The hardest thing for AI to learn is how your company works: how ANIMA builds company context
Using a kickoff meeting as the example, see how four techniques get AI to deliver a first draft that follows your company's way of working.
AI knows everything. Except your company.
What clients care about, which contacts have preferences and sore spots, which pitfalls past projects ran into: this experience used to pass by word of mouth among senior colleagues. We call it “company context.”
ANIMA is enterprise AI from ANIWEAVE. It joins collaboration tools such as LINE, Slack, and Teams, learns how your company works, and has the work prepared before you ask. It builds company context with four techniques, so the first version of any work handed to it already follows how your company does things.
The same kickoff meeting, four details that make the difference
Take the kickoff meeting for a new project as an example. (The scenario and numbers below are illustrative.)
On Thursday, the company holds a kickoff meeting for a new project with a client it has worked with before. On Tuesday afternoon, the project manager asks ANIMA in the project group to prepare the meeting materials. By early Wednesday, the agenda, timeline, and pre-meeting summary are ready.
Without company context, the agenda lists only generic discussion items. With company context, all four areas are arranged according to the company's experience:
- Scope: A generic agenda lists only discussion items. This client tends to add requests during kickoff meetings, so the agenda first lays out what is in scope and out of scope, and added requests are quoted separately.
- Decision focus: A generic agenda follows the order of discussion. The client's senior leaders care more about the launch date than the number of features, so the timeline goes first, with the delivery date of each milestone explained up front.
- Contact's preferences: The usual approach is to send the full materials on the day of the meeting. This project contact likes to receive a one-page summary the day before, with details in an attachment.
- Contact's sore spots: A generic agenda doesn't give meeting minutes any special attention. In the last project, a change was agreed only verbally and never put in writing, and at close-out the two sides disagreed about the extra fees. This time, the last item on the agenda is confirming how the meeting minutes will be signed off.
What are the details that matter in your project meetings? Bring some recent project materials and book a free 30-minute call.
What company context is
Company context is the background AI needs to work inside a company: how the team actually works, the rules and decision criteria currently in force, the latest status of every case, and the reasoning behind past decisions.
Some of this lives in senior colleagues' experience. The rest is scattered across group chats, cloud documents, and various systems. For AI to understand this internal information, it first has to connect to the data sources and keep the permissions already in place. ANIMA then organizes it into a form AI can use directly.
Four techniques that learn how your company works
- Process mining: does two things. First, it analyzes the collaboration records the company has authorized and reconstructs how a piece of work actually runs, including every step, every handoff, and the time each one takes. What to prepare before a kickoff meeting and what to follow up on afterward are learned from this. Second, it ranks work by the time it could save and its impact, then recommends for each item whether to hand it to AI to complete, have AI assist, redesign the workflow, or keep it manual. Once a change is live, it measures the actual results again.
- Knowledge base: takes rules, standards, and decision criteria scattered in different places and reorganizes them into one company handbook, with a single official version as the reference. When a source is updated, the content is revised with it. How a project's scope is defined and how acceptance criteria are written both rely on it.
- Memory: remembers personal preferences, team guidelines, and organizational culture, along with each case's status and the reasoning behind decisions. How this client tends to add requests, and the pre-meeting materials a contact prefers, are kept here.
- Index: builds one master directory, with sources attached, across all conversations, documents, and system records, keeping the origin, time, and version of every item. Finding things is faster and every judgment can be traced to its source. AI also doesn't have to read large amounts of data over and over, which lowers usage costs.
The results of the four techniques come together as an organization map and a personal map: a map of how the company and each colleague work, which ANIMA draws on for every task. This is also the foundation of Workspace Intelligence: understand the company first, so the work is ready before you ask.
How we handle permissions, conflicts, and ownership
Handing company data to AI raises three questions.
Who can see what?
ANIMA keeps the access permissions of each data source. Every user can only reach content they are authorized to see.
What if the information conflicts?
When sources disagree, ANIMA goes with the latest valid version and attaches where each source came from. When a judgment call is needed, it goes to the colleague responsible. Practices learned through process mining are also only switched on after the team confirms them.
Who owns all this?
Company context belongs to your company. Workflows, the knowledge base, and memory can all be viewed and edited, and they are kept in full even if the underlying model changes.
Teams already using ChatGPT and Claude can bring company context to those tools too.
Bring a recent project's materials, book a free 30-minute call, and see live how ANIMA prepares your project meetings.