Our values

What drives Pinevon

Pinevon builds AI around real business information and human judgement, so businesses can find clearer answers, support their customers and stay in control of important decisions.

Our mission

Businesses already hold the facts they need: what they sell, what is in stock, who asked what and what was agreed. We build AI that works from those facts, says plainly when they are missing, and leaves important decisions with the people responsible for them. We try to be honest about what our tools cannot do yet, because a business can only rely on software it understands.

Four commitments, and how to check them

Each one comes with an example you can see in Anchor today.

Grounded answers
The customer assistant searches your products, shows what it used, and hands the conversation to your team when it finds nothing relevant.
How grounding works in Anchor
Honesty about limitations
When the records needed for an answer are not there, Anchor says so instead of showing a zero that looks like a fact. Approving a restock suggestion records your decision; it does not order stock.
What we do and do not claim
Human oversight
Suggested actions wait for an owner or admin to approve them, and approval applies to the exact version that was reviewed.
Human review in Anchor
Practical usefulness
You can bring an existing catalog in from a CSV or Excel file and check a preview before anything is saved, instead of retyping products.
Import a catalog

The principles behind them

In place today = checked against the productWorking towards = intent, not a claim

01
Useful intelligence
Technology should solve a real operational problem. If a feature does not save time, prevent an error or improve a decision, it does not ship.
In place today
Anchor starts from the questions a business asks every day: what is in stock, what a product costs, what sold, which quotes are waiting.
Working towards
Measuring, with the businesses that use Anchor, how much time and how many mistakes it actually saves. We do not publish such numbers yet.
02
Simplicity
Powerful systems should not demand complicated setup. A small business should reach a first useful answer in an afternoon, not a quarter.
In place today
Products can be imported from a CSV or Excel (.xlsx) file, with a preview that explains every row that needs attention before anything is saved.
Working towards
A first useful answer within an afternoon for a small business. It is a goal we design for, not a measured result.
03
Trust
Security, privacy and honest communication are product requirements. We say what we do, what we do not do, and what we have not verified.
In place today
The Trust Center lists what we have not certified, and the changelog records what actually shipped.
Working towards
Independent verification of our security practices once the company is ready for it. Nothing has been audited or certified yet.
04
Connected systems
Business information is more valuable when catalog, orders, customers and conversations share one source of truth instead of living in separate tools.
In place today
Catalog, stock, quotes, orders and customer conversations live in one Anchor workspace. A quote becomes an order when the customer accepts it, and stock updates with it.
Working towards
More connections to the tools businesses already use, chosen by what customers ask for.
05
Human control
AI should assist people and businesses, not silently take irreversible actions. Suggestions wait for approval; questions the records cannot answer are handed to a person.
In place today
Suggested actions wait for an owner or admin, and approval covers exactly the version that was reviewed. The customer assistant hands a conversation to the team when the records do not support an answer.
Working towards
Keeping approval and audit intact as more multi-step work becomes possible, including any future workflow builder.
06
Long-term building
We would rather build one product deeply than announce many. Durable customer value beats feature noise.
In place today
One product, Anchor, with a public page that separates what is available now from what is next and what is only being explored.
Working towards
Growing carefully with the businesses that use Anchor rather than announcing products before they exist.

More decisions the principles shaped

Principles only matter when they change what gets built. These are real choices in Anchor today.

Recording a sale never takes a payment
In-person sales can be recorded for stock and reporting, but payment stays with the tools you already use. We would rather be clear than half-build a payment system.
No customer logos until customers approve them
Our customer pages show an honest empty state and a way to share a story. Nothing is published without written approval.
A workflow builder is still only being explored
Letting a business chain several steps together is useful, but not before approval and audit work the same way they do for single actions.

Tell us where we fall short.

If something we say or build does not match these values, we want to know.

Our mission

Four commitments, and how to check them

Each one comes with an example you can see in Anchor today.

Grounded answers

The customer assistant searches your products, shows what it used, and hands the conversation to your team when it finds nothing relevant.

How grounding works in Anchor

Honesty about limitations

When the records needed for an answer are not there, Anchor says so instead of showing a zero that looks like a fact. Approving a restock suggestion records your decision; it does not order stock.

What we do and do not claim

Human oversight

Suggested actions wait for an owner or admin to approve them, and approval applies to the exact version that was reviewed.

Human review in Anchor

Practical usefulness

You can bring an existing catalog in from a CSV or Excel file and check a preview before anything is saved, instead of retyping products.

Import a catalog

The principles behind them

In place today = checked against the productWorking towards = intent, not a claim

Useful intelligence

Technology should solve a real operational problem. If a feature does not save time, prevent an error or improve a decision, it does not ship.

In place today: Anchor starts from the questions a business asks every day: what is in stock, what a product costs, what sold, which quotes are waiting.
Working towards: Measuring, with the businesses that use Anchor, how much time and how many mistakes it actually saves. We do not publish such numbers yet.

Simplicity

Powerful systems should not demand complicated setup. A small business should reach a first useful answer in an afternoon, not a quarter.

In place today: Products can be imported from a CSV or Excel (.xlsx) file, with a preview that explains every row that needs attention before anything is saved.
Working towards: A first useful answer within an afternoon for a small business. It is a goal we design for, not a measured result.

Trust

Security, privacy and honest communication are product requirements. We say what we do, what we do not do, and what we have not verified.

In place today: The Trust Center lists what we have not certified, and the changelog records what actually shipped.
Working towards: Independent verification of our security practices once the company is ready for it. Nothing has been audited or certified yet.

Connected systems

Business information is more valuable when catalog, orders, customers and conversations share one source of truth instead of living in separate tools.

In place today: Catalog, stock, quotes, orders and customer conversations live in one Anchor workspace. A quote becomes an order when the customer accepts it, and stock updates with it.
Working towards: More connections to the tools businesses already use, chosen by what customers ask for.

Human control

AI should assist people and businesses, not silently take irreversible actions. Suggestions wait for approval; questions the records cannot answer are handed to a person.

In place today: Suggested actions wait for an owner or admin, and approval covers exactly the version that was reviewed. The customer assistant hands a conversation to the team when the records do not support an answer.
Working towards: Keeping approval and audit intact as more multi-step work becomes possible, including any future workflow builder.

Long-term building

We would rather build one product deeply than announce many. Durable customer value beats feature noise.

In place today: One product, Anchor, with a public page that separates what is available now from what is next and what is only being explored.
Working towards: Growing carefully with the businesses that use Anchor rather than announcing products before they exist.

More decisions the principles shaped

Principles only matter when they change what gets built. These are real choices in Anchor today.

Recording a sale never takes a payment

In-person sales can be recorded for stock and reporting, but payment stays with the tools you already use. We would rather be clear than half-build a payment system.

No customer logos until customers approve them

Our customer pages show an honest empty state and a way to share a story. Nothing is published without written approval.

A workflow builder is still only being explored

Letting a business chain several steps together is useful, but not before approval and audit work the same way they do for single actions.