Evidence Signal
22 commands · 11 sales skills · CRM with deal memory · self-personalizing
Project Deep Dive
SalesSidekick for Claude Cowork
A personal sales operating system built inside Claude — processes calls, builds strategy, tracks deals, generates presentations, and learns your business as you use it
A personal sales operating system inside Claude — processes calls, builds strategy, tracks deals, and learns your business as you use it.
Sales productivity toolSelf-personalizationDeal intelligenceEnterprise sales methodology
Flagship Case Study
The systems thinking behind Claude Cowork
Most AI tools stop at prompting. This system was built to behave more like an operating environment that adapts to the user, preserves context, and enforces evidence quality.
The value here is not merely that it runs inside Claude Cowork. The value is that the product carries memory, structure, evidence standards, and graceful degradation so the user can trust it day after day.
Where the gap was
Sellers had models with raw capability, but no persistent operating layer for context, call processing, deal memory, or strategic workflows.
What had to change
The system needed to personalize itself, keep state, and preserve output quality even when optional connectors or third-party tools were unavailable.
Why it matters
It demonstrates product design at the workflow level: commands, memory, evidence discipline, and user adaptation working together as one system.
Positioning Lens
SalesSidekick for Claude Cowork
This project is important because it shows how I productize workflows, not just models. It turns a powerful model into a repeatable, personalized system for doing real work.
Signature workflow decisions
- Built self-personalization into setup so the system configures itself around the operator instead of forcing a generic default.
- Made evidence grading a product behavior, not a documentation promise.
- Designed graceful degradation so the workflow still works when external systems are missing, slow, or incomplete.
The Problem
What problem was worth solving?
I build things to solve a specific pain, not to decorate a portfolio. This is the pressure that made this project necessary.
Sales teams need AI that can execute repeatable workflows — processing calls, building strategy, tracking deals — not just write one-off responses.
Without persistent deal memory and structured workflows, AI-assisted sales work stays fragmented and inconsistent.
SalesSidekick for Claude Cowork packages 22 commands, 11 sales skills, and persistent deal memory into a system that personalizes itself to your business.
The Build
What I built to solve it
A self-personalizing sales assistant: 22 commands cover the full sales workflow from daily planning through deal strategy and call processing. Connected to a Notion CRM (6 databases, 70 fields) for persistent deal memory. Every output is evidence-graded so sellers know what is verified vs. estimated. A setup wizard learns your company, competitors, and preferences — then tailors all outputs.
L1Command System22 slash commands across 7 workflow categories
Invoke
L2Skill Framework11 domain skills — 1,866 lines of structured knowledge
Read/Write
L3Notion CRM6 relational databases with persistent state
Generate
L4Output EngineEvidence-graded artifacts with graceful degradation
Click any node to explore details
Commands
22 sales workflows
Full coverage: daily planning, call processing, deal strategy, territory management, communication, account management, and system utilities.
Sales skills
11 domain frameworks
Repeatable sales methodologies built in: deal qualification (MEDDPICC), competitive analysis, call intelligence, account strategy, and more.
Deal memory
6 Notion databases · 70 fields
Companies, Contacts, Deals, Tasks, Call Notes, LinkedIn Posts — auto-created during setup with full relationships and deal tracking.
Codebase
7,000+ lines across 47 files
2,870 lines of commands, 1,866 lines of sales skills, and a comprehensive system configuration.
Value
What changed because it existed
This is the clearest evidence of practical value, system leverage, and execution quality.
Deal intelligence
Qualification-driven deal analysis
Call and deal workflows structured around qualification depth, risk identification, and concrete next actions.
Presentation generation
Native PowerPoint output
Generates presentation decks directly from deal context and analysis — no copy-pasting between tools.
Persistent memory
Deals, contacts, and notes saved automatically
Everything is written back to the CRM so the system remembers context across sessions — no re-explaining last week's calls.
Multi-output workflows
One command, multiple deliverables
A single call closeout command produces a deal scorecard, follow-up tasks, coaching notes, email draft, risk signals, and competitive intel.
Tech Stack
What it runs on
The stack matters here because it reflects design choices, constraints, and how the system was intended to scale or integrate.
Claude Cowork / Claude Code operating layer22-command workflow frameworkNotion MCP integration (6 databases, 70 fields)PptxGenJS native deck generationGamma presentation connectorGmail / Google Calendar / Google Drive connectorsEvidence grading system (Verified/Estimated/Hypothesis)Self-personalization engine (/setup wizard)
Build Story
How the thinking unfolded
This is the reasoning path behind the output, not just the finished artifact.
01
Learns your business, not just your name
Run setup once and the system rewrites itself — your company context, competitive landscape, brand voice, and professional profile are embedded into every future output.
02
Honest about what it knows
Every output tags claims as Verified, Estimated, or Hypothesis. When more than half the output is based on assumptions, the system warns the seller — no AI hallucination disguised as fact.
03
Real sales methodologies, not generic advice
Deal qualification (MEDDPICC), multi-angle strategic analysis, competitive positioning, and structured call processing — proven enterprise methodologies encoded directly into automated workflows.
Capability Signal
What this project demonstrates
Each project is a proof point. These are the capabilities it most clearly reveals.
Sales Tool Design
Built a 22-command, 11-skill sales system with self-personalization, graceful degradation, and persistent deal memory.
Sales Methodology Automation
MEDDPICC deal scoring, strategic analysis frameworks, competitive positioning, and structured call processing — real methodologies turned into automated workflows.
Persistent Deal Memory
6-database CRM with 70 fields keeps deals, contacts, tasks, and call notes synchronized across sessions — the system remembers everything.
Evidence Grading
Every AI output is evidence-graded (Verified, Estimated, or Hypothesis) with automatic warnings when outputs are assumption-heavy.
Continue
Want to go deeper?
Ask the Latif AI Guide about the architecture, the commercial logic, or what this project says about how I approach hard problems.