Cut OpenAI API Costs in EdTech With Fine-Tuning
EdTech platforms pay OpenAI on every tutoring, feedback and quiz API call. Here is the cost math for serving the same volume from a fine-tuned local model.
Read the articleNotes from
the work.
Guides, research notes and what we learn building models.
Ertas Builder Plan ($25/mo): Who It's For, What You Get, Honest Review
An honest breakdown of the Ertas Builder plan. What's included, who it's right for, when to upgrade, and how the pricing compares to the API costs it replaces.
Read the article
Ertas Pro Plan Review ($50/mo): 200 Credits, 100 GB
An honest review of the Ertas Pro plan. What's included, who it's for, the ROI math for small AI agencies, and when to upgrade to Business.
Read the article
Ertas vs HuggingFace AutoTrain: Visual Fine-Tuning Without the YAML Configs
Comparing Ertas and HuggingFace AutoTrain for no-code LLM fine-tuning. Covers workflow UX, GGUF export, local deployment, pricing, and dataset format differences.
Read the article
Ertas vs Modal Labs: Which Is Better for Agencies Fine-Tuning Client Models?
Comparing Ertas and Modal Labs for AI agency fine-tuning workflows. Covers the GUI vs code divide, multi-client management, cost predictability, and GGUF deployment.
Read the article
Ertas vs Replicate for Fine-Tuning: Cost, Workflow, and GGUF Export Compared
Side-by-side comparison of Ertas and Replicate for fine-tuning language models. Covers workflow, pricing, GGUF export, data privacy, and when to choose each platform.
Read the articleErtas vs Together AI: Fine-Tuning Costs, Local Deployment, and Data Privacy
Comparing Ertas and Together AI for fine-tuning language models. Covers per-token vs flat-cost inference, data privacy, local deployment, and when each platform wins.
Read the articleFine-Tune a Listing Description AI for Real Estate: Step-by-Step
Real estate agents spend 30-45 minutes writing each listing. A fine-tuned model trained on the brokerage's own listings generates on-brand descriptions in 2 minutes. Here's how to build it.
Read the articleFine-Tune a Product Recommendation Model for E-Commerce: Full Walkthrough
Generic recommendation engines miss semantic product relationships. Here's how to fine-tune a model on your catalog and purchase history to build recommendations that increase average order value.
Read the articleFine-Tuned Copywriting Models for Agency Clients: Ad Copy That Actually Converts
Generic AI ad copy is mediocre because it has never seen your client's winning variants. A copywriting model trained on historical performance data generates copy calibrated to convert — not just grammatically correct.
Read the articleThe Fine-Tuned Model Is the Cheapest AI Moat You Can Build
Distribution moats cost millions. Network effect moats require years. A fine-tuned model moat costs $25/month and 4 hours. Here's the math on why this is the most accessible competitive advantage in software.
Read the articleFine-Tune a Tutoring AI for EdTech: Subject-Specific Models That Don't Hallucinate Curriculum
Generic AI tutors hallucinate curriculum and use inconsistent methodology. A fine-tuned model trained on your course content tutors in your pedagogy, at your difficulty level, without inventing facts.
Read the articleFunded Startup vs Vibecoder: Why the Solo Builder Wins on AI in 2026
Conventional wisdom says funded AI startups beat solo builders. For specific AI product types in 2026, this is wrong. Here's where vibecoders have a structural advantage over well-funded teams.
Read the articleLangChain + Ollama: Fine-Tuned Local Models
Swap the OpenAI backend in your LangChain pipelines for a fine-tuned local model on Ollama. Same chain code, two config lines, zero per-token costs.
Read the articleMCP + Fine-Tuned Local Model: Connect Claude to Your Domain-Specific AI
Model Context Protocol (MCP) lets Claude Desktop talk to any server — including your own Ollama-hosted fine-tuned model. Here's the architecture and setup for routing Claude requests to a custom domain model.
Read the articleMCP Servers + Local Models: Zero API Costs for Domain-Specific AI Tools
The combination of MCP servers and fine-tuned local models eliminates per-token costs for AI tools built on Claude, Cursor, and other MCP-compatible clients. Here's the cost math and the architecture.
Read the articleMCP Tools for AI Agency Client Workflows: Deliver Models as Tools, Not Files
AI agencies typically deliver a model file. With MCP, you can deliver a Claude Desktop or Cursor tool that your client uses daily — recurring value that justifies a recurring retainer.
Read the articleMicro-SaaS AI Moat: Why Small Apps Benefit Most From Fine-Tuning
Micro-SaaS founders often assume fine-tuning is for funded startups with ML teams. It is not. Small apps with focused use cases and real user data are the ideal fine-tuning candidate — and the moat compounds fastest.
Read the articleFrom $500/Month OpenAI Bills to $0: Migrating n8n Workflows to Local Models
A practical migration guide for n8n users spending hundreds on OpenAI API calls. Move your workflows to local fine-tuned models without breaking anything.
Read the articleNiche AI Agency vs Generalist: Which Wins Clients in 2026
The data is clear: niche AI agencies close faster, charge more, and retain clients longer. Here's why niching works in AI specifically and how to find yours.
Read the articleOllama's OpenAI-Compatible API: Drop Your Fine-Tuned Model Into Any OpenAI Integration
Ollama exposes an OpenAI-compatible REST API. Any code written for the OpenAI SDK — Langchain, LlamaIndex, your own app — works with your fine-tuned local model by changing one URL. Here's what to know.
Read the articleReal Estate CRM AI Assistant: Fine-Tune a Follow-Up Model on Agent Communication
Real estate agents lose deals by failing to follow up. A fine-tuned model trained on the agent's own communication history drafts personalized follow-ups in the agent's voice — reducing relationship maintenance from 6 hours to 45 minutes per month.
Read the articleReal Estate Lead Qualification AI: Fine-Tune a Scoring Model on Your Conversion History
Real estate teams waste hours on low-intent leads. A fine-tuned classifier trained on your closed and dead leads automatically scores inbound leads so agents focus on the ones that close.
Read the articleReplit App AI Costs Exploding? Replace OpenAI with a Fine-Tuned Local Model
Replit's always-on deployment and easy AI integration create a specific API cost problem. Here's how to replace OpenAI with a fine-tuned local model and cut costs to flat rate.
Read the article