Chat and text
Slower answers, larger bills
Every "where is my order" pays for a model round trip that decides exactly what it decided last time.
The verified fast path for AI agents
Wont learns the tool calls your chat, voice and back-office agents make again and again, proves each one against your model, and then answers it instantly. Anything new, and every write, still goes to your model.
Works withOpenAIAnthropicPipecatDocker
The cost of a repeat
To find an order, a ticket or a booking, an agent asks its model which tool to call, then waits. In our runs that decision took a median of 0.68 to 0.95 seconds, and the same decisions came up again and again.
Chat and text
Every "where is my order" pays for a model round trip that decides exactly what it decided last time.
Voice
On a call, each decision is dead air: about 0.95 s at the median and 2.3 s at the 90th percentile in our live support run.
Back office
Tickets, claims and parcels follow the same steps every time: find the person, fetch the record. The model re-decides it on every request.
How it works
Wont is a proxy you run yourself, between your agent and its model. It never guesses: a step is served only after it has matched your model on your own traffic and a person has said yes.
01
Change one base URL. Every request still goes to your model while Wont records each decision on your machine and counts the ones your agent makes the same way every time, down to where each argument comes from.
02
On live traffic, Wont predicts each repeated step beside your model and keeps score. A step earns a certificate only when a 95% bound puts its error under 2%, over at least 24 hours, 2 days and 10 users.
03
Certified is not enough. A person on your team decides to serve a step, with one command, and can undo it the same way.
04
Wont answers that step in about 2 ms, in your provider's exact format, and your code runs the tool as it always does. A new prompt, model or tool list stops serving until the step is proven again.
Not a cache, and not a router. Wont learns how each argument is derived, so a new customer's ticket number is read from what they wrote, never replayed from the last conversation.
For chat and text agents
A support agent spends much of its day on the same moves: find the customer, fetch the order, check the ticket. Wont takes over the moves it has proven, so your model's time goes to the reply, the judgment and anything new.
find_user_id_by_emailserved by Wont · 1.8 msget_order_detailsserved by Wont · 1.8 mscancel_pending_ordera write · always your modelModel time and tokens Wont would remove, by desk
model timetokens
Exploratory: each desk's share of the agent's whole model bill, replies included, if every proven step were promoted.
For voice agents
On a call, your agent can't think out loud. Each time it decides which tool to call, the caller hears silence: about 0.95 s at the median and 2.3 s at the 90th percentile in our live support run.
Where Wont sits on a call
Measured on spoken calls · Pipecat, Deepgram, Qwen3-30B
Synthetic callers. The agent speaks a short preamble first, so the first sound the caller hears did not change; the silence after it did. Serving the step itself on calls is measured once its certificate passes.
Benchmarks
Six support desks, a real model, simulated customers, and every plan frozen before the run. The misses are published with the wins.
The same 748 retail tool decisions. The Decisions API guesses the next tool on every turn; Wont answers the whole call, tool and arguments, only where it has proof.
Swipe to see the whole chart.
| Measure | Decisions API | Wont |
|---|---|---|
| Predicts | the next tool | tool and every argument |
| Right, where it answers | 69.8% to 96.1% | 98.9% |
| Time per decision | 155 ms | 1.8 ms |
| Cost per 1,000 | $0.33 | $0 |
| Still needs a model call | yes, for the arguments | no |
| Your data leaves | yes | no |
Wont's 98.9% counts the whole call; the Decisions API's scores count the tool only, and rise as its confidence threshold narrows what it answers. Its time was measured over the network from one laptop.
Tool decisions predicted exactly, tool and every argument, on conversations Wont had never seen.
| Desk | Exact | Wrong | What Wont took over |
|---|---|---|---|
| Airline | 49.3% | 2 of 183 | Booking by code, flight status, flight search |
| IT helpdesk | 47.3% | 0 of 106 | Employee by email, ticket, their laptop |
| Insurance | 37.6% | 0 of 68 | Policy by number or email, its claims |
| Parcel delivery | 33.7% | 0 of 60 | Tracking by number, by order |
| Retail | 24.3% | 0 of 182* | Customer by email, order by number |
| Clinic | 13.9% | 1 of 36 | Patient by ID; refused where the agent varied |
* 2 wrong when run live; today's code refuses both. No wrong prediction was ever served: every one was caught in shadow.
Log scale. A served step skips the model call entirely.
What if Wont is wrong?
A wrong answer costs more than a slow one. So Wont serves nothing until a step is proven on your own traffic and a person says yes, and it steps aside the moment anything changes.
Route certificateshadow
find_user_id_by_email, after the caller gives an email and an order number
The voice route after its first day on spoken calls: the bound passed, the time guards had not.
A certificate per route: its evidence, the agent version it holds for, and how each argument is derived.
Nothing is served until someone on your team runs promote, and demote undoes it.
Cancelling, booking, refunding: always your model's decision. Wont never serves one.
A new prompt, model or tool list, or a sampled audit that disagrees, sends the step back to shadow.
wont proxy kill on stops all serving at once; every request goes to your model.
Traffic stays on your machine. Reports and certificates hold no customer values. Your API key goes only to your provider.
0 writes served. 0 wrong answers served. Every disagreement in our runs was caught in shadow, before anything was promoted.
For developers
The demo runs the whole lifecycle against a fake model, in memory, for eight kinds of support desk. Then put the proxy in front of your own agent, in shadow.
Pre-release: version 0.2.0, with the proxy, installs from the repository on request until it is on PyPI.
# the whole lifecycle in seconds: fake model, in memory pip install ".[proxy]" wont proxy demo --scenario all E-commerce support [ok] customer: "Hi, where is my order A-5001?" learned: get_order(order_id <- extract:user:id[-1]) evidence: 300 comparisons, 0 disagreements, bound 1.49%, certified served: get_order({"order_id":"A-5001"}) (0 model calls) never served: cancel_order (a write) Telecom support [ok] customer: "Hi, my data stopped working. My phone number is (713) 554-6001." learned: find_account_by_phone(phone <- digits(extract:user:phone[-1])) ...
# effects.json: which tools only read (writes are never served) { "get_order": "read", "cancel_order": "write" } wont proxy --effects effects.json serve \ --upstream-openai https://api.openai.com # or --upstream-anthropic https://api.anthropic.com from openai import OpenAI client = OpenAI( base_url="http://127.0.0.1:8787/v1", default_headers={ "x-wont-agent": "support-bot", "x-wont-intent": "order_status", "x-wont-user": customer_id, }, )
# what repeats in your traffic (no customer values) wont proxy --effects effects.json report --out wont-report # each route's live evidence and certificate wont proxy --effects effects.json routes wont proxy --effects effects.json cert ROUTE # a person's decision, reversible wont proxy --effects effects.json promote ROUTE wont proxy --effects effects.json demote ROUTE # stop serving everywhere wont proxy kill on
Runs as one Python process or the Docker image in deploy/proxy/, published on 127.0.0.1 only. Pipecat has a full voice example in the repository.
Questions
Every number on this page comes from a run recorded in our repository, with its plan, its data hashes and its misses. Exploratory results are marked.
No. Your model answers everything new, everything uncertain, every reply to your user and every write. Wont takes only the steps it has proven and you have promoted.
Both, and back-office agents too: anything that calls an OpenAI- or Anthropic-compatible model to decide its tools. On a call, the time saved is silence the caller no longer hears.
Before promotion nothing is served, so a miss only raises the bar for certification. After promotion a share of requests is still checked against your model, and one disagreement sends the step back to shadow.
Your traffic stays on your machine, in a local store you can turn off. Reports and certificates hold no customer values, and your API key goes only to your provider.
Wont works best with a label per request, such as a flow node or a menu choice. Without one, Wont grouping requests itself reached 27.5% on retail in an exploratory test. Inferring the label is next.
Not yet. Every result here is a real model with simulated customers. A free shadow-mode pilot measures what Wont would do on your traffic, without serving anything.
A free design-partner pilot. It records, learns and reports, and serves nothing until you promote a certified step.
or write to prem@usewont.com