Nixfred AI Systems Workbench
Practical instruments for reliable AI systems.
Each instrument takes real inputs and returns a result you can inspect, edit, and export. Sample data is included. No account, and no API key for the core experience.
- Registered
- 14
- Available now
- 0
- Categories
- 5
Showing 14 of 14 instruments.
First release queue
Nothing here is usable yet. These 3 instruments are being built first, in this order.
-
Prompt Laboratory
See why a small wording change flips model behavior, by comparing prompt structures side by side.
Runs locally, input stays on device Not yet available -
Token & Cost Planner
Estimate context usage, request cost, and monthly spend with assumptions you can edit and inspect.
Runs locally, input stays on device Not yet available -
Context Packer
Decide what survives a finite context window, and see exactly what gets dropped or compressed.
Runs locally, input stays on device Not yet available
Catalog in build
14 instruments are specified and queued. None can be opened yet. Each one ships when its own specification is built and accepted, not before.
-
Prompt Laboratory
See why a small wording change flips model behavior, by comparing prompt structures side by side.
Runs locally, input stays on device Not yet available -
Token & Cost Planner
Estimate context usage, request cost, and monthly spend with assumptions you can edit and inspect.
Runs locally, input stays on device Not yet available -
Context Packer
Decide what survives a finite context window, and see exactly what gets dropped or compressed.
Runs locally, input stays on device Not yet available -
Retrieval Lab
Watch how chunking, embedding, and ranking choices change what a retrieval system actually returns.
Runs locally, input stays on device Not yet available -
Model Selector
Match a model to a workload using stated constraints instead of vibes or leaderboard rank.
Runs locally, input stays on device Not yet available -
Evaluation Workbench
Design an evaluation that would actually catch the failure you are worried about.
Runs locally, input stays on device Not yet available -
Workflow Decomposer
Break a fuzzy business process into steps a system can actually automate, and mark the ones it cannot.
Runs locally, input stays on device Not yet available -
Agent Designer
Specify an agent completely enough to build it, including its tools, limits, and escalation paths.
Runs locally, input stays on device Not yet available -
Permission Planner
Give an agent the narrowest authority that still lets it finish the job.
Runs locally, input stays on device Not yet available -
Latency Budgeter
Find out which stage of your pipeline spends the most user patience, before you ship it.
Runs locally, input stays on device Not yet available -
Failure Investigator
Work backward from a bad output to the mechanism that produced it, then write the fix down.
Runs locally, input stays on device Not yet available -
Drift Monitor
Tell a real behavior change apart from ordinary sampling noise before you go chasing it.
Runs locally, input stays on device Not yet available -
Signal Tester
Check whether your evaluation signal measures the thing you care about, or something adjacent to it.
Runs locally, input stays on device Not yet available -
AI Stack Mapper
Draw the whole path a request takes, so you can explain the system to someone who has to trust it.
Runs locally, input stays on device Not yet available
No instrument matches that search and filter combination.
Local by default
Every instrument runs in your browser. What you type stays on your device, and there is no server here to send it to. No accounts, no tracking, no analytics.
If an instrument ever needs to send something somewhere, it says so on its own page, before you use it, and it tells you exactly what leaves.
About this workbench
A workbench of small, sharp instruments for the parts of AI system work that are easy to get wrong. It is not a chat product and not a news feed. Read what it is and what it is not on the about page.
Built by Fred Nix. More at nixfred.com.