stage and next action, surfaces open questions.
2
Gate — pre-proposal or proposal? Claude asks based on what was confirmed. Pre-proposal = pains not fully confirmed, price range only, goal is earning the right to scope. Proposal = pains confirmed, fixed price, milestone billing, this is what gets signed.
3
/pricing-workflow [client] — maps confirmed pains to modules, assigns a complexity tier, calculates the price range (optimistic + realistic), surfaces hidden labor cost. Never quote a number without running this first.
4
/ai-proposal — generates the full HTML document and deploys it to the client's Netlify site. Internal amber sections (pricing rationale, deal strategy, contingency notes) are visible only to us.
5
Gate — one-pagers? Claude asks: "This covers N workflows. Want shareable one-pagers?" A one-pager is a standalone page for a single workflow — no pricing, no full proposal — designed to be forwarded by the champion to the team lead who owns that specific process. Claude also writes the pre-filled WhatsApp message for the champion to send. You don't need to be in that internal conversation.
/client-intake
/pricing-workflow
/ai-proposal
/one-pager
deployed
Also: pipeline.json is updated automatically after /client-intake, which means the Venture Signal tables refresh live — no manual sync needed.
Always on — the system's living output

The Venture Signal report

The single document that answers "what are we building, for whom, at what price, and where does the pipeline stand?" Not a deck — an intelligence report built from all client data. The pipeline tables load live from pipeline.json, so they're always current without a manual rebuild.

VS
Pipeline tables — live stage, next action, and temperature for all 6 clients. Refreshes every time /client-intake runs.
VS
Workflow catalog — every module we sell: Track A (Commercial) and Track B (Ops Intelligence). One platform, one ERP connector per client, modules stack on top.
VS
Client Intelligence Map — ICP scoring and opportunity scoring across all 6 clients. Shows which pains repeat most and which clients have the highest potential.
VS
Partner Sync — items flagged specifically for Giuseppe and Raffaello to align on before the next client interaction.
VS
Competitor benchmarks — Mercura (YC W25), Vendavo, Treble. Where we sit, what gap we fill that they don't.
To rebuild: run /venture-signal after a significant new signal (new client, confirmed pain, product decision). Pipeline tables never need a rebuild — they're always live.
These are the three main workflows. There are more — the system will suggest them at the right moment. Every Monday at 8am, /adapttoai-review runs automatically and drops a prioritized action list into the AdapttoAI Telegram group. No trigger needed.
3
What we need to do

Updated Apr 4. Two strategic topics to align on first, then the commercial pipeline below.

Strategic — align before anything else
Before Tue Apr 7

Pricing framework — this changes how we sell

New model built April 2026: module-based annual fee, 12-month minimum, no separate implementation fee. You haven't reviewed it yet. This isn't just a pricing doc — it's a strategic shift toward recurring revenue. The retainer model in your MAGG proposal ($140K MXN/mo, time-based) is the old model. We need to align before Lamosa and before Omar replies.

File: b2b/docs/pricing-philosophy.md — read the prepay table and the floor/ceiling section. The goal is to make every engagement a subscription, not a project.

Discuss

Product — What are we actually selling?

Two open questions: (1) Which features in the MVP prototype map to the modules we're proposing? Does it reflect what we're actually building? (2) Neither of us has a shared view of which modules are proposed to which client — we need this before any new client call.

→ MVP prototype · → Venture Signal — Workflow Catalog

Commercial — active pipeline
Before Tue Apr 7

1. Lamosa — align before the call

Tuesday call with Antu (+ possibly Doménico). Goal: verbal agreement to move to formal proposal. Three things to lock before the call: (1) which pricing tier to lead with — Quick Win $12–17K vs Core $20–28K, (2) Mexico/Peru sign-off — can San Lorenzo Peru decide independently?, (3) who asks what (Giuseppe, you, Diego).

→ Internal proposal · → Meeting prep brief (Antu)

MAGG — Apr 8

2. MAGG — decide pricing model before Omar replies

Proposal sent. Follow up no earlier than Apr 8. The key question isn't the math — Claude can run /pricing-workflow magg and generate module-based numbers in minutes. The real question is strategic: do we shift MAGG to the new subscription model, or hold the retainer approach for ops-heavy engagements? Decide with Giuseppe before Omar comes back, so we're aligned on how to respond.

→ Venture Signal — MAGG analysis · → MAGG proposal (sent)

This week

3. AutoalDía — unblock or park

Pre-proposal exists but isn't finished. Stalled for 7+ days. Blocker: WhatsApp API access path needs to be confirmed before sending. Who owns this client and closes the brief?

→ AutoalDía pre-proposal (draft)

ASAP

4. Eurostar — follow up with Ferdi today

Pre-proposal sent March 2026. No reply. 33+ days of silence — going cold. Send this now. Main message to Ferdi, plus two short forwarding snippets for his After-Sales Manager and Technical Lead (one-pagers already built).

→ Full pre-proposal · → One-pager: After-Sales · → One-pager: Technician App

ASAP

5. Aronlight — resend to Manel with corrected links

Message sent Apr 3 had old Netlify links. Resend now with updated links. Same message, same forwarding snippets — links now point to Cloudflare Pages.

→ Full pre-proposal · → One-pager: Sales · → One-pager: Technical Matching

4
Other documents to know

You won't need these every session, but they're the source of truth the workflows read from. If something doesn't add up in a proposal or pricing conversation, these are the first files to check.

P$

pricing-philosophy.md — read this first

New pricing model built April 2026. Module-based annual fee, 12-month minimum, no separate implementation fee. Team rates, complexity tiers, build cost estimates, competitor benchmarks, and a hidden labor cost calculator. The Learning Log at the bottom gets updated after every deal — won or lost.
File: b2b/docs/pricing-philosophy.md

MR

module-registry.md — what we build

Canonical definitions of every module: A1 (Discount Approval), A2 (Quote Gen), A3 (Spec Matching), B1–B4 (Ops Intelligence). Each entry has scope, complexity tier, ERP dependencies, and open questions to confirm before pricing. Read before any pre-proposal.
File: b2b/docs/module-registry.md

CN

call-notes.md — source of truth per client

Every client folder has one. Confirmed pains, discovery answers, open questions, pipeline stage, internal language. Claude reads this before doing anything for a client. Updated automatically via /client-intake — never edit manually unless you're correcting something specific.
Location: b2b/clients/[client]/call-notes.md

PJ

pipeline.json — the system's single source of truth

Holds current stage, next action, and partner sync items for every client. Drives the Venture Signal tables and the Monday review. Updated automatically after every /client-intake. Edit manually only if you have context from outside the system (e.g. Omar texted you back). Update next_action, status, and the updated date.
File: b2b/venture-signal/pipeline.json