AI agent implementation · flat fee · four weeks

We build the AI
your team can't hire for.

Done-for-you AI agents shipped to your repo in 2 to 6 weeks. Flat fee from $15k. The code is yours. Your inbox is the only meeting.

Brief
Forge
Review
Ship
Send a brief →See a 4-week build, day by day
The Forge contract

One technical partner per engagement. Async by default. Four weeks from discovery to deploy. The agent ships on time, or we extend support free until it does. Code is yours on day twenty-eight. Nothing about that is negotiable — that's the contract.

Forge by AI Toolery Lab · Calgary, AB · forge@aitoolerylab.com
MIT-licensed codeYours, no patents
No callhomeNo telemetry or tracking
GDPR-friendlyEU data stays in EU
Code-yours on day 28Repo transferred at deploy
Calgary AB jurisdictionService Agreement v2
The honest comparison

Four ways to ship the AI feature your team needs. Here's the trade-off, side by side.

Founders evaluating Forge are usually weighing four options. We've laid them out so the choice doesn't require a procurement spreadsheet.

ForgeHire MLESaaS platformAgency
Time to first agent in production4 weeks90 days (find) + 30 days (ramp)1-3 hours of self-config (then maintain forever)8-16 weeks (T&M, scope-creep)
Cost shapeFlat fee $15-50k$220-340k all-in / year + equity dilutionPer-call subscription, scales with usageHourly $200-500, scope-creep risk
Code ownershipYours, in your GitHub on day 28YoursVendor's. Their stack, their lock-inNegotiable, often vendor-side
In-house ML expertise requiredNoYes — that IS the hireSome (prompt engineering / connectors)No — they bring it (and bill it)
Recurring meetings during buildZero. Async only1:1s, sprint planning, retros — every weekSelf-serve (no meetings)Tuesday standup + bi-weekly review
Vendor lock-inNone — code-yoursNoneHigh. Migration cost is your problemVariable. Often retainer-tail
Maintenance after go-live7-30 day support + Care add-on $200/mo (optional)Their job (and salary continues)Vendor's job (and subscription continues)Retainer required, hourly otherwise
Compliance evidence (FCRA / EU AI Act / LL144)Optional AttestProto bundle (+0$)They build it (more salary)Vendor-controlled, variesAdd-on engagement
Outcome guarantee≥80% success criteria or support extends freeNo formal guaranteeSLA + uptime, no outcomeRare — usually scope-completion only
The honest problem

You don't have an ML hire and you can't justify one.

A capable AI engineer in 2026 costs $220-340k all-in. The agent feature you need is one or two well-scoped builds — not a full-time hire. The platforms charge per-call forever. The agencies want retainers. You want the agent shipped, the code in your repo, and your team able to maintain it.

1

Hire an MLE

$220-340k all-in. 60-90 days to find, 30 days to ramp. Cap-table dilution if you give equity. Wrong shape for one or two well-defined builds.

2

Subscribe to a platform

Per-call billing forever. Vendor lock-in. Your data sits in their infrastructure. The day you outgrow it, the migration is your problem.

3

Hire an agency

Retainers. T&M billing. Scope creep. Calls every Tuesday. The agent works at handover and starts breaking three weeks later when nobody on your side knows the codebase.

Forge is the fourth option: one technical partner, flat fee, four weeks, code yours.

The four-week build

What happens between contract signature and your agent in production.

Same arc for every engagement. Asynchronous by default — daily Slack updates, two scheduled checkpoints, no recurring meetings. You ship code with your team while we ship code into your stack.

Week 1

Discovery + signed Statement of Work

Workflow shadowing on a recorded screen-share. Test data review. Exhibit A drafted and signed by Friday. The scope is binding from there — additions become Phase 2 contracts, not surprise-bill line items.

Week 2

Agent core + first integration

Agent built into your stack against scoped credentials. End-of-week demo on the real workflow. You see the artefact, not slides.

Week 3

Remaining integrations + sandbox tuning

Tuned to ≥80% success rate on the agreed test set. End-of-week demo on production-shaped data. Edge cases logged for week 4 polish.

Week 4

Production deploy + handoff

Deploy to your infrastructure. Code repo transferred to your GitHub. Runbook + monitoring dashboard delivered. 7-30 days of free post-launch support depending on tier. Done.

What's different

Three commitments most agencies and platforms refuse to make.

Code yours

Repo handover on day 28

The agent code lives in your GitHub at deploy. No portion of it stays behind a Forge platform. If we vanish tomorrow, your team owns enough to maintain it.

Flat fee

One number on the contract

$15k, $30k, or $50k — picked at signing, not a moving target. We absorb scope risk inside the tier. Out-of-tier asks become Phase 2 conversations, not invoice surprises.

Async only

Zero recurring meetings

Daily Slack standups in a private channel. Two scheduled checkpoints (Discovery + Production). Your founder time stays on your product, not on our calendar.

Estimator

See your tier and timeline before you write a brief.

No form gates, no "talk to sales" maze. Tell us four things about your build and we'll suggest the tier that fits — open the recommendation as a pre-filled brief.

ROI calculator

Will Forge pay for itself? — 4 sliders.

Plug in the actual hours your team spends on the workflow, the loaded hourly cost, and the automation target. Calculator computes annual savings, payback period, and 5-year ROI in real time.

15 hrs/week
$75/hr
Salary + benefits ÷ working hours. Engineers ~$120-180. Ops ~$50-90. Founders ~$200-400.
70%
Forge agents typically reach 70-85% automation. Higher needs more edge cases — that's a Pro tier signal.
Annual savings
$40,950
546 hours/year
Payback period
8.8 mo
build cost ÷ annual savings
3-year net
$92,850
savings minus build cost
5-year ROI
583%
return on Forge investment

Numbers update live. Defaults: 15 h/week, $75/hr ops cost, Standard tier, 70% automation. Adjust to your actual workflow.

Adoption forecaster

Will your team actually use it? — model it before you commission.

The single biggest reason AI builds fail isn't the model — it's that the team never adopts the tool. Model your situation across the five variables that predict 12-week adoption from real Forge engagements. If the score is red, we'll refuse the engagement until it improves. We don't ship code that won't get used.

20 people

12-week adoption
58%
Watch closely
20%
W1
32%
W2
45%
W4
53%
W8
58%
W12
What this means: Adoption is reachable but not automatic. We add a 30-day adoption check-in (free) and recommend a named KPI owner.

Calibrated against post-launch data from Forge engagements + published BCG / McKinsey enterprise AI adoption studies (2024-2026). Not a guarantee — a planning tool.

Selected work

What a finished build looks like.

Reference engagements that ship the patterns Forge specialises in. Each card opens the brief — including tier, timeline, and the metric the customer measured against.

open-source

Three template skeletons. MIT licensed. Fork or hire — your call.

Reference agent skeletons for the integrations Forge gets asked about most. Each runs end-to-end with one API key. DIY in 1-2 weeks, or commission the production-grade version.

These templates are MIT-licensed reference skeletons. Fork them, learn from them, build with them. If you'd rather have us run a 4-week production-grade build on top — that's the Forge service. Either way you win.

Pricing — visible

Three tiers. Pick at signing. No "talk to sales" maze.

Sized to ship the actual thing your team needs in the time you actually have. Pricing is fixed and public on purpose — your time is too valuable to spend on procurement theatre.

✓ Code yours on day 28✓ Outcome guarantee in writing✓ 50% deposit, 50% on deploy✓ MIT-licensed integrations
Starter
$15k
2 weeks · 1 use case
  • 1 functional AI agent
  • 2 integrations of your choice
  • Basic monitoring (logs)
  • 1-hour training session
  • 7-day post-launch support
  • Code repo transferred to your GitHub
Start the conversation
Pro
$50k
6 weeks · multi-agent or custom UI
  • Up to 2 agents OR 1 complex single agent
  • Up to 8 integrations
  • 1 custom UI surface (basic)
  • Advanced monitoring + alerting
  • 4-hour training session
  • 30-day post-launch support
  • Phase 2 scope-of-work draft included

Beyond this scope — multi-agent meshes, full frontend rebuilds, multi-vertical compliance overlays — quote separately, from $75k.

Start the conversation

50% deposit at signing, 50% at production deploy. Stripe Connect or wire. Service contract template available on request before signing.

The commitment

If the agent doesn't meet success criteria during the support window, support extends free until it does — or 60 days, whichever first.

Success criteria are written into Exhibit A in week 1. They're concrete, testable, and agreed by both sides before code starts. We don't ship vague.

MIT-licensed code on handoverAsync-only deliveryNo recurring feesCalgary, AB jurisdiction

Industrial craft for the AI era. Your agent ships in four weeks — code in your repo, nothing left behind on day twenty-eight.

Forge · AI Toolery Lab · Calgary, AB
FAQ

What founders ask before signing.

Why flat fee, not hourly?
Hourly creates the wrong incentives. A flat fee means we're motivated to ship fast and well — if something takes longer than estimated, that's our problem to solve, not your invoice line item. It also means you can budget the project before signing, instead of watching meter run.
What if the agent doesn't work?
Success criteria are agreed in writing during week 1. They're concrete (e.g. "agent handles 80% of qualifying tickets without escalation across a 100-ticket holdout set"). If the agent doesn't meet them in the support window, we extend support free until it does or 60 days elapse, whichever first.
Can we pay after delivery?
No. 50% deposit at signing is firm. It's the discipline test — if the project can't justify 50% of the budget at the start, it isn't ready to start. Balance is due at production deploy, not before.
Do you take calls during the build?
Async by default. Daily Slack updates in a private channel, two scheduled live checkpoints (Discovery review and Production deploy), unlimited written Q&A in between. If something blocks the build, we hop on a 15-minute call. No recurring meetings.
What if our team can't maintain the code afterward?
Then we're a poor fit. Our model assumes your team has at least one engineer who can read Python or TypeScript and operate Git. If you don't, you'd be better served by a managed-platform vendor, and we'll happily point you at honest options.
Do you sign NDAs?
Yes. Standard mutual NDA at the start of every Discovery week if you want one — we have a template ready, or you can use yours. The service agreement also covers confidentiality for 24 months post-engagement.
Who actually does the building?
A focused two-person engineering team — twelve+ years of production systems experience and the last three years on AI agent infrastructure. Each engagement is staffed by a single technical lead end-to-end. No subcontractors without your written approval.

Pick a tier, send the SOW, and we kick off Monday.

Reply with the use case in plain English. I'll respond within four hours with a service-agreement template and one calendar slot for a 30-minute discovery call.

Start your build →Also building AttestProto ↗