Saas Replacement Planner

SaaS Replacement Planner

Evaluate a company's SaaS stack and produce a rigorous, actionable plan that quantifies the ROI of migrating from subscription software to AI-agent-powered alternatives. Bias toward replacement where the economics support it, but stay honest: not every tool can be replaced today.

Contents

• references/analysis-framework.md -- the six-step per-tool analysis (classification, feasibility tiers, build-cost formulas, Claude+MCP architecture, risk scoring, priority matrix)
• references/output-template.md -- full saas-replacement-plan.md document structure
• references/patterns.md -- proven replacement patterns by category and edge-case handling
• references/analysis-guidelines.md -- estimation rigor, honesty rules, OneWave AI thesis, and output quality standards

Workflow

1. Gather input. Collect the SaaS tool list. Parse screenshots, CSVs, or bank statements; organize informal lists into a structured format (tool, monthly/annual cost, seats, primary use case). Ask clarifying questions only when critical data is missing: seat counts, primary workflows, integration dependencies, mission-critical vs nice-to-have, and compliance requirements. 2. Research current pricing. For each tool, verify pricing with WebSearch when numbers are missing or look off. Check per-seat pricing, outdated plans, API access for the replacement, and data export capability. 3. Analyze each tool. Run the complete six-step framework in references/analysis-framework.md for every tool. Do not skip tools or give superficial analysis. Apply matching patterns from references/patterns.md. 4. Build the priority matrix. Plot all tools on the Impact vs Effort axes. Sequence replacements and group tools that share infrastructure (e.g., all needing Supabase) to reduce incremental build cost. 5. Generate the timeline. Create a realistic timeline accounting for engineering capacity, parallel running periods, dependencies between replacements, and quick wins that fund later investments. 6. Write the plan. Generate saas-replacement-plan.md in the current working directory using the structure in references/output-template.md. Follow the rigor and quality standards in references/analysis-guidelines.md. 7. Present key findings. After writing the file, summarize total potential savings, the top 3 quick wins, any surprising findings, and the recommended first action.

Core Principle

Every SaaS subscription is a recurring tax on the business; every agent replacement is an investment in owned infrastructure that compounds over time. Make the numbers speak clearly and let the ROI make the argument. See references/analysis-guidelines.md for the full OneWave AI thesis and how to frame the analysis.

Analysis Framework

Run this full framework for every SaaS tool in the stack. Do not skip tools or give superficial analysis.

1. Tool Classification

Categorize the tool into one functional domain:

• Communication & Collaboration: Slack, Teams, Zoom, email tools
• CRM & Sales: Salesforce, HubSpot, Pipedrive, Apollo, Outreach
• Marketing & Content: Mailchimp, Buffer, Hootsuite, SEMrush, Ahrefs
• Project Management: Asana, Monday, Jira, Linear, ClickUp
• Customer Support: Zendesk, Intercom, Freshdesk, Help Scout
• Analytics & BI: Mixpanel, Amplitude, Tableau, Looker, Google Analytics
• Finance & Accounting: QuickBooks, Xero, Stripe, Brex
• HR & People: Gusto, BambooHR, Rippling, Lattice
• Developer Tools: GitHub, Vercel, AWS, Datadog, PagerDuty
• Document & Knowledge: Notion, Confluence, Google Workspace, Dropbox
• Design: Figma, Canva, Adobe CC
• Security & Compliance: Okta, 1Password, Vanta, Drata
• Data & Integration: Zapier, Make, Segment, Fivetran

2. Replacement Feasibility Assessment

Rate each tool on a four-tier scale.

FULL REPLACEMENT -- An AI agent can completely replace this tool within 3 months.

• Core functionality is primarily data processing, content generation, routing, or decision-making
• No proprietary network effects or marketplace dependencies
• API access to underlying data sources is available
• Examples: most email marketing tools, basic CRM, content scheduling, expense categorization

PARTIAL REPLACEMENT -- An AI agent can replace 50-80% of functionality, with the remainder handled by a simpler/cheaper alternative or custom integration.

• Core workflows automate, but some UI-heavy or collaborative features require a lightweight frontend
• Some integration lock-in but data is exportable
• Examples: project management (agent handles routing/updates, simple UI for boards), analytics (agent handles queries/reports, lightweight dashboard for visualization)

AUGMENTATION -- Keep the tool but layer an AI agent on top to reduce seats, automate workflows, and cut costs by 30-60%.

• The tool provides essential infrastructure or has strong network effects
• An agent can automate repetitive tasks within the tool via API
• Examples: Slack (keep, add agent for triage/routing), GitHub (keep, add agent for reviews/CI)

NOT FEASIBLE -- Replacement is not practical today due to regulatory requirements, deep platform lock-in, or infrastructure dependencies.

• Compliance mandates require the specific vendor
• The tool IS the platform (e.g., AWS for hosting, Stripe for payments processing)
• Switching cost exceeds 3-year savings
• Examples: core cloud infrastructure, payment processors, identity providers with SOC2/compliance requirements

3. Build Cost Estimation

For each FULL or PARTIAL tool, estimate the build cost.

One-Time Build Costs:

• Engineering hours (at $150/hr blended rate, adjustable)
• Claude API costs during development and testing
• Infrastructure setup (Supabase, Vercel, etc.)
• Data migration effort
• Integration development with remaining tools

Ongoing Operating Costs:

• Claude API usage (estimate tokens/month based on workflow volume)
• Infrastructure hosting (typically $20-100/month for most agent workloads)
• Maintenance engineering hours (estimate 2-4 hrs/month per agent)
• MCP server hosting if applicable

Cost Calculation Formula:

Annual SaaS Cost = (monthly_price * seats * 12)
Year 1 Agent Cost = build_cost + (monthly_operating * 12)
Year 2+ Agent Cost = monthly_operating * 12
Break-Even Month = build_cost / (monthly_saas - monthly_operating)
3-Year ROI = ((annual_saas * 3) - (year1_cost + year2_cost + year3_cost)) / (year1_cost + year2_cost + year3_cost) * 100

4. Claude + MCP Alternative Architecture

For each replaceable tool, design the agent-based alternative.

Agent Architecture:

• Claude model tier (Haiku for simple routing, Sonnet for most workflows, Opus for complex reasoning)
• System prompt and tool configuration the agent needs
• MCP servers that provide the required integrations
• Whether the agent should be autonomous, human-in-the-loop, or scheduled

Required MCP Integrations -- map each replacement to specific MCP servers:

Data Migration Path:

• How to export data from the current tool
• Where to store it (typically Supabase Postgres)
• Schema design for the replacement
• Migration timeline and rollback plan

5. Risk Assessment

Evaluate each replacement on a 1-5 scale:

• Data Loss Risk: Can all critical data be exported and preserved?
• Workflow Disruption: How much will daily workflows change?
• Team Adoption: Will the team resist the change?
• Reliability Gap: Is the agent solution as reliable as the SaaS?
• Compliance Impact: Any regulatory implications of switching?
• Vendor Lock-in Escape: How difficult is it to leave the current tool?
• Feature Gap: What capabilities are lost in the transition?

Aggregate risk score: (sum of all scores) / 35 * 100 = risk percentage.

6. Priority Matrix

Score each opportunity on two axes.

Impact Score (1-10):

• Annual cost savings (1-3 points based on dollar amount)
• Workflow improvement potential (1-3 points)
• Strategic alignment with AI-first operations (1-2 points)
• Data ownership and portability gain (1-2 points)

Effort Score (1-10):

• Engineering complexity (1-3 points, lower is easier)
• Integration dependencies (1-3 points, lower is fewer)
• Data migration complexity (1-2 points, lower is simpler)
• Team change management (1-2 points, lower is easier)

Priority Quadrants:

• Q1 Quick Wins (High Impact, Low Effort): do these first
• Q2 Strategic Bets (High Impact, High Effort): plan and resource these
• Q3 Fill-ins (Low Impact, Low Effort): do when convenient
• Q4 Reconsider (Low Impact, High Effort): probably not worth it

Analysis Guidelines & Quality Standards

Be Rigorous With Numbers

• Always show the math
• Use conservative estimates for agent costs (round up)
• Use actual SaaS pricing; do not guess, look it up via WebSearch if needed
• Add a 20% buffer to build estimates to account for implementation risk
• Include the opportunity cost of engineering time spent building replacements

Be Honest About Limitations

• Some tools genuinely cannot be replaced today; say so clearly
• Network-effect tools (Slack, GitHub) are usually AUGMENTATION, not replacement
• Compliance-critical tools need careful analysis; err on the side of caution
• If data export is not possible, flag this as a blocker
• Note when a replacement requires capabilities that do not exist yet

Be Specific About Alternatives

• Do not just say "an AI agent can do this"; describe exactly how
• Specify which MCP servers and what the agent's system prompt looks like conceptually
• Describe the user experience: how does someone interact with the replacement?
• Address the cold-start problem: what happens during migration?

Consider the Human Element

• Not every efficiency gain is worth the disruption
• Some tools are beloved by teams; factor in morale and adoption
• Training time is a real cost; include it in estimates
• Some workflows benefit from the structure a SaaS tool imposes

OneWave AI Alignment

Frame the analysis to demonstrate this thesis with real numbers from the user's own stack: 1. Most SaaS is overpriced relative to the value delivered 2. AI agents can replicate core SaaS functionality at 10-30% of the cost 3. Custom agents provide better integration and flexibility than off-the-shelf SaaS 4. The shift from SaaS to agents is inevitable; early movers gain competitive advantage 5. Data ownership returns to the company when SaaS is replaced with agents 6. Agent-based systems compound in value as they learn from your data

Quality Standards

The output saas-replacement-plan.md must meet these standards: 1. Every tool analyzed: no tool in the input list is skipped 2. Numbers add up: all financial projections are internally consistent 3. Actionable specificity: every recommendation includes concrete next steps 4. Honest assessment: clearly flag tools that should NOT be replaced 5. Complete architecture: each replacement includes enough technical detail to begin implementation 6. Risk transparency: all material risks are identified with mitigation strategies 7. Timeline realism: the implementation timeline accounts for actual engineering capacity 8. Stakeholder ready: the document is presentable to a CEO, CTO, or CFO without additional formatting

Output Template

Generate a saas-replacement-plan.md file in the current working directory using this structure. The document must be comprehensive, well-formatted, and ready to present to a CEO, CTO, or CFO without additional formatting.

# SaaS Replacement Plan
Generated: [date]
Prepared for: [company/user name if provided]

## Executive Summary

**Current Annual SaaS Spend**: $XX,XXX
**Projected Year 1 Spend (with replacements)**: $XX,XXX
**Projected Year 2+ Annual Spend**: $XX,XXX
**3-Year Net Savings**: $XX,XXX
**Number of Tools Analyzed**: XX
**Recommended for Full Replacement**: XX
**Recommended for Partial Replacement**: XX
**Recommended for Augmentation**: XX
**Not Feasible to Replace**: XX

## Current SaaS Stack Overview

| Tool | Category | Monthly Cost | Annual Cost | Seats | Primary Use |
|------|----------|-------------|-------------|-------|-------------|
| ... | ... | ... | ... | ... | ... |

**Total Monthly Spend**: $X,XXX
**Total Annual Spend**: $XX,XXX

## Priority Matrix

### Q1 -- Quick Wins (Do First)
[Tools with high impact, low effort -- start here]

### Q2 -- Strategic Bets (Plan Next)
[Tools with high impact, high effort -- resource and schedule]

### Q3 -- Fill-ins (When Convenient)
[Tools with low impact, low effort -- batch these together]

### Q4 -- Reconsider (Probably Skip)
[Tools with low impact, high effort -- not worth it now]

## Detailed Replacement Analysis

### [Tool Name] -- [FULL/PARTIAL/AUGMENTATION/NOT FEASIBLE]

**Current Cost**: $XXX/month ($X,XXX/year) for X seats
**Category**: [category]
**Feasibility**: [rating with justification]

**Replacement Architecture:**
- Agent Type: [autonomous/human-in-loop/scheduled]
- Model: [Haiku/Sonnet/Opus]
- MCP Integrations: [list]
- Data Store: [e.g., Supabase Postgres]

**Build Estimate:**
| Item | Cost |
|------|------|
| Engineering (XX hours) | $X,XXX |
| Infrastructure Setup | $XXX |
| Data Migration | $XXX |
| **Total One-Time** | **$X,XXX** |

**Monthly Operating Cost**: $XXX
- Claude API: $XX
- Infrastructure: $XX
- Maintenance: $XX

**ROI Analysis:**
- Monthly Savings: $XXX
- Break-Even: Month X
- Year 1 Net: +/- $X,XXX
- 3-Year Net Savings: $XX,XXX
- 3-Year ROI: XXX%

**Risk Assessment:**
| Risk Factor | Score (1-5) | Notes |
|-------------|-------------|-------|
| Data Loss | X | ... |
| Workflow Disruption | X | ... |
| Team Adoption | X | ... |
| Reliability | X | ... |
| Compliance | X | ... |
| Vendor Lock-in | X | ... |
| Feature Gap | X | ... |
| **Aggregate Risk** | **XX%** | |

**Implementation Steps:**
1. [Step with timeline]
2. [Step with timeline]
3. [Step with timeline]

[Repeat for each tool]

## Implementation Timeline

### Phase 1: Quick Wins (Weeks 1-4)
- [Tool replacements with specific milestones]

### Phase 2: Strategic Replacements (Months 2-4)
- [Tool replacements with specific milestones]

### Phase 3: Optimization & Augmentation (Months 4-6)
- [Remaining replacements and augmentations]

### Phase 4: Review & Iterate (Month 6+)
- [Performance review, cost validation, iteration]

## Financial Summary

### Cost Comparison Table

| Tool | Current Annual | Year 1 (Build+Run) | Year 2+ Annual | 3-Year Savings |
|------|---------------|--------------------|--------------|----|
| ... | ... | ... | ... | ... |
| **TOTALS** | **$XX,XXX** | **$XX,XXX** | **$XX,XXX** | **$XX,XXX** |

### Savings Trajectory

- **Month 1-3**: Net investment period (building agents)
- **Month 4-6**: Break-even on quick wins
- **Month 7-12**: Cumulative savings begin
- **Year 2**: Full savings realized
- **Year 3**: Maximum ROI achieved

### Investment Required

- **Total One-Time Build Cost**: $XX,XXX
- **Monthly Operating (all agents)**: $X,XXX
- **Annual Operating**: $XX,XXX
- **Payback Period**: X months

## Risk Mitigation Strategy

### High-Risk Replacements
[Tools with aggregate risk > 60% -- detailed mitigation plans]

### Rollback Plans
[For each Phase 1-2 replacement, document how to revert]

### Parallel Running Period
[Recommend running old and new systems simultaneously for X weeks per tool]

### Monitoring & Validation
[KPIs to track for each replacement to ensure quality parity]

## Technical Architecture

### Agent Infrastructure
- **Runtime**: Claude Code / Claude API
- **Database**: Supabase (Postgres)
- **Hosting**: Vercel (Edge Functions for lightweight agents)
- **Orchestration**: MCP protocol for tool integration
- **Monitoring**: [Recommended approach]

### MCP Server Requirements
[List all MCP servers needed across all replacements]

### Data Architecture
[How data flows between agents and storage]

## Recommendations

### Immediate Actions (This Week)
1. [Specific action]
2. [Specific action]
3. [Specific action]

### 30-Day Goals
1. [Specific goal with measurable outcome]
2. [Specific goal with measurable outcome]

### 90-Day Goals
1. [Specific goal with measurable outcome]
2. [Specific goal with measurable outcome]

## Appendix

### Methodology Notes
[Assumptions, rate cards, estimation approach]

### Tool-Specific Research
[Links, documentation, API availability notes per tool]

### Glossary
- **MCP**: Model Context Protocol -- standard for connecting AI models to external tools and data
- **Agent**: An AI system that can take actions autonomously via tool use
- **Human-in-the-Loop**: Agent that drafts actions for human approval before execution

Common SaaS Replacement Patterns

Use these proven patterns to accelerate analysis. Match each tool to the closest pattern, then adapt.

Email Marketing (Mailchimp, ConvertKit)

• Replacement: Claude agent + Gmail MCP + Supabase for subscriber management
• Why it works: email marketing is content generation + list management + scheduling, all agent-native tasks
• Typical savings: 70-90%

Basic CRM (Pipedrive, HubSpot Starter)

• Replacement: Claude agent + Supabase (contacts/deals tables) + Gmail MCP + Google Calendar MCP
• Why it works: CRM at its core is a database with workflow automation; agents excel at both
• Typical savings: 60-80%

Content Scheduling (Buffer, Hootsuite)

• Replacement: Claude agent + platform APIs + Supabase for content calendar
• Why it works: content scheduling is API calls on a timer with some content generation
• Typical savings: 80-95%

Help Desk (Zendesk, Intercom basic)

• Replacement: Claude agent + email/chat integration + Supabase knowledge base
• Why it works: most support tickets are repetitive and can be handled or triaged by an agent
• Typical savings: 50-70%

Expense Management (Expensify, Ramp basic)

• Replacement: Claude agent + bank API + Supabase + receipt OCR
• Why it works: categorization and policy checking are pattern-matching tasks agents handle well
• Typical savings: 60-80%

Meeting Scheduling (Calendly, SavvyCal)

• Replacement: Claude agent + Google Calendar MCP + email
• Why it works: availability checking and scheduling is a well-defined agent task
• Typical savings: 90-100%

Survey/Forms (Typeform, SurveyMonkey)

• Replacement: Claude agent + conversational interface + Supabase
• Why it works: an agent can conduct dynamic surveys that adapt in real time, better than static forms
• Typical savings: 80-95%

Analytics Reporting (basic BI tools)

• Replacement: Claude agent + Supabase (direct SQL) + scheduled reports via Slack/email
• Why it works: most analytics requests are natural-language queries against structured data
• Typical savings: 60-80%

Workflow Automation (Zapier, Make)

• Replacement: Claude agent with MCP integrations + Supabase Edge Functions
• Why it works: agents handle conditional logic, error handling, and complex routing better than visual workflow builders
• Typical savings: 70-90%

Document Generation (PandaDoc, DocuSign basic)

• Replacement: Claude agent + document templates + email for delivery
• Why it works: document assembly from templates is a core language model capability
• Note: e-signatures still require a specialized service, so this is a PARTIAL replacement
• Typical savings: 40-60%

Edge Cases and Nuances

Tools With Network Effects

Slack, GitHub, Figma, and similar tools derive value from being where everyone already is. These are almost never full replacements. The play is AUGMENTATION: add agents that reduce time spent in these tools and cut the number of paid seats needed.

Compliance-Mandated Tools

If a tool is required for SOC2, HIPAA, or similar compliance, mark replacement NOT FEASIBLE unless the replacement can be certified. Document this clearly and do not recommend risky transitions.

Tools With Proprietary Data Formats

Some tools lock data in proprietary formats. If export is limited or lossy, this significantly increases migration risk and cost. Flag these explicitly.

Free Tier Tools

If a tool is on a free tier, replacement may not save money but could still be worth it for integration benefits, data ownership, or reduced complexity. Analyze these separately.

Multi-Tool Bundles

Google Workspace, Microsoft 365, and similar bundles often cost less per-tool than the sum of individual replacements. Analyze bundles holistically, not tool-by-tool.

API Rate Limits

When estimating Claude API costs, account for volume carefully. A tool processing 10,000 customer support tickets per month has materially different API costs than one handling 100.