CRM+ AI for PT. Adhi Karya Persero
The client’s marketing and tender processes were fragmented and handled manually across units. CRM+ centralizes all data, standardizes workflows, and adds AI-driven scoring for better evaluation. The result is faster coordination, smarter decision-making, and increased tender capacity.
71%
time saved using automated data scrapers compared to manual data collection.
120%
faster report creation and export.
12%
increase in the number of tenders evaluated after CRM+ implementation.

Product Manager
6 Months
2 Frontend Engineers,
2 Backend Engineers,
1 Quality Assurance,
1 System Tester,
1 UX Designer
IDR 550 - 750 million (2025)
500+
Overview
Adhi Karya, one of Indonesia’s largest state-owned construction companies, previously managed marketing and tender activities in a fragmented and manual way across individual business units. Key client, project, and tender data were stored in Excel files without standardized formats—resulting in frequent data duplication, inconsistent information, and delayed analysis. CRM+ was developed as an AI-powered centralized platform to streamline tender workflows, manage partner qualification more efficiently, and enable data-driven decision-making. By unifying marketing and tender processes across units, CRM+ replaces manual work with integrated systems that give stakeholders real-time visibility into project pipelines, risks, and opportunities.
My Contribution
This project marked my transition from building “functional systems” to shaping decision-making products at enterprise scale.
I was responsible for defining how AI could meaningfully support complex tender workflows—not as a gimmick, but as a practical tool for teams dealing with long sales cycles, high-value projects, and strict governance. My role covered end-to-end product ownership: from uncovering real operational pain points, translating them into AI-backed product concepts, aligning stakeholders, to guiding implementation and validation.
More than shipping features, my focus was ensuring the product reduced cognitive load, shortened preparation cycles, and helped leaders make faster, more confident decisions.

Problems
Identifying the Core Problem
Tender preparation in large construction organizations is inherently complex. One tender requires:
Inputs from multiple departments
Referencing past tenders manually
Repeated validation of partner eligibility and documents
In reality, teams relied on spreadsheets, shared folders, and tribal knowledge. This led to:
Long tender preparation cycles
High dependency on specific individuals
Inconsistent data and duplicated work
Limited visibility for executives into pipeline health and risk
Despite having large volumes of historical data, none of it was being used to predict, guide, or accelerate decisions.
This gap became the opportunity.

Proposed Solution: CRM+ with AI as a Decision Layer
Rather than positioning AI as “automation,” I framed it as a decision-support layer on top of CRM workflows.
Key principles I defined:
AI should reduce manual referencing, not replace human judgment
Insights must be explainable and actionable
Outputs must fit naturally into existing tender workflows
Core AI-Driven Concepts
Historical tender intelligence to surface relevant past submissions automatically
Partner qualification signals based on experience, documents, and past performance
Pipeline risk visibility to help management identify bottlenecks and weak points early
All AI outputs were designed to support—not interrupt—existing CRM usage.
Ideation & Validation Process
Aligning Stakeholders Early
I worked closely with business users, management, and engineers to stress-test assumptions:
What decisions are hardest to make today?
What data is trusted—and what isn’t?
Where would AI actually save time?
Feasibility-Driven Design
Instead of designing idealistic AI features, I:
Prioritized low-risk, high-impact AI use cases
Defined clear input–output expectations for engineers
Ensured models could evolve incrementally as data quality improved
This avoided over-engineering while still delivering visible value early.
The Result
CRM+ evolved from a traditional CRM into a strategic tender management platform:
Tender data became centralized and reusable
AI-assisted insights reduced manual searching and comparison
Executives gained clearer visibility into project pipelines and risks
Most importantly, the system shifted how teams prepare and evaluate tenders—from reactive and manual to structured and insight-driven.
Challenges & Learnings
Balancing AI Ambition with Reality
Not all data was clean or structured. I learned to:
Design AI features that work with imperfect data
Clearly communicate limitations to stakeholders
Treat AI as an evolving capability, not a one-time launch
Enterprise Change Takes More Than Features
The hardest part wasn’t technology—it was trust.
Building confidence in AI outputs required:
Transparency
Gradual rollout
Strong alignment with existing decision habits
Key Takeaways
AI is most powerful when it augments human decisions, not replaces them
Enterprise products succeed when they respect existing workflows
Shipping value early beats waiting for “perfect” data
A good PM designs not just systems—but how people think and decide
Get Athos Pro

