Ertas Agency Plan ($69.50/mo Early Bird): 10 Client Projects, 5 Seats — Is It Worth It?

The Ertas Agency plan is $69.50 per month during Early Bird (normally $149/month, locked for life on pre-subscription). It is designed for small AI agencies and consultants managing fine-tuning projects for multiple clients simultaneously.

This is an honest breakdown of what the plan includes, who it is designed for, how the ROI math works, and when it is not enough.

What's Included in Agency

400 credits per month. Significantly more than Builder's 100. For an agency running 8-10 clients with monthly retraining cycles, 400 credits provides comfortable headroom (roughly 25-50 training runs per month depending on dataset size and model).

10 client-labeled projects. This is the core Agency feature. Each project is isolated: datasets, training runs, model versions, and GGUF exports are separated per client. Your client A data cannot be accidentally included in client B training. Each project has its own name (your client's name or project identifier) visible in the interface.

5 seats. Five team members can access the Agency account with role-appropriate permissions. Typical small agency use: account owner + 1-2 developers + 1-2 account managers who run evaluations or check training progress.

3 concurrent training jobs. Run three client models simultaneously instead of queuing them. At 8-10 clients with overlapping retraining schedules, this is the difference between delivering on time and managing an awkward queue.

200 GB storage. Enough for 30-40 exported GGUF models plus all their training datasets. A 7B Q4 GGUF is ~4-5 GB; a 14B Q4 GGUF is ~8-9 GB.

Higher-tier GPUs. Agency clients get access to better GPU hardware for training, which means faster training runs and the ability to handle larger datasets.

Dataset synthesis and bulk evaluation. Same as Builder — generate synthetic training data, run bulk evaluations on test sets, use auto-evals to catch quality regressions.

The Per-Client Architecture

Understanding the per-client project model is important for evaluating whether Agency fits your workflow.

Each of your 10 project slots represents one client (or one major use case within a client). Inside each project:

• All dataset versions are stored (so you can retrain from any historical snapshot)
• All training runs are logged with parameters and results
• All exported GGUF files are available for download
• Evaluation results are tracked over time

The LoRA adapter model means that each client's fine-tuned model is a lightweight adapter (50-200 MB) on top of a shared base model. Clients share base model weights but have entirely separate adapters. This is how you manage 10 client models at 1/10th the storage cost of 10 full model copies.

Who Agency Is Right For

AI agency with 3-10 active client fine-tuning projects. You are delivering custom fine-tuned models to clients, managing ongoing retraining cycles, and need project isolation, team access, and concurrent job slots. This is exactly the Agency plan use case.

Freelance AI consultant with concurrent client work. You work independently but juggle 4-6 clients simultaneously. The 5 seats let you add a contract developer or assistant. Per-client projects keep your client work clean.

Small internal AI team building models for multiple departments. Marketing model, sales model, support model — each needs separate training data and separate GGUF output. Agency gives you the structure to manage this without cross-contamination.

The ROI Calculation

The question is not whether $69.50/month is expensive — it is whether it is expensive relative to what it enables you to bill.

Scenario 1: 5 clients at $500/month retainer each

• Monthly revenue: $2,500
• Agency plan cost: $69.50
• Plan as % of revenue: 2.8%
• For every client you add beyond 1, the plan cost is increasingly trivial

Scenario 2: 3 clients at $800/month project fee

• Monthly revenue: $2,400
• Agency plan cost: $69.50
• Plan as % of revenue: 2.9%

Scenario 3: 10 clients at $300/month retainer (model maintenance only)

• Monthly revenue: $3,000
• Agency plan cost: $69.50
• Plan as % of revenue: 2.3%

The Agency plan pays for itself with a single client paying $70/month or more. Any agency with more than one active fine-tuning client should be on Agency — the question is not whether to, but when.

What Each Client Actually Needs (and What Agency Delivers)

Comparing the Cost to DIY Infrastructure

Some agencies consider building their own fine-tuning infrastructure instead of paying for Agency. Here is the honest comparison:

DIY may make sense for very large agencies with dedicated ML engineers. For small agencies and consultants, paying $69.50/month for infrastructure that would otherwise cost hundreds in developer time is a straightforward decision.

Agency vs Agency Pro: When to Upgrade

Agency Pro is $169/month (Early Bird; normally $349/month). The upgrade unlocks:

• 30 client projects (vs 10)
• 15 seats (vs 5)
• 8 concurrent jobs (vs 3)
• 1,000 credits/month (vs 400)
• White-label API (your clients can call the model through your branded API endpoint)
• Unlimited dataset synthesis and evaluation

Upgrade to Agency Pro when:

• You have more than 10 active client projects
• You need white-label API delivery (clients call your branded endpoint, not Ertas)
• You have a team larger than 5 people
• You consistently run into the 3 concurrent job limit

Stay on Agency when:

• You have 10 or fewer active clients
• You deliver GGUF files directly (no API endpoint needed)
• Your team is 5 people or fewer
• 3 concurrent jobs is sufficient for your retraining schedule

The Early Bird Lock-In

Agency at $69.50/month (Early Bird) vs $149/month (standard).

• Savings over 12 months: $948
• Savings over 24 months: $1,896
• Savings over 36 months: $2,844

For a 2-person agency running for 3 years, the difference between locking in early bird and paying standard pricing is nearly $2,900. This is not promotional language — it is the arithmetic of a 53% price difference compounded over time.

Pre-subscription means monthly billing, cancel anytime, refund guarantee. The lifetime price lock applies as long as the subscription remains active.

How to Know You Are Ready for Agency

You are ready for Agency if:

• You have or expect to have 3+ active client fine-tuning projects
• You need more than one team member to access the platform
• You are delivering fine-tuned models as part of a paid service
• You are billing clients for AI automation work that involves custom models

If you have only 1-2 projects and work solo, Builder ($14.50/month) may be sufficient. The upgrade to Agency makes sense when client project count exceeds 3 or when you need team access.

---

Further Reading

• Ertas Builder Plan Review — Is the Builder plan right before upgrading to Agency?
• Agency AI Cost Reduction — How agencies cut costs 99%+ with fine-tuned local models
• Manage Multiple Fine-Tuned Models — Operational workflows for multi-client model management
• White-Label AI Platform for Agencies — When white-labeling becomes important
• Recurring Revenue with Fine-Tuned Models — Building retainer income around model maintenance