Conversational AI with memory and your documents built in.
Send a message, receive a grounded answer. VectorForgeAI keeps the thread, retrieves the right passages from your documents, runs your functions, and scores each conversation, so none of it has to live in your database.
Do you ship to Norway?
Yes. Orders to Norway are sent with DHL and clear customs in Oslo.
How long does it take to get there?
Processing this turn
- 1 Rewrite “there” resolved to Norway from the last 3 turns
- 2 Retrieve 3 documents, 3 chunks each
- 3 Rerank 30 chunks scored, top 5 kept
- 4 Answer Low temperature, names its source
Deliveries to Norway take three to five working days after dispatch.
0.81 shipping-zones.md, chunk 4
Inside one request
What happens between the question and the answer
Most chat APIs provide a model and leave retrieval to you. These four stages run automatically on every message you send.
-
1
Resolve the follow-up
A question like “how long does it take to get there?” means nothing on its own. The last three turns are folded into the search query before anything is retrieved.
-
2
Retrieve
Passages are pulled from the collections attached to the conversation. Documents per turn and chunks per document are both configurable.
-
3
Rerank
Candidate chunks are scored against the resolved query and reduced to the ones that earn their place in the prompt.
-
4
Answer
The model replies at temperature 0.15 with the retrieved passages in context, and the response names the documents it used.
Memory
Your application stores one identifier
The first call returns a conversation_id. Send it with the next message and the history, retrieved context, and tool definitions are all preserved.
- user
- Your own identifier for the person in the conversation. Threads are scoped to it, so one customer can never be served another's conversation.
- system_context
- The instructions that define the assistant. Set it once, or change it mid-conversation when the context changes.
- collections
- The document collections this conversation is permitted to read from.
curl -X POST https://api.vectorforgeai.com/v1/messages \
-H "Authorization: Bearer $API_KEY" \
-H "Team-Token: $TEAM_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"user": "user_12345",
"conversation": "jkl789mno012",
"message": "How long does it take to get there?"
}'
Tool calling
Connect the assistant to your own systems
Describe a function and its parameters. When the assistant needs it, the response returns type: "tool_calls" instead of an answer. You execute it where your data lives and post the result back.
- Your code executes the function, so credentials never leave your infrastructure.
- Parameters are typed: string, number, boolean, enum, object, or array.
- Schemas are enforced strictly, so arguments arrive in the shape you declared.
- Set no_tools on a request when you want a plain answer instead.
curl -X POST https://api.vectorforgeai.com/v1/messages \
-H "Authorization: Bearer $API_KEY" \
-H "Team-Token: $TEAM_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"user": "user_12345",
"conversation": "jkl789mno012",
"tool_results": [{
"tool_id": "call_abc123",
"tool_name": "order_status",
"arguments": { "order_id": "A-4471" },
"result": "Dispatched 12 March, arriving 17 March"
}]
}'
# No message field. The assistant answers using the result.
Quality and insight
Ratings, reviews, and outcome scores on every thread
Every conversation is rated and reviewed, so you can see which questions the assistant could not answer and what it was missing.
Ratings per answer
Place a thumbs up and down next to each reply and post it to the rating endpoint. The score attaches to that message, not the whole thread.
A review of every thread
Each conversation is reviewed for what the person wanted, whether they got it, and what would have helped: a missing document, or a tool you have not built yet.
Happiness, scored 1 to 10
Estimated from the tone of the person's messages and how the thread ended.
The dashboard turns these signals into activity over time, outcome mix, average thread length, which tools are called, and how many people return.
Operating at scale
Predictable behaviour for long threads, errors, limits, and retention
- Long threads
- Past roughly 75,000 tokens a conversation moves to a smaller model, so a long thread stays affordable instead of failing.
- Upstream errors
- Requests are retried before they are abandoned, and a blocked response returns a clear error rather than an empty answer.
- Budget reached
- A conversation that reaches your usage limit is marked, rather than silently dropping messages.
- Retention
- Conversations are deleted after 30 days. Closing the account removes everything.
Run your first grounded conversation today
One POST returns an answer. Everything described on this page is enabled by default.