Model selection and cost optimization
Which model, at which size, for which step — chosen by measured quality and cost on your own tasks rather than by the leaderboard.
What you get
Each deliverable names the format it arrives in. These are the formats the stack produces today.
Evaluation set
From your own tasks
Per-step model recommendation
Document
Cost model
Spreadsheet
Routing rules
Configuration
How it works
- 1
Build the evaluation set
Real tasks from your workflow, scored the same way every run.
- 2
Measure candidates
Frontier, small and open-weight models, per step.
- 3
Recommend and route
The cheapest model that clears the quality bar, per step, written into routing rules.
What runs where
On the platform
- The platform's evaluation harness is the starting point
Built for you
- Your evaluation set, cost model and routing rules
What stays under your control
The promises that hold on every engagement, and the ones we do not make on this one.
On every engagement
- Outputs cite the document or conversation they came from, so a person can check before acting.
- A person approves anything that reaches a customer or writes to a system of record.
- Delivery is Slack, email or a document you own. Other channels are not offered.
- Fixed price, agreed before we start. Nothing numeric is published on this site.
The work runs on the TurnkeyGTM platform where it fits, and on plain software where it does not.
Not promised here
- Figures are per engagement and are not published on the page.
- Recommendations are revisited when models change; the set makes that cheap.
Questions we get asked
More in automation & engineering
AI workflow automation
The workflow your team repeats every week, automated with AI where it helps and plain software where it does not, run beside your current process until you trust it.
See the deliverablesCustom hosted and on-premise AI
AI deployments designed for where your data has to live — a dedicated hosted environment or your own hardware — on open-weight models and self-hostable components.
See the deliverablesAI output evaluation and quality gates
Frozen scorers and checks that decide whether an AI output is good enough to ship, so quality is measured rather than eyeballed.
See the deliverablesHow an engagement runs
One call to scope it, one fixed price to approve, one build you can trust before it goes live, then a handover.
The full process- 1
Scope the work
A thirty-minute call on what your team does by hand every week, where the information lives, and who needs the output.
- 2
Fixed price, ROI up front
You get a written proposal with a fixed price and an honest estimate of the hours it gives back.
- 3
Build beside your process
We design and build the automation and run it next to the manual version until the outputs match and you trust them.
- 4
Hand over, keep it running
Documentation and runbooks so your team owns it.
Tell us where this fits in your week.
A thirty-minute call is enough to scope it. Fixed price, agreed before we start.