Ahmad Faisal Rahman · Co-founder/CTO · Fintech / F&B · SMB · 28 employees
Two gaps dominate. First, the only AI signal in the record is Ahmad Faisal Rahman's LinkedIn post about Cursor two days ago — a single-founder enthusiasm signal with no evidence of licence counts, IDE rules, MCP servers, or daily usage across the 28-person org, so we cannot size Day 1 or forecast adoption. Second, nothing in the record shows AI touching PR review, testing, or agent guardrails — a material omission for a fintech/F&B handling payments data, which makes Day 2's guardrail content the highest-risk, highest-value block. Highest-leverage 90-day move: run a two-week instrumented baseline (licence audit, per-sub-team usage telemetry, one PR-review sample) before cohort 1, then deliver a single right-sized cohort and re-measure at day 60 to prove the RM 3,500/seat against real cycle-time movement.
The founder's LinkedIn post about Cursor two days ago is the only tooling evidence in the record — it proves intent, not a stack. With 28 total employees and no licence count, IDE-rule inventory, or MCP/secrets setup on file, Day 1 would have to open with a stack audit rather than a build.
Nothing in the record shows AI touching PR review, testing, code generation, or agent guardrails — the four things Day 2 exists to install. For a fintech/F&B handling payments, the absence of any stated blast-radius or agent-approval policy is the most decision-relevant gap in the whole file.
The co-founder/CTO personally posting about Cursor two days ago is a strong top-down signal — culture is the one dimension with genuine evidence. It lands at 55 rather than higher because the record shows enthusiasm only: no stated blast-radius tolerance, no workflow-change commitment, and the deal is still New lead at 10% probability.
No cycle time, PRs/dev/week, time-to-merge, or deploy frequency appears in the record, and engineering headcount inside the 28 employees is unknown. Without a baseline the workshop cannot show a before/after, so Day 1 should spend its first hour pulling these four numbers straight from the VCS.
The record contains exactly one adoption data point — a single founder visibly using Cursor. There is no licence utilisation figure and no view of spread across sub-teams, so the 35-seat cohort ceiling is irrelevant here; the live question is whether this org can fill even a 10-seat cohort.
Defect rate, escaped bugs, time-to-feature, and dev satisfaction are all absent from the record, so there is no quality baseline to defend RM 3,500/seat against. The only outcome-adjacent facts are the 64/100 health score and RM 42,000 ARR — a small base where one escaped regression is proportionally expensive.
No proposals yet. The AI writes one from the readiness check.